0b - US Patent For Adaptive Scaling in Trajectory Generation Patent (Patent # 11,048,260 Issued June 29, 2021) - Justia Patents Search

12/5/22, 11:33 PM US Patent for Adaptive scaling in trajectory generation Patent (Patent # 11,048,260 issued June 29, 2021)
29, 2021) - Justia Patents Search
Adaptive scaling in trajectory

generation
Nov 2, 2018 - Zoox, Inc.
Techniques for generating trajectories and drivable

areas for navigating a vehicle in an environment are
discussed herein. The techniques can include receiving
a reference trajectory representing an initial trajectory
for a vehicle, such as an autonomous vehicle, to
traverse the environment. A point density can be
determined for various portions of the reference
trajectory. In some cases, the point density can be
based at least in part on a cost associated with a
curvature value associated the reference trajectory or a
cost associated with a distance between the reference
trajectory and an obstacle in the environment.
Further, the techniques can include evaluating a cost
function at points on the reference trajectory to
generate a target trajectory with respect to the
https://patents.justia.com/patent/11048260 1/48
12/5/22, 11:33 PM US Patent for Adaptive scaling in trajectory generation Patent (Patent # 11,048,260 issued June 29, 2021) - Justia Patents Search
reference trajectory, and controlling the autonomous

vehicle to traverse the environment based at least in
part on the target trajectory.
Latest Zoox, Inc. Patents:

Sharing occlusion data
Emergency sound localization
Component tray assembly for vehicles

Automatic creation and updating of maps
Display screen or portion thereof having an animated graphical user interface
Skip to: Description · Claims · References Cited · Patent History · Patent History
Description
BACKGROUND
Various methods, apparatuses, and systems are utilized by autonomous vehicles to guide
such autonomous vehicles through environments including various static and dynamic
objects. For instance, autonomous vehicles utilize route planning methods, apparatuses,
and systems to guide autonomous vehicles through congested areas with other moving
vehicles (autonomous or otherwise), moving people, stationary buildings, etc. In some
examples, generating a route may be computationally intensive and/or may not provide a
safe or comfortable route for passengers.
BRIEF DESCRIPTION OF THE DRAWINGS

The detailed description is described with reference to the accompanying figures. In the
figures, the left-most digit(s) of a reference number identifies the figure in which the
reference number first appears. The use of the same reference numbers in different figures
indicates similar or identical components or features.
FIG. 1 is a schematic diagram illustrating an example implementation of a system for

generating a trajectory, generating a drivable area, and evaluating costs, in accordance with
embodiments of the disclosure.
FIG. 2 is an illustration of a reference trajectory comprising regions of differing point

densities, in accordance with embodiments of the disclosure.
FIG. 3 is an illustration of determining weight(s) for a cost associated with generating a

target trajectory based on a reference trajectory when navigating around an obstacle, in
accordance with embodiments of the disclosure.
FIG. 4 is an illustration of determining weight(s) for a cost associated with generating a

target trajectory based on a reference trajectory when changing lanes, in accordance with
embodiments of the disclosure.
FIG. 5A is an illustration of a dilated region, a collision region, and a safety region

associated with a drivable area, in accordance with embodiments of the disclosure.
FIG. 5B is an illustration of regions associated with a drivable area based at least in part on
a classification of an object in an environment and/or on a velocity of a vehicle in the
environment, in accordance with embodiments of the disclosure.
FIG. 6 is an illustration of updating a region based on identifying an object such as a

bicyclist proximate to the vehicle, in accordance with embodiments of the disclosure.
FIG. 7 depicts a block diagram of an example system for implementing the techniques
described herein.
FIG. 8 depicts an example process for determining a point density for points associated
with a reference trajectory and evaluating costs at the points to generate a target trajectory,
in accordance with embodiments of the disclosure.
FIG. 9 depicts an example process for determining weights for points associated with a
reference trajectory and evaluating, based on the weights, costs at the points to determine a
target trajectory, in accordance with embodiments of the disclosure.
DETAILED DESCRIPTION
As discussed above, generating a route for an autonomous vehicle through an environment
may be computationally intensive and/or may not provide a safe or comfortable route for
passengers. This application describes techniques for reducing a computational burden of
planning a route through an environment, may improve an accuracy and precision of
trajectory generation, and/or may improve safety and comfort of routes for passengers. For
instance, the techniques described herein may reduce the computational burden by
adaptively scaling a density of trajectory points (e.g., points along the trajectory used for
determining associated costs and controls) based on the level of activity (e.g., cost(s)
associated with curvature(s) and/or object(s)) along the route. By using a lower density of
trajectory points for portions of the route that have lower activity (e.g., lower cost(s)
associated with curvature(s) and/or fewer nearby object(s)), the computational intensity of
trajectory generation can be reduced. Such costs as referred to herein may be, as non-
limiting examples, proportional with respect to the curvature or distance, an L1, L2,
quadratic, Huber, polynomial, exponential, function or the like, including any combination
thereof. As another example, the techniques described herein may increase accuracy
and/or precision of trajectory generation by using a relatively higher density of trajectory
points for portions of the route that have higher activity (e.g., higher cost(s) associated with
curvature(s) and/or more nearby object(s)). By way of another example, cost functions
quantitatively balance goals of comfort, vehicle dynamics, safety, and the like. Techniques
discussed herein include adaptively scaling weights associated with one or more costs to
enhance safety and/or comfort when determining the contours of a trajectory through an
environment. Further, regions establishing buffers around objects in an environment can
be increased or decreased in size depending on a classification of an object (e.g.,
pedestrians, vehicles, etc.) and/or depending on a velocity of the autonomous vehicle in the
environment.
This disclosure is directed to techniques for generating trajectories and drivable areas for
navigating a vehicle in an environment. For example, an autonomous vehicle can receive a
reference trajectory representing an initial path or trajectory for the autonomous vehicle to
follow in an environment. A point density of points on the reference trajectory can be based
at least in part on a cost associated with a curvature of the reference trajectory and/or on a
cost associated with a distance between the reference trajectory and an obstacle in the
environment. A weight associated with various costs can be based on triggers such as
jumps or discontinuities in a reference trajectory or obstacle costs meeting or exceeding a
threshold. Further, a drivable area can represent a region in the environment where the
autonomous vehicle can travel. Boundaries of the drivable region can include information
about object(s) in the environment and probabilistic distances between the boundaries and
the object(s). In some examples, regions associated with the drivable area can be based on
a classification of objects in the environment and/or on a velocity of the autonomous
vehicle in the environment. A target trajectory can be generated with respect to the
reference trajectory based at least in part on evaluating one or more costs at points on the
reference trajectory relative to the drivable area, whereby a density of points on the
reference trajectory and weights associated with cost(s) can be determined as discussed
herein.
In some examples, a planning system of an autonomous vehicle can include one or more
layers for generating and optimizing one or more trajectories for the autonomous vehicle to
traverse an environment. For example, a first layer of the planning system can receive or
determine a lane reference (also referred to as a reference trajectory), which may
correspond to or be associated with a center of a road segment. Costs associated with
points on the lane reference can be evaluated and optimized to generate a first target
trajectory. For example, a state of the vehicle can be evaluated along each point on the lane
reference (or reference trajectory) to evaluate changing states of the vehicle over time (e.g.,
sometimes referred to as a “rollout”). In some examples, the first target trajectory can be
provided to a second layer of the planning system, whereby the first target trajectory is
used as a reference trajectory. Costs associated with points on the reference trajectory can
be evaluated and optimized to generate a second target trajectory. In some instances, the
second target trajectory can be optimized further or can be used to control the autonomous
vehicle. In some examples, the first layer can optimize the reference trajectory with respect
to a distance between points and/or the second layer can optimize the reference trajectory
with respect to a time between points, although other combinations are contemplated here.
In some examples, a reference trajectory representing an initial path or trajectory for an

autonomous vehicle to follow can be generated or received by a computing device of the
autonomous vehicle. In some examples, the reference trajectory can correspond to a
centerline of a road segment, although the reference trajectory can represent any path in an
environment. In some examples, points on the reference trajectory can be established,
selected, sampled, or otherwise determined based at least in part on a cost associated with
a curvature value associated with the reference trajectory and/or based at least in part on a
cost associated with a distance between a point associated with a region of the reference
trajectory and one or more obstacles in the environment. For example, a region associated
with a relatively high curvature value may represent a maneuver around an obstacle in the
environment or may represent a maneuver such as turning a corner, parking, and the like.
By way of another example, a region of the reference trajectory can be relatively close to an
obstacle in the environment. Accordingly, in the case of a high cost associated with a
curvature and/or of a high cost associated with a distance to an obstacle, the region can be
associated with a relatively higher point density of points on the reference trajectory. In
some cases, one or more costs can be evaluated at the points on the reference trajectory to
generate a target trajectory, whereby the autonomous vehicle can be controlled to follow
the target trajectory. Accordingly, dynamically varying the density of points on the
reference trajectory based on a cost associated with a curvature of the reference trajectory
and/or based on a cost associated with a distance between the point on the reference
trajectory and a point associated with an obstacle can affect an amount of downstream
processing in connection with generating a target trajectory.
In some examples, a target trajectory can be generated with respect to the reference
trajectory by evaluating one or more costs associated with points on the reference
trajectory. In general, the one or more costs may include, but is not limited to a reference
cost, an obstacle cost, a lateral cost, a longitudinal cost, and the like. Additional details of
the one or more costs are provided below. Further, a cost can be associated with a weight
that can increase or decrease a cost associated with a trajectory or a cost associated with a
point of the trajectory. In some examples, the operations discussed herein can include
decreasing a weight associated with a cost based on triggers such as discontinuities in a
reference trajectory or one or more costs meeting or exceeding a threshold. By way of
example, and without limitation, a weight associated with a reference cost (discussed
below) can be decreased when the vehicle is navigating around a double-parked vehicle or
while changing lanes. In some examples, a reference cost can comprise a cost associated
with a difference between a point (also referred to as a reference point) on the reference
trajectory and a corresponding point (also referred to as a point or a target point) on the
target trajectory, whereby the difference represents one or more difference in a yaw, lateral
offset, velocity, acceleration, curvature, curvature rate, and the like. In some examples,
decreasing a weight associated with a reference cost can reduce a penalty associated with
the target trajectory being located a distance away from the reference trajectory, which can
provide smoother transitions leading towards safer and/or more comfortable vehicle
operations. For example, decreasing the reference costs associated with a portion of the
reference trajectory can lead to increasing a buffer between the vehicle and an obstacle in
the environment.
In other examples, such as while changing lanes, selecting a weight for a reference cost can
be combined with other checks to provide safe and comfortable vehicle dynamics. For
example, during a lane change action a target trajectory can be checked to maintain a
lateral displacement less than a threshold distance given a distance traveled forward by the
vehicle and/or given a period of travel time. By way of example, and without limitation, a
vehicle may be constrained to traveling 1 meter in a lateral direction for every 2 meters the
vehicle travels in a forward direction, thereby reducing a sideways force during the lane
change. By way of another example, and without limitation, the vehicle may further be
constrained to traveling 1 meter in a lateral direction for a sliding window of time, such as
two seconds. Of course, other distances and times are contemplated herein.
In some examples, an obstacle cost can comprise a cost associated with a distance between
a point on the reference trajectory or the target trajectory and a point associated with an
obstacle in the environment. By way of example, the point associated with the obstacle can
correspond to a point on a boundary of a drivable area or can correspond to a point
associated with the obstacle in the environment. In some examples, an obstacle in the
environment can include, but is not limited to a static object (e.g., building, curb, sidewalk,
lane marking, sign post, traffic light, tree, etc.) or a dynamic object (e.g., a vehicle, bicyclist,
pedestrian, animal, etc.). In some examples, a dynamic object can also be referred to as an
agent. In some examples, a static object or a dynamic object can be referred to generally as
an object or an obstacle.
In some examples, a lateral cost can refer to a cost associated with steering inputs to the
vehicle, such as maximum steering inputs relative to a velocity of the vehicle.
In some examples, a longitudinal cost can refer to a cost associated with a velocity and/or
acceleration of the vehicle (e.g., maximum braking and/or acceleration).
As introduced above, the vehicle can determine a drivable area that represents a region in
the environment where the vehicle can travel. In some examples, a computing device of an
autonomous vehicle can receive sensor data captured by one or more sensors of the
autonomous vehicle and can determine one or more objects in the environment and/or
attributes of the one or more objects in the environment. In some examples, the
autonomous vehicle can utilize the object(s) and/or the attributes of the object(s) to
determine which object(s) should be included in determining extents of the drivable area.
Accordingly, the autonomous vehicle can plan a trajectory (e.g., a reference trajectory
and/or the target trajectory) within the extents of the drivable area.
In some examples, the drivable area can comprise a dilated region, a collision region,
and/or a safety region. For example, the dilated region can be statically or dynamically
generated with respect to a lane boundary to represent the largest extent of the drivable
area, and can comprise information about object(s) in the environment and probabilistic
distances between the boundaries and the object(s) and the reference and/or target
trajectories. For example, the dilated region can represent a buffer associated with a
boundary based at least in part on a distance (e.g., half of a width of a vehicle) plus some
distance based on an uncertainty of sensor noise, which may be based at least in part on an
object classification. Further, and in some examples, the collision region can represent a
smaller drivable area than the dilated region (e.g., representing a greater distance between
an obstacle and a boundary of the collision region) representing a region for the
autonomous vehicle to avoid to further reduce a likelihood that the autonomous vehicle
will collide with an object in the environment. In some examples, a cost associated with
entering the collision region can be relatively high (relative to the safety region). In some
examples, the safety region can represent a region smaller than the collision region and the
dilated region to provide a buffer between the autonomous vehicle and the object in the
environment. In some examples, a cost associated with entering the safety region can be
lower than a cost associated with the collision region. In some examples, the collision
region and/or the safety region can also be associated with information about object(s) in
the environment and probabilistic distances between the boundaries and the object(s). In
some examples, the autonomous vehicle can evaluate costs based at least in part on
distance(s) between points on the reference trajectory and/or the target trajectory and one
or more points associated with the regions, as discussed herein. In some examples, the
cost(s) associated with the region(s) may vary. For example, a cost and/or weight
associated with the safety region may be relatively less than a cost and/or weight associated
with the collision region.
In some examples, a size of the region(s) can be based at least in part on a classification
type of objects that the regions represent or are otherwise associated with in the
environment. For example, a safety region associated with a pedestrian may be associated
with a first distance or size and a safety region associated with a vehicle may be associated
with a second distance or size that is different than the first distance or size. By way of
example, and without limitation, a safety region associated with the pedestrian can be
larger than a safety region associated with a vehicle, which may result in the autonomous
vehicle giving a larger buffer to the pedestrian than to the vehicle, assuming other factors
are the same.
In some examples, a size of the region(s) can be based at least in part on a velocity of the
autonomous vehicle. For example, a safety region associated with a pedestrian may be
associated with a first distance or size while the autonomous vehicle is traveling at a first
velocity (or is predicted to travel at a reference point corresponding to the region) and may
be associated with a second distance or size while the autonomous vehicle is traveling at a
second velocity (or is predicted to travel at a reference point corresponding to the region)
that is different than the first velocity. By way of example, and without limitation, a size of
the safety region can be increased as the velocity of the autonomous vehicle increases,
which may result in the autonomous vehicle giving a larger buffer to the pedestrian at a
higher velocity, assuming other factors are the same.
The techniques discussed herein can improve a functioning of a computing device in a

number of additional ways. In some cases, varying a density of points associated with the
reference trajectory can reduce an amount of processing by increasing a density of points in

regions of high interest or in regions with high activity and by decreasing a density of
points in regions of low interest or low activity. Further, because additional processing
steps, such as evaluating costs, are based on the points on the reference trajectory, reducing
a number of points to consider can represent a large reduction in overall processing
without sacrificing accuracy or precision. In some examples, selecting weights associated
with a cost can result in safer trajectories (e.g., by increasing a distance between an object
and a vehicle in the environment) and/or can result in more comfortable trajectories (e.g.,
by reducing lateral acceleration during a lane change action). In some examples, varying a
size of regions associated with a drivable area can improve safety by generating and/or
modifying the regions based on a classification type and/or a velocity of the autonomous
vehicle. Further, the techniques discussed herein can be used alone or in combination to
improve safety and/or comfort in a variety of systems that generate trajectories. These and
other improvements to the functioning of the computer are discussed herein.
The techniques described herein can be implemented in a number of ways. Example

implementations are provided below with reference to the following figures. Although
discussed in the context of an autonomous vehicle, the methods, apparatuses, and systems
described herein can be applied to a variety of systems (e.g., a sensor system or a robotic
platform), and are not limited to autonomous vehicles. In another example, the techniques
can be utilized in an aviation or nautical context. Additionally, the techniques described
herein can be used with real data (e.g., captured using sensor(s)), simulated data (e.g.,
generated by a simulator), or any combination of the two.
FIG. 1 is a schematic diagram illustrating an example implementation of a system for

generating a trajectory, generating a drivable area, and evaluating costs, in accordance with
embodiments of the disclosure. More specifically, FIG. 1 illustrates an example
environment 100 in which a vehicle 102 is positioned. In the illustrated example, the
vehicle 102 is driving in the environment 100, although in other examples the vehicle 102
may be stationary and/or parked in the environment 100. One or more objects, or agents,
also are in the environment 100. For instance, FIG. 1 illustrates additional vehicles 104a
and 104b and pedestrians 106a and 106b in the environment 100. Of course, any number
and/or type of objects can additionally or alternatively be present in the environment 100.
For the purpose of illustration, the vehicle 102 can be an autonomous vehicle configured to
operate according to a Level 5 classification issued by the U.S. National Highway Traffic
Safety Administration, which describes a vehicle capable of performing all safety-critical
functions for the entire trip, with the driver (or occupant) not being expected to control the
vehicle at any time. In such an example, since the vehicle 102 can be configured to control
all functions from start to stop, including all parking functions, it can be unoccupied. This
is merely an example, and the systems and methods described herein can be incorporated
into any ground-borne, airborne, or waterborne vehicle, including those ranging from
vehicles that need to be manually controlled by a driver at all times, to those that are
partially or fully autonomously controlled. Additional details associated with the vehicle
102 are described below.
In the example of FIG. 1, the vehicle 102 can be associated with one or more sensor
systems 108. The sensor system(s) 108 can generate sensor data 110, which can be
utilized by vehicle computing device(s) 112 associated with the vehicle 102 to recognize the
one or more objects, e.g., the vehicles 104 and the pedestrians 106. The sensor system(s)
108 can include, but is/are not limited to, light detection and ranging (LIDAR) sensors,
radio detection and ranging (RADAR) sensors, Time of Flight sensors, ultrasonic
transducers, sound navigation and ranging (SONAR) sensors, location sensors (e.g., global
positioning system (GPS), compass, etc.), inertial sensors (e.g., inertial measurement units,
accelerometers, magnetometers, gyroscopes, etc.), cameras (e.g., RGB, IR, intensity, depth,
etc.), wheel encoders, microphones, environment sensors (e.g., temperature sensors,
humidity sensors, light sensors, pressure sensors, etc.), etc.
In at least one example, the vehicle computing device(s) 112 can include a perception
system, which can perform object detection, segmentation, and/or classification based at
least in part on the sensor data 110 received from the sensor system(s) 108. For instance,
the perception system can detect the vehicles 104a and 104b and/or the pedestrians 106a
and 106b in the environment 100 based on the sensor data 110 generated by the sensor
system(s) 108. Additionally, the perception system can determine an extent of the vehicles
104a and 104b and/or the pedestrians 106a and 106b (e.g., height, weight, length, etc.), a
pose of the vehicles 104a and 104b and/or the pedestrians 106a and 106b (e.g., x-
coordinate, y-coordinate, z-coordinate, pitch, roll, yaw), etc. The sensor system(s) 108 can
continuously generate the sensor data 110 (e.g., in near-real time), which can be utilized by
the perception system (and other systems of the vehicle computing device(s) 112).
The vehicle computing device(s) 112 can also include a drivable area component 114, a
trajectory generation component 116, and a cost(s) component 118. In some examples,
although discussed separately, the trajectory generation component 116 and the cost(s)
component 118 (and/or other components discussed herein) can be integrated into a single
component or algorithm. In general, the drivable area component 114 can include
functionality to generate and/or determine a drivable area 120, in accordance with the
embodiments discussed herein. In general, the trajectory generation component 116 can
include functionality to generate a reference trajectory 122 and/or a target trajectory 124
within the drivable area 120, in accordance with the embodiments discussed herein. In
general, the cost(s) component can include functionality to evaluate one or more costs to
generate the target trajectory 124 with respect to the reference trajectory 122.
In some examples, the drivable area 120 may be a virtual area in the environment 100 in
which the vehicle 102 can travel safely through the environment 100, for example, relative
to objects including the vehicles 104a, 104b and the pedestrians 106a, 106b.
As illustrated in FIG. 1 and detailed further herein, the drivable area 120 may be defined by
virtual boundaries 126 and 128. In the example drivable area 120 illustrated in FIG. 1, the
drivable area 120 may have a variable width 130, for example, in the width or lateral
direction of the vehicle 102. FIG. 1 illustrates a first width 130a and a second width 130b,
which collectively, and along with other widths at other positions in the drivable area 120,
may be referred to as the width 130. As described herein, attributes of the virtual
boundaries 126 and 128, including the width 130, may be determined based at least in
part on the sensor data 110 and/or determinations made by the perception system. For
example, in some implementations, the virtual boundaries 126 and 128 may be
determined based on information about the objects in the environment 100, which may
include information about semantic classifications and/or probabilistic models. In at least
some examples, the virtual boundaries 126 and 128 may be encoded with information
(such as lateral distance to the nearest object, semantic classification of the nearest object,
related probabilities of such distance and classification, etc.). In some examples, the
drivable area 120 can be generated in accordance with the techniques discussed in U.S.
patent application Ser. No. 15/982,694, filed May 17, 2018, which is hereby incorporated
by reference, in its entirety.
As also illustrated in FIG. 1, upon determining the drivable area 120, the vehicle
computing device(s) 112 also may determine a reference trajectory 122 along which the
vehicle 102 may travel through the environment 100. The planner system can determine
routes and/or trajectories to use to control the vehicle 102 based at least in part on the
sensor data 110 received from the sensor system(s) 108 and/or any determinations made
by the perception system. For instance, the planning system can evaluate one or more costs
to generate the target trajectory 124 with respect to the reference trajectory 122 to navigate
safely through the environment relative to the vehicle 104a, 104b and the pedestrians
106a, 106b.
More specifically, FIG. 1 illustrates a scenario in which the vehicle 102 is travelling through
the environment 100, generally in the direction of arrow 132. The vehicle 102 is travelling
on a road 134 having a first lane 136 and a second lane 138. The vehicle 102 is in the first
lane 136, behind the vehicle 104a, and, in the example, is travelling relatively faster than
the vehicle 104a. For example, the vehicle 104a may be slowing down to turn left, to
parallel park on the left side of the road 134, to drop off a passenger or delivery on the left
shoulder of the road, or the like. In examples of this disclosure, the vehicle 102, for
example, by executing the drivable area component 114, may determine the drivable area
120 as a precursor to determining whether the vehicle 102 can safely navigate around the
slowing vehicle 104a. More specifically, the vehicle computing device(s) 112, using the
sensor data 110, may determine information about objects in the environment 100, which
may include the vehicle 104a, the vehicle 104b, the pedestrian 106a, the pedestrian 106b,
and/or additional objects. For example, the vehicle computing device(s) 112, using the
drivable area component 114, may fuse or combine information about each of the objects
to configure the drivable area 120. Such a drivable area 120 may, in turn, be used to
determine the reference trajectory 122 and/or the target trajectory 124 (e.g., by providing
constraints and/or boundaries) along which the vehicle 102 may travel. In some examples,
the reference trajectory 122 can be generated prior to or substantially simultaneously
(within technical tolerances) as the drivable area 120.
With further reference to FIG. 1, the drivable area 120 may have varying widths, for
example, the first width 130a and a second width 130b, and those widths are determined
at least in part based on information about the objects in the environment 100. For
example, at a position immediately in front of the vehicle 102 in the direction illustrated by
arrow 132, the width 130a generally spans the entire width of the road 134. At edges of the
first lane 136 and the second lane 138, a slight offset relative to the shoulder may be
provided, although such an offset may be excluded in other embodiments. Further along
the road 134 in the direction of the arrow 132, the drivable area 120 begins to narrow as
the boundary 128 gradually moves away from the left shoulder until the drivable area 120
is completely confined in the right lane 138. This narrowing of the drivable area 120 is a
result of the presence and/or the relative deceleration of the vehicle 104a. Moreover,
immediately adjacent the vehicle 104a, the drivable area 120 may further narrow, for
example, to provide a minimum lateral offset relative to the vehicle 104a. Similarly, on the
right side of the drivable area 120, offsets may be provided proximate the first pedestrian
106a and the vehicle 104b, which is parked on the right shoulder of the road 134. In some
examples, such a drivable area 120 may also be determined based on a predicted trajectory
of objects, such as that of the pedestrian 106a. As a non-limiting example depicted in FIG.
1, the drivable area 120 indents slightly near the pedestrian 106a in anticipation of the
pedestrian 106a continuing walking in the direction of travel 132. The length of such an
indent may be based on, for example, the velocities of the vehicle 102 and the pedestrian
106a.
As described further herein, the width 130 of the drivable area 120 varies as a result of
information about objects in the environment 100. Thus, for example, offsets are provided
proximate the vehicle 104a, the vehicle 104b, and the pedestrian 106a. These offsets may
serve to provide a safety margin or minimum distance of the vehicle 102 from the objects,
as the vehicle 102 travels in the drivable area 120. In some implementations, the offsets
and/or the minimum distances may differ among and between objects in the environment
100. For example, semantic information about the objects may be used to determine
appropriate offsets. Thus, in FIG. 1, pedestrians may be associated with a first classification
and vehicles may be associated with a second classification. The first classification may
correspond to a minimum offset larger than a minimum offset to which the second
classification corresponds. Moreover, vehicles within the second classification may be
further classified as static, as in the case of the vehicle 104b or dynamic, as in the case of
the vehicle 104a. An offset relative to the moving vehicle 104a may be greater than the
offset associated with the parked vehicle 104b, for example. In further examples,
probabilistic information about the objects may also or alternatively be used to determine
the offsets. For example, there may be some uncertainty associated with the actions of the
moving pedestrian 106a and/or the moving vehicle 104a, for example, because the
pedestrian 106a or the driver of the vehicle 104a may change course. Probabilistic
information may also be used to account for sensor inaccuracies, for example, associated
with the sensor system(s) 108.
The drivable area 120 may allow for more efficient navigation around the vehicle 104a
when compared with conventional navigation techniques. In conventional solutions, for
example, the vehicle 102 may slow down as it approaches the vehicle 104a, for example, to
maintain a minimum distance between the vehicle 102 and the vehicle 104a, and only
upon stopping or slowing to a threshold speed may the vehicle 102 begin to seek out an
alternative route around the vehicle 104a. Alternatively, the vehicle 102 may idle in the
lane 136 until the vehicle 104a turns, parks, moves, accelerates, or the like. However, in
implementations of this disclosure, the drivable area 120 considers multiple objects as well
as information about the objects to provide a more robust understanding of the
environment 100. This understanding may enhance decision making and/or allow for
more effective and efficient control of the vehicle 102 as it travels through the environment
100. With specific regard to FIG. 1, such enhanced decision-making may allow the vehicle
to traverse along the target trajectory 124, without the need to slow down behind the
vehicle 104a and/or await additional decision making. Further, the drivable area 120
provides boundaries for a wide range of possible trajectories which may be employed to
safely navigate a portion of a path. By providing such a wide range of safe trajectories, the
vehicle 102 is able to quickly overcome any hazards, while incorporating information about
all objects in a scene.
FIG. 1 illustrates a single example of using a drivable area to navigate in an environment.

Myriad other examples also are contemplated. For example, the drivable area component
114 may be used in any environment to better navigate with respect to objects.
FIG. 2 is an illustration of a reference trajectory comprising regions of differing point

densities, in accordance with embodiments of the disclosure. More specifically, FIG. 2
illustrates an example environment 200 in which the vehicle 102 is positioned. FIG. 2
further illustrates an obstacle 202 in the environment 200, whereby a reference trajectory
204 represents an initial path for the vehicle 102 to navigate around the obstacle 202 in
the environment 200. In some examples, the obstacle 202 corresponds to the vehicle
104a of FIG. 1, although the obstacle 202 may represent any static object or dynamic
object in the environment 200. Further, the reference trajectory 204 may correspond to
the reference trajectory 122 of FIG. 1, although the reference trajectory 204 may represent
any path or trajectory in the environment 200.
In some examples, the trajectory generation component 116 (or another component of the
vehicle computing device(s) 112) can determine a density of points associated with various
regions of the reference trajectory 204. By way of example and without limitation, the
vehicle computing device(s) 112 can identify region(s) of the reference trajectory 204
corresponding to regions of interest or corresponding to regions of high activity. In some
examples, the region(s) can be identified based at least in part on a cost associated with a
curvature value associated with the reference trajectory 204 and/or based at least in part
on a cost associated with a distance between a point on the reference trajectory 204 and a
distance to a point associated with an obstacle in the environment 200.
As illustrated, the reference trajectory 204 includes portions 206, 208, 210, and 226,
although the reference trajectory 204 can include any number of portions.
The portion 206 can be associated with a point density 212, the portion 208 can be
associated with a point density 214, the portion 210 can be associated with a point density
216, and the portion 226 can be associated with a point density 228.
In some examples, the point density 214 can be higher than the point density 212 and/or
216 because a cost associated with a curvature value associated with the portion 208 is
higher than a curvature value associated with the portions 206 and/or 210. In some
examples, the point density 228 can be higher than the point density 212 and/or 217
because a cost associated with a distance between the portion 226 and the obstacle 202 is
smaller than a distance between the portions 206, 210 and the obstacle 202. That is, the
point density 214 associated with the portion 208 can be based at least in part on a cost
associated with a curvature value associated with the portion 208 and the point density
228 associated with the portion 226 can be based at least in part on a cost associated with
a distance between the portion 226 and the obstacle 202.
In some examples, the point density 214 can correspond to a density of points in space or
time. For example, as discussed above, a first layer of a planning system may optimize a
trajectory based on a distance (which may be a Euclidian distance, length along an inertial
reference, etc.) between points while a second layer of a planning system may optimize the
trajectory based on a time period between points, or vice versa. In some examples, the
difference 218 between adjacent reference points 220 and 222 can represent a distance in
space (e.g., 5 cm, 10 cm, 50 cm, 1 m, and the like) or a distance in time (e.g., 0.05 seconds,
0.1 seconds, 0.2 seconds, 0.5 seconds, 1 second, and the like). In an example where the
point density 214 is higher than the point density 212, the difference 218 between the
adjacent points 220 and 222 can be smaller than a difference 224 between adjacent
reference points associated with the portion 206.
As can be understood in the context of this disclosure, one or more costs can be evaluated
at discrete points on the reference trajectory 204 to generate a target trajectory that the
vehicle 102 can follow to traverse through the environment 200.
FIG. 3 is an illustration of determining weight(s) associated with a cost associated with

generating a target trajectory based on a reference trajectory when navigating around an
obstacle, in accordance with embodiments of the disclosure. More specifically, FIG. 3
illustrates an example environment 300 in which the vehicle 102 is positioned. FIG. 3
further illustrates an obstacle 302 in the environment 300, whereby a reference trajectory
304 represents an initial path for the vehicle 102 to navigate around the obstacle 302 in
the environment 200. In some examples, the reference trajectory 304 can correspond to a
maneuver for the vehicle 102 to navigate around a double-parked vehicle in the
environment 300. In some examples, the obstacle 302 may represent any static object or
dynamic object in the environment 300. Further, the reference trajectory 304 may
represent any path in the environment 300.
In general, the vehicle computing device(s) 112 can evaluate one or more costs at reference
points 306 and 308 to generate a target trajectory 310 having points 312 and 314
corresponding to the reference points 306 and 308, respectively. A dotted line between
points on the reference trajectory 304 and the target trajectory 310 indicates
corresponding points. In some cases, each point on the reference trajectory 304 can be
associated with a corresponding point on the target trajectory 310, although in some cases
the target trajectory 310 can be associated with points that do not necessarily correspond
to point(s) on the reference trajectory. In some cases, a point (e.g., the points 312, 314, and
326) on the target trajectory 310 can be orthogonal to a corresponding reference point
(e.g., the reference points 306, 308, and 324, respectively) on the reference trajectory
304, based at least in part on a curvature value at the reference trajectory 304.
In general, the one or more costs may include, but is not limited to a reference cost, an
obstacle cost, a lateral cost, a longitudinal cost, and the like. A cost can be associated with a
weight (e.g., weights 316 and 318) that can increase or decrease a cost associated with a
trajectory or a cost associated with a point of the trajectory. In some examples, the vehicle
computing device(s) 112 can perform operations to decrease a weight associated with a cost
based on actions or maneuvers to be performed by the vehicle.
A start trigger 320 can represent a condition causing weights associated with cost(s) to be
varied, as discussed herein. In some examples, a trigger can correspond to obstacle costs
meeting or exceeding a threshold (e.g., discussed in context of FIG. 3) and/or jumps or
discontinuities in a reference trajectory (e.g., discussed in the context of FIG. 4). For
example, during an operation to evaluate initial costs associate with reference points on the
reference trajectory 304, an obstacle cost associated with a reference point 324 may meet
or exceed a threshold cost by virtue of the reference point 324 being located within the
obstacle 302. Accordingly, the start trigger can be initiated along the reference trajectory
304 based at least in part on a distance or time in relation to the reference point 324. An
end trigger 322 can represent a period of time or distance associated with the cost(s) (e.g.,
an obstacle cost) associated with the reference trajectory 304 being below a threshold
value. Based at least in part on the start trigger 320 and the end trigger 322, a weight
associated with the points of the reference trajectory 304 and/or the target trajectory 310
can be adjusted to increase or decrease the weights. For example, a weight associated with
the reference point 306 can correspond to the weight 316 and a weight associated with the
reference point 308 can correspond to the weight 318. Accordingly, evaluating costs
associated with the points 306 and 308 can be based at least in part on the weights 316
and 318, respectively. By reducing the weight associated with the reference cost when
passing the obstacle 302, the vehicle is better able to explore trajectories which may be
further from the reference trajectory 304 to overcome the obstacle 302. By way of
example, a point 326 corresponding to the reference point 324 is allowed to be located
further from the reference point 324 to safely overcome the obstacle 302.
In some examples, the weights associated with the reference points 306, 308, and 324
can be used to determine the points 312, 314, and 326 respectively, and accordingly, can
determine the contours of the target trajectory 310. By way of example, and without
limitation, a weight associated with a reference cost can be decreased when the vehicle is
navigating around the obstacle 302. In some examples, a reference cost can comprise a
cost associated with a difference between a point (e.g., the reference point 308) on the
reference trajectory 304 and a corresponding point (e.g., the point 314) on the target
trajectory 310. In some example, the difference can represent one or more difference in a
yaw, lateral offset, velocity, acceleration, curvature, curvature rate, and the like. In some
examples, decreasing a weight associated with a reference cost can reduce a penalty
associated with the target trajectory 310 being located a distance away from the reference
trajectory 304, which can provide smoother transitions leading towards safer and/or more
comfortable vehicle operations. For example, decreasing the reference costs associated
with a portion of the reference trajectory 304 associated with the start trigger 320 and the
end trigger 322 can lead to increasing a buffer between the vehicle 102 and the obstacle
302 in the environment 300 and/or otherwise allowing the vehicle 102 to explore
trajectories not previously available.
FIG. 4 is an illustration of determining weight(s) associated with a cost associated with

generating a target trajectory based on a reference trajectory when changing lanes, in
accordance with embodiments of the disclosure. More specifically, FIG. 4 illustrates an
example environment 400 in which the vehicle 102 is positioned. FIG. 4 further illustrates
an obstacle 402 in the environment 400, whereby a reference trajectory 404 (comprising
portions 404A and 404B) represents an initial path or trajectory for the vehicle 102 to
change lanes to navigate around the obstacle 402 in the environment 400. In some
examples, the obstacle 402 corresponds to the vehicle 104a of FIG. 1, although the
obstacle 402 may represent any static object or dynamic object in the environment 200.
Further, the reference trajectory 404 may represent an alternative implementation
compared to the reference trajectory 122 of FIG. 1 (e.g., representing a lane change action
from the lane 136 to the lane 138). However, the reference trajectory 404 may represent
any path or trajectory in the environment 400.
In general, the vehicle computing device(s) 112 can evaluate one or more costs at a
reference point 406 to generate a target trajectory 408 having a point 410 corresponding
to the reference point 306. A dotted line between the reference point 406 and the point
410 indicates a corresponding point. Although only one point on the reference trajectory
404 is shown in FIG. 4, it can be understood that a plurality of points on the reference
trajectory 404 can correspond to a plurality of corresponding points on the target
trajectory 408.
In general, the one or more costs may include, but is not limited to a reference cost, an
obstacle cost, a lateral cost, a longitudinal cost, and the like. A cost can be associated with a
weight (e.g., weights 412 and 414) that can increase or decrease a cost associated with a
trajectory or a cost associated with a point of the trajectory. In some examples, the vehicle
computing device(s) 112 can perform operations to decrease a weight associated with a cost
based on actions or maneuvers to be performed by the vehicle.
A start trigger 416 can represent a jump or discontinuity of the reference trajectory 404,
which can represent a lane change action to change from the lane 136 to the lane 138 in
the environment 400. A curve 418 can represent the varying weight(s) of cost(s) over the
reference trajectory 404B, whereby a shape of the curve 418 can be adjusted based on a
lateral distance for the vehicle 102 to traverse with respect to a distance and/or time.
Based at least in part on the start trigger 416 and the curve 418, a weight associated with
the points of the reference trajectory 404 and/or the target trajectory 408 can be adjusted
to increase or decrease the weight. In some examples, the curve 418 can represent a linear
function, quadratic function, stepwise function, and the like, associated with the weight.
For example, a weight associated with the reference point 406 can correspond to the
weight 414. Accordingly, evaluating cost(s) associated with the point 406 can be based at
least in part on the weight 414. Of course, evaluating the costs can be performed for a
plurality of points associated with a plurality of weights, as discussed herein.
In some examples, a lane change action can be associated with other constraints to provide
a smooth and comfortable behavior of the vehicle 102. For example, a detail of the target
trajectory 408 is illustrated as a detail 420 of a portion of the target trajectory 422. In an
example 424, the vehicle computing device(s) 112 can determine whether a lateral
displacement 426 of the vehicle 102 is less than a threshold amount given a longitudinal
displacement 428. By way of example and without limitation, the lateral displacement 426
may be 1 meter (or N meters) and the longitudinal displacement 428 may be 5 meters (or
M meters). In an example 430, the vehicle computing device(s) 112 can determine whether
a lateral displacement 432 of the vehicle 102 is less than a threshold amount given a time
period 434. By way of example and without limitation, the lateral displacement 432 may
be 1 meter (or N′ meters) and the longitudinal displacement 428 may be 2 seconds (or S
seconds). In some examples, a shape of the curve 418 can be adjusted (e.g., to vary weights
more or less rapidly) to ensure that the lateral displacement 426 and 432 of the vehicle
102 is below a threshold. Thus, in some examples, the vehicle computing device(s) 112 can
vary weight(s) associated with cost(s) and can ensure that a lateral displacement of the
vehicle 102 is less than a threshold amount given a distance to travel and given a time
period to determine the target trajectory 408 for the vehicle 102 to follow as the vehicle
102 traverses the environment 400.
FIG. 5A is an illustration of a dilated region, a collision region, and a safety region

associated with a drivable area, in accordance with embodiments of the disclosure. In
particular, FIG. 5 illustrates an example environment 500 in which the vehicle 102 is
positioned in a lane 502 of a road 504. FIG. 5 further illustrates a pedestrian 106a
traversing the environment 500 and the vehicle 104a parked on the side of the road 504.
In some examples, the vehicle computing device(s) 112 can determine a drivable area 506
comprising a plurality of regions, as illustrated in a detail 508. In particular, the drivable
area 506 comprises a dilated region 510, a collision region 512, and a safety region 514.
For example, the dilated region 510 can represent the largest extent of the drivable area
506 and can comprise information about object(s) (e.g., 106a and 104a) in the
environment 500 and probabilistic distances between the boundaries (e.g., defined by the
regions 510, 512, and 514) and the object(s). As a non-limiting example, uncertainties
from measurement and/or movement of the object(s) may be effectively “encoded” into
such boundaries. Further, and in some examples, the collision region 512 can represent a
drivable region smaller than the dilated region 510 representing a region for the vehicle
102 to avoid to further reduce a likelihood that the vehicle 102 will collide with the
pedestrian 106a and the vehicle 104a in the environment 500. In some examples, the
safety region 514 can represent a drivable region smaller than the collision region 512 and
the dilated region 510 to provide a buffer between the vehicle 102 and the pedestrian 106a
and the vehicle 104a in the environment 500. In some examples, the collision region 512
and/or the safety region 514 can also be associated with information about object(s) in the
environment and probabilistic distances between the boundaries and the object(s). In some
examples, the vehicle 102 can evaluate costs based at least in part on distance(s) between
points on a reference trajectory 516 (and/or on a target trajectory) and one or more points
associated with the regions 510, 512, and/or 516, as discussed herein. In some examples,
the cost(s) associated with the region(s) 510, 512, and 514 may vary. For example, a cost
associated with the safety region 514 may be relatively less than a cost associated with the
collision region 512, and vice versa.
FIG. 5B is an illustration of regions associated with a drivable area based at least in part on
a classification of an object in an environment and/or on a velocity of a vehicle in the
environment, in accordance with embodiments of the disclosure. An example 518
illustrates the detail 508 of FIG. 5A with the dilated region 510, the collision region 512,
and the safety region 514.
For example, the safety region 514 represents a safety region while the vehicle 102 is
associated with a velocity V1.
By way of another example, the safety region 514* represents a safety region while the
vehicle 102 is associated with a velocity V2. As the safety regions 514 and 514* can be
associated with information about object(s) in the environment 500 (e.g., classification
information) and probabilistic distances between the boundaries and the object(s), a size of
different portions of the safety regions 514 and 514* can vary based on classification
information of the different portions and a velocity of the vehicle 102. For example, a
portion 520 can represent a portion of the safety regions 514 and 514* associated with the
pedestrian 106a, while the portion 522 can represent a portion of the safety regions 514
and 514* associated with the vehicle 104a. Portions 524, 526, and 528 may not be
associated with the objects 106a or 104a, or may be associated with the extents of the road
504.
In some examples, a size of the safety regions 514 and 514* can be based at least in part on
classification information of the associated object and/or on a velocity of the vehicle 102 in
the environment. For example, a distance 530 associated with the portion 520 of the
safety region 514* can represent an increase in a size of the safety region 514 for the
portion associated with the pedestrian 106a while the vehicle 102 is at a velocity V2. A
distance 532 associated with the portion 522 of the safety region 514* can represent an
increase in a size of the safety region 514 for the portion associated with the vehicle 104a
while the vehicle 102 is at a velocity V2. In some examples, the distance 530 can be larger
than the distance 532. In some examples, the velocity V2 can be higher than the velocity V1.
In some examples, varying a size of the safety regions 514 and 514* can further include
modifying an extent of the drivable area to generate a modified drivable area.
In some examples, the portions 524, 526, and/or 528 of the safety region 514 that are not
associated with the pedestrian 106a or the vehicle 104a may or may not change based on a
velocity of the vehicle 102.
FIG. 6 is an illustration of updating a region based on identifying an object such as a

bicyclist proximate to the vehicle, in accordance with embodiments of the disclosure. In
particular, FIG. 6 illustrates an example environment 600 in which the vehicle 102 is
positioned in a lane 602 of a road 604. FIG. 6 further illustrates a pedestrian 106a
traversing the environment 500, the vehicle 104a parked on the side of the road 604, and
a bicyclist 606 (e.g., in a bicycle lane) proximate to the vehicle 102.
A detail 608 illustrates the vehicle 102, the bicyclist 606 proximate to the vehicle 102,
and a drivable area 610 comprising a plurality of regions. For simplicity of discussion, a
lane boundary 612 and a lane bias region 614 are illustrated in FIG. 6, although it may be
understood that additional regions may comprise a dilated region, a collision region, a
safety region, and the like. In some examples, the vehicle computing device(s) 112 can
determine the lane bias region 614 based at least in part on the bicyclist 606 (and/or a
bicycle lane, parking lane, parked car, and the like) being proximate to the vehicle 102 in
the environment 600.
For example, the vehicle computing device(s) 112 can determine a predicted trajectory for
the bicyclist 606 to determine that the bicyclist 606 may pass the vehicle 102, and
accordingly, can expand a size of the lane boundary 612 to generate the lane bias region
614. Accordingly, for a given reference trajectory 616, the vehicle computing device(s) 112
can generate a target trajectory 618 that biases a location of the vehicle 102 away from the
bicyclist 606 (and/or bicycle lane) to allow the bicyclist to pass the vehicle 102 with more
comfortable space. In some examples, a determination of whether to generate the lane bias
region 614 can be based at least in part on one or more of a velocity of the bicyclist 606, a
velocity of the bicyclist 606 relative to a velocity of the vehicle 102, a velocity of the vehicle
102 being below a threshold, a position of the bicyclist 606 in the lane 602 of the road
604, a position of the vehicle 102 and/or information from a map, and the like. In this
manner, the vehicle computing device(s) 112 can evaluate cost(s) based on the lane bias
region 614 to generate the target trajectory 620 that provides additional space for the
bicyclist 606 to pass the vehicle 102.
FIG. 7 depicts a block diagram of an example system 700 for implementing the techniques
described herein. In at least one example, the system 700 can include a vehicle 702.
The vehicle 702 can include a vehicle computing device 704, one or more sensor systems
706, one or more emitters 708, one or more communication connections 710, at least one
direct connection 712, and one or more drive systems 714.
The vehicle computing device 704 can include one or more processors 716 and memory
718 communicatively coupled with the one or more processors 716. In the illustrated
example, the vehicle 702 is an autonomous vehicle; however, the vehicle 702 could be any
other type of vehicle. In the illustrated example, the memory 718 of the vehicle computing
device 704 stores a localization component 720, a perception component 722, one or
more maps 724, one or more system controllers 726, and a planning component 728
comprising the drivable area component 114, the trajectory generation component 116, the
cost(s) component 118, a safety region component 730, a trajectory sampling component
732, and a weight(s) component 734. Though depicted in FIG. 7 as residing in memory
718 for illustrative purposes, it is contemplated that the localization component 720, the
perception component 722, the one or more maps 724, the one or more system controllers
726, the planning component 728, the drivable area component 114, the trajectory
generation component 116, the cost(s) component 118, the safety region component 730,
the trajectory sampling component 732, and the weight(s) component 734 may
additionally, or alternatively, be accessible to the vehicle 702 (e.g., stored remotely).
In at least one example, the localization component 720 can include functionality to
receive data from the sensor system(s) 706 to determine a position and/or orientation of
the vehicle 702 (e.g., one or more of an x-, y-, z-position, roll, pitch, or yaw). For example,
the localization component 720 can include and/or request/receive a map of an
environment and can continuously determine a location and/or orientation of the
autonomous vehicle within the map. In some instances, the localization component 720
can utilize SLAM (simultaneous localization and mapping), CLAMS (calibration,
localization and mapping, simultaneously), relative SLAM, bundle adjustment, non-linear
least squares optimization, or the like to receive image data, LIDAR data, RADAR data,
IMU data, GPS data, wheel encoder data, and the like to accurately determine a location of
the autonomous vehicle. In some instances, the localization component 720 can provide
data to various components of the vehicle 702 to determine an initial position of an
autonomous vehicle for generating a trajectory and/or for generating map data, as
discussed herein.
In some instances, the perception component 722 can include functionality to perform
object detection, segmentation, and/or classification. In some examples, the perception
component 722 can provide processed sensor data that indicates a presence of an entity
that is proximate to the vehicle 702 and/or a classification of the entity as an entity type
(e.g., car, pedestrian, cyclist, animal, building, tree, road surface, curb, sidewalk, unknown,
etc.). In additional or alternative examples, the perception component 722 can provide
processed sensor data that indicates one or more characteristics associated with a detected
entity (e.g., a tracked object) and/or the environment in which the entity is positioned. In
some examples, characteristics associated with an entity can include, but are not limited to,
an x-position (global and/or local position), a y-position (global and/or local position), a z-
position (global and/or local position), an orientation (e.g., a roll, pitch, yaw), an entity
type (e.g., a classification), a velocity of the entity, an acceleration of the entity, an extent of
the entity (size), etc. Characteristics associated with the environment can include, but are
not limited to, a presence of another entity in the environment, a state of another entity in
the environment, a time of day, a day of a week, a season, a weather condition, an
indication of darkness/light, etc.
The memory 718 can further include one or more maps 724 that can be used by the vehicle
702 to navigate within the environment. For the purpose of this discussion, a map can be
any number of data structures modeled in two dimensions, three dimensions, or N-
dimensions that are capable of providing information about an environment, such as, but
not limited to, topologies (such as intersections), streets, mountain ranges, roads, terrain,
and the environment in general. In some instances, a map can include, but is not limited
to: texture information (e.g., color information (e.g., RGB color information, Lab color
information, HSV/HSL color information), and the like), intensity information (e.g.,
LIDAR information, RADAR information, and the like); spatial information (e.g., image
data projected onto a mesh, individual “surfels” (e.g., polygons associated with individual
color and/or intensity)), reflectivity information (e.g., specularity information,
retroreflectivity information, BRDF information, BSSRDF information, and the like). In
one example, a map can include a three-dimensional mesh of the environment. In some
instances, the map can be stored in a tiled format, such that individual tiles of the map
represent a discrete portion of an environment, and can be loaded into working memory as
needed, as discussed herein. In at least one example, the one or more maps 724 can
include at least one map (e.g., images and/or a mesh). In some examples, the vehicle 702
can be controlled based at least in part on the maps 724. That is, the maps 724 can be used
in connection with the localization component 720, the perception component 722,
and/or the planning component 728 to determine a location of the vehicle 702, identify
objects in an environment, and/or generate routes and/or trajectories to navigate within an
environment.
In some examples, the one or more maps 724 can be stored on a remote computing
device(s) (such as the computing device(s) 738) accessible via network(s) 736. In some
examples, multiple maps 724 can be stored based on, for example, a characteristic (e.g.,
type of entity, time of day, day of week, season of the year, etc.). Storing multiple maps 724
can have similar memory requirements, but increase the speed at which data in a map can
be accessed.
In at least one example, the vehicle computing device 704 can include one or more system
controllers 726, which can be configured to control steering, propulsion, braking, safety,
emitters, communication, and other systems of the vehicle 702. These system controller(s)
726 can communicate with and/or control corresponding systems of the drive system(s)
714 and/or other components of the vehicle 702.
In general, the planning component 728 can determine a path for the vehicle 702 to follow
to traverse through an environment. For example, the planning component 728 can
determine various routes and trajectories and various levels of detail. For example, the
planning component 728 can determine a route to travel from a first location (e.g., a
current location) to a second location (e.g., a target location). For the purpose of this
discussion, a route can be a sequence of waypoints for travelling between two locations. As
non-limiting examples, waypoints include streets, intersections, global positioning system
(GPS) coordinates, distributed points along a line connecting the first location to the
second location constrained by a map and/or map information, etc. Further, the planning
component 728 can generate an instruction for guiding the autonomous vehicle along at
least a portion of the route from the first location to the second location. In at least one
example, the planning component 728 can determine how to guide the autonomous
vehicle from a first waypoint in the sequence of waypoints to a second waypoint in the
sequence of waypoints. In some examples, the instruction can be a trajectory, or a portion
of a trajectory. In some examples, multiple trajectories can be substantially simultaneously
generated (e.g., within technical tolerances) in accordance with a receding horizon
technique, wherein one of the multiple trajectories is selected for the vehicle 702 to
navigate.
In some instances, the planning component 728 can include a prediction component to
generate predicted trajectories of objects in an environment. For example, a prediction
component can generate one or more predicted trajectories for vehicles, pedestrians,
animals, and the like within a threshold distance from the vehicle 702. In some instances,
a prediction component can measure a trace of an object and generate a trajectory for the
object based on observed and predicted behavior. Examples of generating predicted
trajectories are discussed in U.S. patent application Ser. No. 16/151,607, filed Oct. 4, 2018
and Ser. No. 15/982,658, filed May 17, 2018. Application Ser. Nos. 16/151,607 and
15/982,658 are herein incorporated by reference, in their entirety.
In some instances, the drivable area component 114 can include functionality to generate
and/or determine a drivable area in an environment. For example, the drivable area
component 114 can receive sensor data from the sensor system(s) 706 and/or can receive
information about obstacles and/or objects in the environment from the perception
component 722. Based at least in part on the sensor data and/or on information associated
with the objects (e.g., location, pose, extent, classification, velocity, predicted trajectories,
etc.) the drivable area component 114 can determine a drivable area comprising one or
more regions, including but not limited to a dilated region, a collision region, and a safety
region, as discussed herein. In some instances, the drivable area component 114 can
determine an extent of the region(s) based on a classification type associated with objects
proximate to the region(s) and/or based on a velocity of the vehicle 702. Additional details
of the drivable area component 114 are discussed throughout this disclosure.
In some instances, the trajectory generation component 116 can include functionality to
generate a reference trajectory and/or a target trajectory within the drivable area. For
example, the trajectory generation component 116 can receive or determine a reference
trajectory, which can correspond to a centerline of a road segment or other path through an
environment. In some instances, the trajectory generation component 116 can generate
segments that can correspond to a motion primitive generated in accordance with the
techniques discussed in U.S. patent application Ser. No. 15/843,596, filed Dec. 15, 2017.
Further, a segment defined by a desired curvature can be generated in accordance with the
techniques discussed in U.S. patent application Ser. No. 15/843,512, filed Dec. 15, 2017.
Application Ser. Nos. 15/843,596 and 15/843,512 are herein incorporated by reference, in
their entirety. Additional details of the trajectory generation component 116 are discussed
throughout this disclosure.
In some instances, the cost(s) component 118 can include functionality to evaluate one or
more costs to generate a target trajectory with respect to the reference trajectory. As
discussed above, the one or more costs may include, but is not limited to, a reference cost,
an obstacle cost, a lateral cost, and longitudinal cost, and the like. In some examples, one or
more costs can be evaluated in accordance with the techniques discussed in U.S. patent
application Ser. No. 16/147,492, filed Sep. 28, 2018, which is hereby incorporated by
reference, in its entirety. Additional details of the cost(s) component 118 are discussed
throughout this disclosure.
In some instances, the safety region component 730 can include functionality to generate
an updated safety region based at least in part on a classification of objects in an
environment and/or on a velocity of the vehicle 702 traversing the environment. As
illustrated in FIGS. 5A, 5B, and 6, the safety region component 730 can determine a size of
a safety region (or other region), which can be used by the cost(s) component 118 in
generating a target trajectory. Additional details of the safety region component 730 are
discussed throughout this disclosure.
In some instances, the trajectory sampling component 732 can include functionality to
determine a point density for portions of the reference trajectory based at least in part on a
cost associated with a curvature value associated with a portion of the reference trajectory
and/or based at least in part on a cost associated with a distance between a point on the
reference trajectory and a point associated with an obstacle in the environment. As
illustrated in FIG. 2, the trajectory sampling component 732 can determine a density of
points for a reference trajectory when a cost associated with a curvature value is relatively
high and/or when cost associated with a distance between a point on the reference
trajectory and a point associated with an obstacle in the environment is relatively high
(e.g., indicating that an object in the environment is located relatively close to a reference
trajectory). Increasing the number of points around a region of high activity or a region of
interest allows for cost(s) to be evaluated at more points, thereby ensuring a safe and
comfortable ride for the vehicle 702. Additional details of the trajectory sampling
component 732 are discussed throughout this disclosure.
In some instances, the weight(s) component 734 can include functionality to select or
determine weight(s) associated with evaluating one or more cost functions when
generating a target trajectory. For example, and as illustrated in FIGS. 3 and 4, an action
start (such as the start of a double-parked vehicle action or a lane change action) can vary a
weight of a cost in order to allow the planning component 728 generate a target trajectory
with more flexibility for the vehicle 702 to traverse the environment. Further, weight(s)
can be adjusted in combination with other operations (e.g., when expanding a lane
boundary based on a proximity of a bicyclist, bike lane, parking lane, parked car, and the
like). Additional details of the weight(s) component 734 are discussed throughout this
disclosure.
In some instances, aspects of some or all of the components discussed herein can include
any models, algorithms, and/or machine learning algorithms. For example, in some
instances, the components in the memory 718 (and the memory 742, discussed below) can
be implemented as a neural network.
As described herein, an exemplary neural network is a biologically inspired algorithm

which passes input data through a series of connected layers to produce an output. Each
layer in a neural network can also comprise another neural network, or can comprise any
number of layers (whether convolutional or not). As can be understood in the context of
this disclosure, a neural network can utilize machine learning, which can refer to a broad
class of such algorithms in which an output is generated based on learned parameters.
Although discussed in the context of neural networks, any type of machine learning can be
used consistent with this disclosure. For example, machine learning algorithms can
include, but are not limited to, regression algorithms (e.g., ordinary least squares
regression (OLSR), linear regression, logistic regression, stepwise regression, multivariate
adaptive regression splines (MARS), locally estimated scatterplot smoothing (LOESS)),
instance-based algorithms (e.g., ridge regression, least absolute shrinkage and selection
operator (LASSO), elastic net, least-angle regression (LARS)), decisions tree algorithms
(e.g., classification and regression tree (CART), iterative dichotomiser 3 (ID3), Chi-squared
automatic interaction detection (CHAID), decision stump, conditional decision trees),
Bayesian algorithms (e.g., naïve Bayes, Gaussian naïve Bayes, multinomial naïve Bayes,
average one-dependence estimators (AODE), Bayesian belief network (BNN), Bayesian
networks), clustering algorithms (e.g., k-means, k-medians, expectation maximization
(EM), hierarchical clustering), association rule learning algorithms (e.g., perceptron, back-
propagation, hopfield network, Radial Basis Function Network (RBFN)), deep learning
algorithms (e.g., Deep Boltzmann Machine (DBM), Deep Belief Networks (DBN),
Convolutional Neural Network (CNN), Stacked Auto-Encoders), Dimensionality Reduction
Algorithms (e.g., Principal Component Analysis (PCA), Principal Component Regression
(PCR), Partial Least Squares Regression (PLSR), Sammon Mapping, Multidimensional
Scaling (MDS), Projection Pursuit, Linear Discriminant Analysis (LDA), Mixture
Discriminant Analysis (MDA), Quadratic Discriminant Analysis (QDA), Flexible
Discriminant Analysis (FDA)), Ensemble Algorithms (e.g., Boosting, Bootstrapped
Aggregation (Bagging), AdaBoost, Stacked Generalization (blending), Gradient Boosting
Machines (GBM), Gradient Boosted Regression Trees (GBRT), Random Forest), SVM
(support vector machine), supervised learning, unsupervised learning, semi-supervised
learning, etc.
Additional examples of architectures include neural networks such as ResNet70,

ResNet101, VGG, DenseNet, PointNet, and the like.
In at least one example, the sensor system(s) 706 can include LIDAR sensors, RADAR
sensors, ultrasonic transducers, sonar sensors, location sensors (e.g., GPS, compass, etc.),
inertial sensors (e.g., inertial measurement units (IMUs), accelerometers, magnetometers,
gyroscopes, etc.), cameras (e.g., RGB, IR, intensity, depth, etc.), time of flight sensors,
microphones, wheel encoders, environment sensors (e.g., temperature sensors, humidity
sensors, light sensors, pressure sensors, etc.), etc. The sensor system(s) 706 can include
multiple instances of each of these or other types of sensors. For instance, the LIDAR
sensors can include individual LIDAR sensors located at the corners, front, back, sides,
and/or top of the vehicle 702. As another example, the camera sensors can include
multiple cameras disposed at various locations about the exterior and/or interior of the
vehicle 702. The sensor system(s) 706 can provide input to the vehicle computing device
704. Additionally or alternatively, the sensor system(s) 706 can send sensor data, via the
one or more networks 736, to the one or more computing device(s) 738 at a particular
frequency, after a lapse of a predetermined period of time, in near real-time, etc.
The vehicle 702 can also include one or more emitters 708 for emitting light and/or
sound, as described above. The emitters 708 in this example include interior audio and
visual emitters to communicate with passengers of the vehicle 702. By way of example and
not limitation, interior emitters can include speakers, lights, signs, display screens, touch
screens, haptic emitters (e.g., vibration and/or force feedback), mechanical actuators (e.g.,
seatbelt tensioners, seat positioners, headrest positioners, etc.), and the like. The emitters
708 in this example also include exterior emitters. By way of example and not limitation,
the exterior emitters in this example include lights to signal a direction of travel or other
indicator of vehicle action (e.g., indicator lights, signs, light arrays, etc.), and one or more
audio emitters (e.g., speakers, speaker arrays, horns, etc.) to audibly communicate with
pedestrians or other nearby vehicles, one or more of which comprising acoustic beam
steering technology.
The vehicle 702 can also include one or more communication connection(s) 710 that
enable communication between the vehicle 702 and one or more other local or remote
computing device(s). For instance, the communication connection(s) 710 can facilitate
communication with other local computing device(s) on the vehicle 702 and/or the drive
system(s) 714. Also, the communication connection(s) 710 can allow the vehicle to
communicate with other nearby computing device(s) (e.g., other nearby vehicles, traffic
signals, etc.). The communications connection(s) 710 also enable the vehicle 702 to
communicate with a remote teleoperations computing device or other remote services.
The communications connection(s) 710 can include physical and/or logical interfaces for
connecting the vehicle computing device 704 to another computing device or a network,
such as network(s) 736. For example, the communications connection(s) 710 can enable
Wi-Fi-based communication such as via frequencies defined by the IEEE 802.11 standards,
short range wireless frequencies such as Bluetooth, cellular communication (e.g., 2G, 3G,
4G, 4G LTE, 5G, etc.) or any suitable wired or wireless communications protocol that
enables the respective computing device to interface with the other computing device(s).
In at least one example, the vehicle 702 can include one or more drive systems 714. In
some examples, the vehicle 702 can have a single drive system 714. In at least one
example, if the vehicle 702 has multiple drive systems 714, individual drive systems 714
can be positioned on opposite ends of the vehicle 702 (e.g., the front and the rear, etc.). In
at least one example, the drive system(s) 714 can include one or more sensor systems to
detect conditions of the drive system(s) 714 and/or the surroundings of the vehicle 702. By
way of example and not limitation, the sensor system(s) can include one or more wheel
encoders (e.g., rotary encoders) to sense rotation of the wheels of the drive systems, inertial
sensors (e.g., inertial measurement units, accelerometers, gyroscopes, magnetometers,
etc.) to measure orientation and acceleration of the drive system, cameras or other image
sensors, ultrasonic sensors to acoustically detect objects in the surroundings of the drive
system, LIDAR sensors, RADAR sensors, etc. Some sensors, such as the wheel encoders
can be unique to the drive system(s) 714. In some cases, the sensor system(s) on the drive
system(s) 714 can overlap or supplement corresponding systems of the vehicle 702 (e.g.,
sensor system(s) 706).
The drive system(s) 714 can include many of the vehicle systems, including a high voltage
battery, a motor to propel the vehicle, an inverter to convert direct current from the battery
into alternating current for use by other vehicle systems, a steering system including a
steering motor and steering rack (which can be electric), a braking system including
hydraulic or electric actuators, a suspension system including hydraulic and/or pneumatic
components, a stability control system for distributing brake forces to mitigate loss of
traction and maintain control, an HVAC system, lighting (e.g., lighting such as head/tail
lights to illuminate an exterior surrounding of the vehicle), and one or more other systems
(e.g., cooling system, safety systems, onboard charging system, other electrical components
such as a DC/DC converter, a high voltage j unction, a high voltage cable, charging system,
charge port, etc.). Additionally, the drive system(s) 714 can include a drive system
controller which can receive and preprocess data from the sensor system(s) and to control
operation of the various vehicle systems. In some examples, the drive system controller can
include one or more processors and memory communicatively coupled with the one or
more processors. The memory can store one or more components to perform various
functionalities of the drive system(s) 714. Furthermore, the drive system(s) 714 also
include one or more communication connection(s) that enable communication by the
respective drive system with one or more other local or remote computing device(s).
In at least one example, the direct connection 712 can provide a physical interface to
couple the one or more drive system(s) 714 with the body of the vehicle 702. For example,
the direct connection 712 can allow the transfer of energy, fluids, air, data, etc. between the
drive system(s) 714 and the vehicle. In some instances, the direct connection 712 can
further releasably secure the drive system(s) 714 to the body of the vehicle 702.
In some examples, the vehicle 702 can send sensor data to one or more computing
device(s) 738 via the network(s) 736. In some examples, the vehicle 702 can send raw
sensor data to the computing device(s) 738. In other examples, the vehicle 702 can send
processed sensor data and/or representations of sensor data to the computing device(s)
738. In some examples, the vehicle 702 can send sensor data to the computing device(s)
738 at a particular frequency, after a lapse of a predetermined period of time, in near real-
time, etc. In some cases, the vehicle 702 can send sensor data (raw or processed) to the
computing device(s) 738 as one or more log files.
The computing device(s) 738 can include processor(s) 740 and a memory 742 storing a
planning component 744.
In some instances, the planning component 744 can substantially correspond to the
planning component 728 and can include functionality to generate trajectories for the
vehicle 702 in an environment.
The processor(s) 716 of the vehicle 702 and the processor(s) 740 of the computing
device(s) 738 can be any suitable processor capable of executing instructions to process
data and perform operations as described herein. By way of example and not limitation, the
processor(s) 716 and 740 can comprise one or more Central Processing Units (CPUs),
Graphics Processing Units (GPUs), or any other device or portion of a device that processes
electronic data to transform that electronic data into other electronic data that can be
stored in registers and/or memory. In some examples, integrated circuits (e.g., ASICs, etc.),
gate arrays (e.g., FPGAs, etc.), and other hardware devices can also be considered
processors in so far as they are configured to implement encoded instructions.
Memory 718 and 742 are examples of non-transitory computer-readable media. The
memory 718 and 742 can store an operating system and one or more software
applications, instructions, programs, and/or data to implement the methods described
herein and the functions attributed to the various systems. In various implementations, the
memory can be implemented using any suitable memory technology, such as static random
access memory (SRAM), synchronous dynamic RAM (SDRAM), nonvolatile/Flash-type
memory, or any other type of memory capable of storing information. The architectures,
systems, and individual elements described herein can include many other logical,
programmatic, and physical components, of which those shown in the accompanying
figures are merely examples that are related to the discussion herein.
In some instances, the memory 718 and 742 can include at least a working memory and a
storage memory. For example, the working memory may be a high-speed memory of
limited capacity (e.g., cache memory) that is used for storing data to be operated on by the
processor(s) 716 and 740. In some instances, the memory 718 and 742 can include a
storage memory that may be a lower-speed memory of relatively large capacity that is used
for long-term storage of data. In some cases, the processor(s) 716 and 740 cannot operate
directly on data that is stored in the storage memory, and data may need to be loaded into a
working memory for performing operations based on the data, as discussed herein.
It should be noted that while FIG. 7 is illustrated as a distributed system, in alternative

examples, components of the vehicle 702 can be associated with the computing device(s)
738 and/or components of the computing device(s) 738 can be associated with the vehicle
702. That is, the vehicle 702 can perform one or more of the functions associated with the
computing device(s) 738, and vice versa.
FIGS. 8 and 9 illustrate example processes in accordance with embodiments of the

disclosure. These processes are illustrated as logical flow graphs, each operation of which
represents a sequence of operations that can be implemented in hardware, software, or a
combination thereof. In the context of software, the operations represent computer-
executable instructions stored on one or more computer-readable storage media that, when
executed by one or more processors, perform the recited operations. Generally, computer-
executable instructions include routines, programs, objects, components, data structures,
and the like that perform particular functions or implement particular abstract data types.
The order in which the operations are described is not intended to be construed as a
limitation, and any number of the described operations can be combined in any order
and/or in parallel to implement the processes.
FIG. 8 depicts an example process 800 for determining a point density for points
associated with a reference trajectory and evaluating costs at the points to generate a target
trajectory, in accordance with embodiments of the disclosure. For example, some or all of
the process 800 can be performed by one or more components in FIG. 1 or 7, as described
herein. For example, some or all of the process 800 can be performed by the vehicle
computing device(s) 112 and/or the computing device(s) 704.
At operation 802, the process can include receiving a reference trajectory representing an
initial trajectory for an autonomous vehicle to traverse an environment. In some instances,
the reference trajectory can comprise a segment of a continuous trajectory or a piecewise
trajectory, depending on an action to be performed and a path to be followed in an
environment. In some examples, a reference trajectory can be based at least in part on
static and/or dynamic obstacles in an environment, a minimum or maximum longitudinal

acceleration or velocity, a maximum steering angle, vehicle dynamics (e.g., a notion that
the wheels do not slip, etc.).
At operation 804, the process can include determining a first point density of first points
associated with a first portion of the reference trajectory. For example, the operation 804
can include determining a cost associated with a curvature value associated with a portion
of the reference trajectory and/or a cost associated with a distance to a point associated
with an obstacle that is below a threshold value. In some examples, the operation 804 can
include accessing a lookup table to determine a point density based on the cost and/or to
access a function to determine a point density. For example, a first cost below a first
threshold (e.g., 0.1 or x) a first point density can be selected as the point density for a
region. Further, where a second cost is above the first threshold (e.g., 0.1 or y) but below a
second threshold (e.g., 0.5 or y), a second point density can be based on linearly
interpolating between density values. Further, where a third cost is above the second
threshold (e.g., 0.5 or y), a third point density can be selected as the point density for a
region. Of course, costs can be selected in a variety of manners.
Further, in some examples, costs associated with a curvature and/or distance can be
performed by a second layer of a planner system, as discussed above. For example, the
second layer can receive a first target trajectory from the first layer of the planner system,
whereby the first target trajectory can include curvature values. Further, the operation 804
can include evaluating the first target trajectory over time with respect to objects in an
environment to determine an initial cost associated with a distance between a point on the
first target trajectory (e.g., a reference trajectory) and a point associated with an obstacle in
the environment and can subsequently determine the point density based at least in part
on the cost.
At operation 806, the process can include determining a second point density of second
points associated with a second portion of the reference trajectory. For example, the
operation 806 can include determining a region associated with a low curvature value
(e.g., a curvature value below a threshold) and/or associated with a distance to a point
associated with an obstacle that meets or exceeds a threshold value.
At operation 808, the process can include evaluating cost function(s) at a first point of the
first points and at a second point of the second points to generate a target trajectory. For
example, the operation 808 can include evaluating one or more of a reference cost, an
obstacle cost, a lateral cost, a longitudinal cost, and the like.
At operation 810, the process can include determining whether cost(s) are below a
threshold. For example, the operation 810 can include using technique(s) involving a
gradient descent to optimize the cost function(s). In some examples, the operation 810 can
include determining whether a cost is below a threshold value, determining whether a
change in cost between optimization operations is below a threshold value, and/or
determining whether a number of optimization iterations has been performed. If the costs
meet or exceed a threshold (e.g., “no” in the operation 810) the process returns to the
operation 808 where another trajectory can be generated (e.g., by varying locations of the
points and/or controls (e.g., steering angles, acceleration, etc.)) and updated costs can be
determined. If the costs are below a threshold (e.g., “yes” in the operation 810), the process
can continue to operation 812.
At operation 812, the process can include controlling the autonomous vehicle to traverse
the environment based at least in part on the target trajectory. In some instances, the target
trajectory can be provided to a trajectory smoother component and/or the trajectory
tracker component to refine the target trajectory and/or to generate control signals for the
various motors and steering actuators of the autonomous vehicle.
FIG. 9 depicts an example process 900 for determining weights associated with points
along a reference trajectory (e.g., those points determined in accordance with FIG. 8) and
evaluating, based on the weights, costs at the points to determine a target trajectory, in
accordance with embodiments of the disclosure. For example, some or all of the process
900 can be performed by one or more components in FIG. 1 or 7, as described herein. For
example, some or all of the process 900 can be performed by the vehicle computing
device(s) 112 and/or the computing device(s) 704.
At operation 902, the process can include receiving a reference trajectory representing an
initial path for an autonomous vehicle to traverse an environment. In some examples, this
operation 902 can substantially correspond to the operation 802.
At operation 904, the process can include determining portion(s) of the reference
trajectory. For example, the operation 904 can include determining a first portion 906, a
second portion 908, and an N-th portion 910. In some examples, the portions 906, 908,
and 910 can be discrete portions of (and/or points along) the reference trajectory. In some
examples, the operation 904 can include determining the portions based at least in part on
actions or maneuvers to be performed by the autonomous vehicle. For example, an action
or maneuver may include, but is not limited to, a double-parked vehicle action or a lane
change action. In some examples, the operation 904 can include determining that the
autonomous vehicle is proximate to a bicyclist or another object in the environment and
determining a portion of the reference trajectory based at least in part on a predicted

trajectory of the bicyclist. In some examples, the operation 904 can include determining a
location associated with the vehicle based on map data that may indicate weight(s) or a
trigger to vary weights, as discussed herein.
At operation 912, the process can include determining weight(s) associated with reference
cost(s) to be evaluated at point(s) associated with the portion(s) of the reference trajectory.
In some cases, a reference cost is associated with a difference between a point on the
reference trajectory and a corresponding point. For example, the operation 912 can include
determining a first weight 914, a second weight 916, and an M-th weight 918. In some
examples, the operation 912 can include determining a weight that corresponds to
individual portions as determined in the operation 904. In some examples, the operation
912 can include determining a weight that corresponds to individual points (e.g., reference
points) associated with individual portions. That is, if the portion 906 comprises 10 points,
the operation 912 can include determining a unique weight for cost(s) associated with each
of the 10 points. Of course, other configurations are contemplated herein.
At operation 920, the process can include evaluating a reference cost function at the
point(s) to generate a target trajectory comprising the corresponding point. In some
examples, the operation 920 can include scaling a cost by a weight as determined in the
operation 912. For example, the operation 920 can include scaling a cost up or down
based on the weight associated with the cost, which, in turn, can lead to changes in the
contour of the target trajectory based on the weights associated with the various costs, as
discussed herein. In some instances, the operation 920 can include evaluating other cost
functions, and is not limited to evaluating a reference costs function. Thus, the operations
provide a flexible framework to generate trajectories
At operation 922, the process can include controlling the autonomous vehicle to traverse
the environment based at least in part on the target trajectory. In some examples, this
operation 922 can substantially correspond to the operation 812.
Example Clauses
A: A system comprising: one or more processors; and one or more computer-readable
media storing instructions executable by the one or more processors, wherein the
instructions, when executed, cause the system to perform operations comprising: receiving
a reference trajectory representing an initial trajectory for an autonomous vehicle to
traverse an environment; determining a first point density of first points associated with a
first portion of the reference trajectory; determining a second point density of second
points associated with a second portion of the reference trajectory, wherein the second
point density is less than the first point density; evaluating a cost function at a first point of
the first points and at a second point of the second points to generate a target trajectory
comprising first corresponding points and second corresponding points, wherein a first
corresponding point of the first corresponding points corresponds to the first point of the
first points, and wherein a second corresponding point of the second corresponding points
corresponds to a second point of the second points; and controlling the autonomous
vehicle to traverse the environment based at least in part on the target trajectory.
B: The system of paragraph A, wherein the first point density is based, at least in part, on
one or more of: a time period between adjacent points of the first points; or a distance
between the adjacent points of the first points.
C: The system of paragraph A or B, wherein the cost function comprises a cost based at
least in part on a distance between the first corresponding point of the target trajectory and
a point associated with an obstacle in the environment.
D: The system of any of paragraphs A-C, wherein the first point density is based at least in
part on determining that at least one of: a first cost associated with a curvature value
associated with the first point of the reference trajectory; or a second cost associated with a
distance between the first point and a point associated with an obstacle in the environment.
E: The system of paragraph D, wherein determining the first point density comprises
accessing a lookup table or a function to determine the first point density based at least in
part on the first cost or the second cost.
F: The system of any of paragraphs A-E, wherein a first time period associated with the first
point and an adjacent first point of the first points is less than a second time period
associated with the second point and an adjacent second point of the second points.
G: A method comprising: receiving a first trajectory representing an initial trajectory for a

vehicle to traverse an environment; determining a point density for points associated with
a portion of the first trajectory based at least in part on a first cost associated with a
curvature value of the portion or a second cost associated with a distance between a first
point of the portion and a second point associated with an obstacle in the environment;
determining, based at least on a cost associated with the first point, a second trajectory;
and controlling the vehicle to traverse the environment based at least in part on the second
trajectory.
H: The method of paragraph G, wherein at least one of: the point density increases as the
first cost associated with the curvature value increases; or the point density increases as the
second cost associated with the distance between a first point and the second point
increases.
I: The method of paragraph G or H, wherein: the point density is a first point density; the
portion is a first portion; the points are first points; the curvature value is a first curvature
value; the distance is a first distance; the obstacle is a first obstacle; and the method further
comprises: determining a second point density for second points associated with a second
portion the first trajectory based at least in part on a third cost associated with a second
curvature value of the second portion or a fourth cost associated with a second distance
between a third point of the second portion and a fourth point associated with a second
obstacle in the environment, wherein the first point density is different than the second
point density.
J: The method of any of paragraphs G-I, wherein the distance is a first distance, and
wherein the point density is based at least in part on one of: a second distance between the
first point on the portion and an adjacent first point on the portion; or a time period
associated with a difference in a first time associated with the first point and a second time
associated with the adjacent first point.
K: The method of any of paragraphs G-J, wherein the cost is based at least in part on the
distance between the first point of the portion and the second point associated with the
obstacle in the environment.
L: The method of any of paragraphs G-K, wherein determining the second trajectory
further comprises: evaluating the cost for a plurality of points associated with the first
trajectory to generate the second trajectory.
M: The method of any of paragraphs G-L, further comprising: receiving the first trajectory
from a first layer of a planning system of the vehicle, wherein the first trajectory is
generated with respect to a distance between points on the first trajectory and the second
trajectory is generated, by a second layer of the planning system, with respect to a time
period between points on the second trajectory.
N: The method of paragraph M, wherein the point density is non-uniform with respect to
the portion of the first trajectory.
O: A non-transitory computer-readable medium storing instructions that, when executed,

cause one or more processors to perform operations comprising: receiving a first trajectory
representing an initial trajectory for a vehicle to traverse an environment; determining a

point density for points associated with a portion of the first trajectory based at least in part
on a first cost associated with a curvature value of the portion or a second cost associated
with a distance between a first point of the portion and a second point associated with an
obstacle in the environment; generating, based at least in part on a cost associated with the
first point, a second trajectory, the second trajectory comprising a target point; and
controlling the vehicle to traverse the environment based at least in part on the second
trajectory.
P: The non-transitory computer-readable medium of paragraph O, wherein at least one of:

the point density increases as the first cost associated with the curvature value increases; or
the point density increases as the second cost associated with the distance between a first
point and the second point increases.
Q: The non-transitory computer-readable medium of paragraph O or P, wherein: the point

density is a first point density; the portion is a first portion; the points are first points; the
curvature value is a first curvature value; the distance is a first distance; the obstacle is a
first obstacle; and the operations further comprise: determining a second point density for
second points associated with a second portion the first trajectory based at least in part on
a third cost associated with a second curvature value of the second portion or a fourth cost
associated with a second distance between a third point of the second portion and a fourth
point associated with a second obstacle in the environment, wherein the first point density
is different than the second point density.
R: The non-transitory computer-readable medium of any of paragraphs O-Q, wherein the

distance is a first distance, and wherein the cost is based at least in part on a second
distance between the target point of the second trajectory and the second point associated
with the obstacle in the environment.
S: The non-transitory computer-readable medium of any of paragraphs O-R, wherein a

first time period associated with the first point and an adjacent first point of the first points
is less than a second time period associated with the second point and an adjacent second
point of the second points.
T: The non-transitory computer-readable medium of paragraph S, the operations further

comprising: receiving the first trajectory from a first layer of a planning system of the
vehicle, wherein the first trajectory is generated with respect to a Euclidian distance
between points on the first trajectory and the second trajectory is generated, by a second
layer of the planning system, with respect to a time period between points on the second
trajectory.
U: A system comprising: one or more processors; and one or more computer-readable

media storing instructions executable by the one or more processors, wherein the
instructions, when executed, cause the system to perform operations comprising: receiving
a reference trajectory representing an initial path for an autonomous vehicle to traverse an
environment; determining a first portion of the reference trajectory; determining a second
portion of the reference trajectory that is separate from the first portion; determining a first
weight associated with a first reference cost to be evaluated at a first point associated with
the first portion of the reference trajectory, wherein the first reference cost is associated
with a first difference between the first point on the reference trajectory and a
corresponding first point; determining a second weight associated with a second reference
cost to be evaluated at a second point associated with the second portion of the reference
trajectory, wherein the second reference cost is associated with a second difference
between the second point on the reference trajectory and a corresponding second point,
wherein the second weight is larger than the first weight; evaluating a reference cost
function at the first point and the second point to generate a target trajectory comprising
the corresponding first point and the corresponding second point; and controlling the
autonomous vehicle to traverse the environment based at least in part on the target
trajectory.
V: The system of paragraph U, wherein the first portion of the reference trajectory is
associated with at least one of a discontinuity in the reference trajectory or an obstacle cost
that meets or exceeds a threshold obstacle cost.
W. The system of paragraph U or V, wherein the first difference comprises one or more of a
difference in a yaw, a lateral offset, a velocity, an acceleration, a curvature, or a curvature
rate.
X: The system of any of paragraphs U-W, wherein the first weight is based at least in part
on one or more of: determining that a lateral displacement associated with a portion of the
target trajectory corresponding to the first portion of the reference trajectory is below a
first threshold distance for a given a time period; or determining that the lateral
displacement associated with the portion of the target trajectory corresponding to the first
portion of the reference trajectory is below a second threshold distance for a given distance.
Y: The system of any of paragraphs U-X, wherein the first reference cost comprises a
penalty associated with generating the target trajectory that deviates, at the first point and
represented as the first difference, from the reference trajectory.
Z: A method comprising: receiving a first trajectory for an autonomous vehicle to traverse

an environment; determining a first portion of the first trajectory; determining a second
portion of the first trajectory that is separate from the first portion; determining a first
weight associated with the first portion of the first trajectory; determining a second weight
associated with second portion of the first trajectory, wherein the first weight is different
than the second weight; evaluating, with respect to the first trajectory and based at least in
part on the first weight and the second weight, a reference cost to generate a second
trajectory; and controlling the autonomous vehicle to traverse the environment based at
least in part on the second trajectory.
AA: The method of paragraph Z, wherein evaluating the reference cost comprises:
evaluating a first reference cost associated with the first portion, the first reference cost
proportional to the first weight; and evaluating a second reference cost associated with the
second portion, the second reference cost proportional to the second weight.
AB: The method of paragraph Z or AA, wherein the first portion of the first trajectory is
associated with points, and wherein the point is a first point, the method further
comprising: determining a point density for the points associated with the first portion of
the first trajectory based at least in part on a first cost associated with a curvature value of
the first portion or a second cost associated with a distance between the first point of the
first portion and a second point associated with an obstacle in the environment.
AC: The method of any of paragraphs Z-AB, wherein the reference cost is associated with a
difference between a first point on the first trajectory and a corresponding first point on the
second trajectory.
AD: The method of paragraph AC, wherein the difference comprises one or more of a
difference in a yaw, a lateral offset, a velocity, an acceleration, a curvature, or a curvature
rate.
AE: The method of any of paragraphs Z-AD, wherein the first portion of the first trajectory
is associated with at least one of a discontinuity in the first trajectory or an obstacle cost
that meets or exceeds a threshold obstacle cost.
AF. The method of paragraph AE, wherein: the discontinuity in the first trajectory
represents a lane change action; or the obstacle cost meeting or exceeding the threshold
obstacle cost represents a least a portion of the first trajectory passing within an area
representing an obstacle in the environment.
AG: The method of any of paragraphs Z-AF, further comprising: determining a drivable
area through the environment for the autonomous vehicle, wherein the drivable area
represents a region in the environment on which the autonomous vehicle is configured to
traverse and an object in the environment; and evaluating an obstacle cost based at least in
part on the drivable area; wherein the second trajectory is based at least in part on the
obstacle cost.
AH: The method of paragraph AG, wherein determining the drivable area is based at least
in part on a velocity associated with the autonomous vehicle or a classification of the object.
AI: A non-transitory computer-readable medium storing instructions that, when executed,

cause one or more processors to perform operations comprising: receiving a first trajectory
for an autonomous vehicle to traverse an environment; determining a first portion of the
first trajectory; determining a second portion of the first trajectory that is separate from the
first portion; determining a first weight associated with the first portion of the first
trajectory; determining a second weight associated with second portion of the first
trajectory, wherein the first weight is different than the second weight; evaluating, with
respect to the first trajectory and based at least in part on the first weight and the second
weight, a reference cost to generate a second trajectory; and controlling the autonomous
vehicle to traverse the environment based at least in part on the second trajectory.
AJ: The non-transitory computer-readable medium of paragraph AI, wherein: evaluating

the reference cost comprises: evaluating a first reference cost associated with the first
portion, the first reference cost proportional to the first weight; and evaluating a second
reference cost associated with the second portion, the second reference cost proportional to
the second weight.
AK: The non-transitory computer-readable medium of paragraph AI or AJ, wherein the

reference cost is associated with a difference between a first point on the first trajectory and
a corresponding first point on the second trajectory.
AL: The non-transitory computer-readable medium of any of paragraphs AI-AK, wherein

the first portion of the first trajectory is associated with at least one of a discontinuity in the
first trajectory or an obstacle cost that meets or exceeds a threshold obstacle cost.
AM: The non-transitory computer-readable medium of any of paragraphs AI-AL, wherein

the operations further comprise: determining a drivable area through the environment for
the autonomous vehicle, wherein the drivable area represents a region in the environment
on which the autonomous vehicle is configured to traverse and an object in the
environment; and evaluating an obstacle cost based at least in part on the drivable area;
wherein the second trajectory is based at least in part on the obstacle cost.
AN: The non-transitory computer-readable medium of paragraph AM, wherein

determining the drivable area is based at least in part on a velocity associated with the
autonomous vehicle or a classification of the object.
While the example clauses described above are described with respect to one particular
implementation, it should be understood that, in the context of this document, the content
of the example clauses can also be implemented via a method, device, system, a computer-
readable medium, and/or another implementation.
CONCLUSION
While one or more examples of the techniques described herein have been described,
various alterations, additions, permutations and equivalents thereof are included within
the scope of the techniques described herein.
In the description of examples, reference is made to the accompanying drawings that form
a part hereof, which show by way of illustration specific examples of the claimed subject
matter. It is to be understood that other examples can be used and that changes or
alterations, such as structural changes, can be made. Such examples, changes or alterations
are not necessarily departures from the scope with respect to the intended claimed subject
matter. While the steps herein can be presented in a certain order, in some cases the
ordering can be changed so that certain inputs are provided at different times or in a
different order without changing the function of the systems and methods described. The
disclosed procedures could also be executed in different orders. Additionally, various
computations that are herein need not be performed in the order disclosed, and other
examples using alternative orderings of the computations could be readily implemented. In
addition to being reordered, the computations could also be decomposed into sub-
computations with the same results.
Claims
1. A system comprising:
one or more processors; and
one or more computer-readable media storing instructions executable by the one or

more processors, wherein the instructions, when executed, cause the system to
perform operations comprising: receiving a reference trajectory representing an
initial trajectory for an autonomous vehicle to traverse an environment; determining
at least one of: a first cost associated with a curvature value associated with a first
point of the reference trajectory; or a second cost associated with a distance between
the first point and a point associated with an obstacle in the environment;
determining, based at least in part on at least one of the first cost or the second cost, a
first point density of first points associated with a first portion of the reference
trajectory; determining a second point density of second points associated with a
second portion of the reference trajectory, wherein the second point density is less
than the first point density; evaluating a cost function at the first point of the first
points and at a second point of the second points to generate a target trajectory
comprising first corresponding points and second corresponding points, wherein a
first corresponding point of the first corresponding points corresponds to the first
point of the first points, and wherein a second corresponding point of the second
corresponding points corresponds to a second point of the second points; and
controlling the autonomous vehicle to traverse the environment based at least in part
on the target trajectory.
2. The system of claim 1, wherein the first point density is further based, at least in part, on
one or more of:
a time period between adjacent points of the first points; or
a distance between the adjacent points of the first points.
3. The system of claim 1, wherein the distance is a first distance, and wherein the cost
function comprises a cost based at least in part on a second distance between the first
corresponding point of the target trajectory and a point associated with the obstacle in the
environment.
4. The system of claim 1, wherein determining the first point density comprises accessing a
lookup table or a function to determine the first point density based at least in part on the
first cost or the second cost.
5. The system of claim 1, wherein a first time period associated with the first point and an
adjacent first point of the first points is less than a second time period associated with the
second point and an adjacent second point of the second points.
6. A method comprising:
receiving a first trajectory representing an initial trajectory for a vehicle to traverse an

environment;
determining a point density for points associated with a portion of the first trajectory
based at least in part on a first cost associated with a curvature value of the portion or
a second cost associated with a distance between a first point of the portion and a
second point associated with an obstacle in the environment;
determining, based at least on a cost associated with the first point, a second
trajectory; and
controlling the vehicle to traverse the environment based at least in part on the
second trajectory.
7. The method of claim 6, wherein at least one of:
the point density increases as the first cost associated with the curvature value
increases; or
the point density increases as the second cost associated with the distance between a
first point and the second point increases.
8. The method of claim 6, wherein:
the point density is a first point density;
the portion is a first portion;
the points are first points;
the curvature value is a first curvature value;
the distance is a first distance;
the obstacle is a first obstacle; and
the method further comprises: determining a second point density for second points
associated with a second portion the first trajectory based at least in part on a third
cost associated with a second curvature value of the second portion or a fourth cost
associated with a second distance between a third point of the second portion and a
fourth point associated with a second obstacle in the environment, wherein the first
point density is different than the second point density.
9. The method of claim 6, wherein the distance is a first distance, and wherein the point
density is based at least in part on one of:
a second distance between the first point on the portion and an adjacent first point on
the portion; or
a time period associated with a difference in a first time associated with the first point
and a second time associated with the adjacent first point.
10. The method of claim 6, wherein the cost is based at least in part on the distance
between the first point of the portion and the second point associated with the obstacle in
the environment.
11. The method of claim 6, wherein determining the second trajectory further comprises:
evaluating the cost for a plurality of points associated with the first trajectory to
generate the second trajectory.
12. The method of claim 6, further comprising:
receiving the first trajectory from a first layer of a planning system of the vehicle,
wherein the first trajectory is generated with respect to a distance between points on
the first trajectory and the second trajectory is generated, by a second layer of the
planning system, with respect to a time period between points on the second
trajectory.
13. The method of claim 12, wherein the point density is non-uniform with respect to the
portion of the first trajectory.
14. The method of claim 6, wherein determining the point density comprises accessing a
lookup table or a function to determine the point density based at least in part on the first
cost or the second cost.
15. A non-transitory computer-readable medium storing instructions that, when executed,

cause one or more processors to perform operations comprising:
receiving a first trajectory representing an initial trajectory for a vehicle to traverse an

environment;
determining a point density for points associated with a portion of the first trajectory
based at least in part on a first cost associated with a curvature value of the portion or
a second cost associated with a distance between a first point of the portion and a
second point associated with an obstacle in the environment;
generating, based at least in part on a cost associated with the first point, a second
trajectory, the second trajectory comprising a target point; and
controlling the vehicle to traverse the environment based at least in part on the
second trajectory.
16. The non-transitory computer-readable medium of claim 15, wherein at least one of:
the point density increases as the first cost associated with the curvature value
increases; or
the point density increases as the second cost associated with the distance between a
first point and the second point increases.
17. The non-transitory computer-readable medium of claim 15, wherein:
the point density is a first point density;
the portion is a first portion;
the points are first points;
the curvature value is a first curvature value;
the distance is a first distance;
the obstacle is a first obstacle; and
the operations further comprise: determining a second point density for second
points associated with a second portion the first trajectory based at least in part on a
third cost associated with a second curvature value of the second portion or a fourth
cost associated with a second distance between a third point of the second portion
and a fourth point associated with a second obstacle in the environment, wherein the
first point density is different than the second point density.
18. The non-transitory computer-readable medium of claim 15, wherein the distance is a
first distance, and wherein the cost is based at least in part on a second distance between
the target point of the second trajectory and the second point associated with the obstacle
in the environment.
19. The non-transitory computer-readable medium of claim 15, wherein a first time period
associated with the first point and an adjacent first point of the first points is less than a
second time period associated with the second point and an adjacent second point of the
second points.
20. The non-transitory computer-readable medium of claim 19, the operations further
comprising:
receiving the first trajectory from a first layer of a planning system of the vehicle,
wherein the first trajectory is generated with respect to a Euclidian distance between
points on the first trajectory and the second trajectory is generated, by a second layer
of the planning system, with respect to a time period between points on the second
trajectory.
Referenced Cited
U.S. Patent Documents
9120485 September 1, 2015 Dolgov
9523984 December 20, 2016 Herbach et al.
20130110368 May 2, 2013 Zagorski
20150120137 April 30, 2015 Zeng et al.
20150345959 December 3, 2015 Meuleau
20150353085 December 10, 2015 Lee
20170120804 May 4, 2017 Kentley et al.
20170168485 June 15, 2017 Berntorp et al.
20180150081 May 31, 2018 Gross et al.
20180164822 June 14, 2018 Chu et al.
20180261093 September 13, 2018 Xu et al.
20180330618 November 15, 2018 Bai et al.
20180338229 November 22, 2018 Nemec et al.
20190009787 January 10, 2019 Ishioka
20190078896 March 14, 2019 Zhu
20190120640 April 25, 2019 Ho et al.
20190196485 June 27, 2019 Li
20190235505 August 1, 2019 Li
20190235506 August 1, 2019 Bardapurkar et al.
20190347515 November 14, 2019 Kehl et al.
20190391580 December 26, 2019 Di Cairano
20200139959 May 7, 2020 Akella et al.
20200398833 December 24, 2020 Hudecek et al.
20200398894 December 24, 2020 Hudecek et al.
Foreign Patent Documents
109878513 June 2019 CN
102013214225 January 2015 DE
102016222782 May 2018 DE
3709280 September 2020 EP
Other references
The PCT Search Report and Written Opinion dated Apr. 14, 2020, for PCT
Application No. PCT/US2019/058760, 16 pages.
Non Final Office Action dated Oct. 14, 2020 for U.S. Appl. No. 16/566,821, “Dynamic
Lane Biasing”, Hudecek, 19 pages.
Bloomberg QuickTake, “Zoox/s $800M Robo Taxi Could Beat Tesla and Uber”,
retrieved on Nov. 5, 2020 at <<https://www.youtube.com/watch?
v=0jDLwnTyybo>>, Youtube, Jul. 17, 2018, 1pp.
PCT Search Report and Written Opinion dated Nov. 20, 2020 for PCT Application
No. PCT/US2020/050225, 33 pages.
Office Action dated Oct. 2, 2020 for U.S. Appl. No. 16/566,828 “Dynamic Collision
Checking” Hudecek, 18 pages.
Office Action for U.S. Appl. No. 16/179,711, dated Mar. 23, 2021, Akella, “Cost Scaling
in Trajectory Generation”, 33 pages.
Patent History
Patent number: 11048260
Type: Grant
Filed: Nov 2, 2018
Date of Patent: Jun 29, 2021
Patent Publication Number: 20200142417
Assignee: Zoox, Inc. (Foster City, CA)
Inventor: Janek Hudecek (Redwood City, CA)
Primary Examiner: Tamara L Weber
Application Number: 16/179,679
Classifications
Current U.S. Class: Non/e
International Classification: G05D 1/02 (20200101); G05D 1/00 (20060101); B60W
30/18 (20120101);

0b - US Patent For Adaptive Scaling in Trajectory Generation Patent (Patent # 11,048,260 Issued June 29, 2021) - Justia Patents Search

Uploaded by

Copyright:

Available Formats

0b - US Patent For Adaptive Scaling in Trajectory Generation Patent (Patent # 11,048,260 Issued June 29, 2021) - Justia Patents Search

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

0b - US Patent For Adaptive Scaling in Trajectory Generation Patent (Patent # 11,048,260 Issued June 29, 2021) - Justia Patents Search

Uploaded by

Copyright:

Available Formats

12/5/22, 11:33 PM US Patent for Adaptive scaling in trajectory generation Patent (Patent # 11,048,260 issued June 29, 2021)

29, 2021) - Justia Patents Search

Adaptive scaling in trajectory

Techniques for generating trajectories and drivable

reference trajectory, and controlling the autonomous

Latest Zoox, Inc. Patents:

Component tray assembly for vehicles

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram illustrating an example implementation of a system for

embodiments of the disclosure.

FIG. 2 is an illustration of a reference trajectory comprising regions of differing point

FIG. 3 is an illustration of determining weight(s) for a cost associated with generating a

FIG. 4 is an illustration of determining weight(s) for a cost associated with generating a

FIG. 5A is an illustration of a dilated region, a collision region, and a safety region

FIG. 6 is an illustration of updating a region based on identifying an object such as a

In some examples, a reference trajectory representing an initial path or trajectory for an

The techniques discussed herein can improve a functioning of a computing device in a

reference trajectory can reduce an amount of processing by increasing a density of points in

The techniques described herein can be implemented in a number of ways. Example

FIG. 1 is a schematic diagram illustrating an example implementation of a system for

FIG. 1 illustrates a single example of using a drivable area to navigate in an environment.

FIG. 2 is an illustration of a reference trajectory comprising regions of differing point

FIG. 3 is an illustration of determining weight(s) associated with a cost associated with

FIG. 4 is an illustration of determining weight(s) associated with a cost associated with

FIG. 5A is an illustration of a dilated region, a collision region, and a safety region

FIG. 6 is an illustration of updating a region based on identifying an object such as a

As described herein, an exemplary neural network is a biologically inspired algorithm

Additional examples of architectures include neural networks such as ResNet70,

It should be noted that while FIG. 7 is illustrated as a distributed system, in alternative

FIGS. 8 and 9 illustrate example processes in accordance with embodiments of the

static and/or dynamic obstacles in an environment, a minimum or maximum longitudinal

determining a portion of the reference trajectory based at least in part on a predicted

G: A method comprising: receiving a first trajectory representing an initial trajectory for a

O: A non-transitory computer-readable medium storing instructions that, when executed,

representing an initial trajectory for a vehicle to traverse an environment; determining a

P: The non-transitory computer-readable medium of paragraph O, wherein at least one of:

Q: The non-transitory computer-readable medium of paragraph O or P, wherein: the point

R: The non-transitory computer-readable medium of any of paragraphs O-Q, wherein the

S: The non-transitory computer-readable medium of any of paragraphs O-R, wherein a

T: The non-transitory computer-readable medium of paragraph S, the operations further

U: A system comprising: one or more processors; and one or more computer-readable

represented as the first difference, from the reference trajectory.

Z: A method comprising: receiving a first trajectory for an autonomous vehicle to traverse

AI: A non-transitory computer-readable medium storing instructions that, when executed,

AJ: The non-transitory computer-readable medium of paragraph AI, wherein: evaluating

AK: The non-transitory computer-readable medium of paragraph AI or AJ, wherein the

AL: The non-transitory computer-readable medium of any of paragraphs AI-AK, wherein

AM: The non-transitory computer-readable medium of any of paragraphs AI-AL, wherein

AN: The non-transitory computer-readable medium of paragraph AM, wherein

one or more processors; and

one or more computer-readable media storing instructions executable by the one or

a time period between adjacent points of the first points; or

a distance between the adjacent points of the first points.

receiving a first trajectory representing an initial trajectory for a vehicle to traverse an

7. The method of claim 6, wherein at least one of:

8. The method of claim 6, wherein:

the point density is a first point density;

the portion is a first portion;

the points are first points;