Kinematic Modeling of Quadruped Robot

Smita A. Ganjare, V S Narwane & Ujwal Deole

1&2

K. J Somaiya College of Engineering, Mumbai University,

Product Development, Larsen & Toubro LTD, Powai
E-mail : [email protected], [email protected], [email protected]

Abstract This Paper presents design of a quadruped

robot which tries to walk with medium speed on flat
terrain. The design of the quadruped is inspired by the
four legged animal, two joints of the leg enable to perform
two basic motions: lifting and stepping. The kinematic
modeling proposes necessary conditions for stable walk on
flat terrain. Brief guidelines for the design of the leg
mechanism are presented through the study of various
joints, links and degrees of freedom. Also a specific
semicircular trajectory is proposed which provide the
movement pattern of the legs.

Laboratory robots did a good job of demonstrated the

feasibility of dynamically balanced legged systems; they
had two primary limitations that would need to be
addressed to build practical legged vehicles. One is the
need for on-board power so the robot could operate in
the field without hoses and wires. Another is the need
for control algorithms that provide locomotion and
stability on rough terrain [5].
Here main concentrate is on the design of
quadruped involving two degrees of freedom where the
robot is able for lifting and stepping only.

Key Words Legged robots, quadruped robot, forward and

inverse kinematics, trajectory planning.

I.

The assumptions for deriving the kinematic model

are as follows:

INTRODUCTION

a)

Surface with alternating pair of legs.

b) The body is held at a constant height and parallel to

the ground plane during locomotion.

This paper presents an introduction to mobile

robotics. The development of mobile robotic platforms
is an important and active area of research. Within this
domain, the major focus has been to develop on wheeled
or tracked systems that cope very effectively with flat
and well-structured solid surfaces (e.g. laboratories and
roads) [1]. In recent years, there has been considerable
success with robotic vehicles even for off-road
conditions. However, wheeled robots still have major
limitations and difficulties in navigating uneven and
rough terrain. These limitations and the capabilities of
legged animals encouraged researchers for the past
decades to focus on the construction of biologically
inspired legged machines. These robots have the
potential to outperform the more traditional designs with
wheels and tracks in terms of mobility and versatility
[2].

c)

The center of gravity of the body is assumed to be

at the geometric center of the body.

This paper initially introduces the formation of coordinate system linkage of quadruped robot and then the
forward and inverse kinematic equations of the
quadruped robot are formulated. Finally, 3D model of
the robot is prepared using UG-NX.
II. FORWARD KINEMATIC MODEL
The mechanical structure of the robot is mainly
divided into two parts: body and legs. The body is the
rigid box; the four legs are distributed symmetrically
along the body and each leg has the same structure. The
links are made up of a series of rigid linkages connected
by rotating joints. There are two joints for each leg
which are hip joint and knee joint [4].

Over the past 40 years, a variety of engineers and

scientists have embraced the opportunity of legged
locomotion, building a diverse set of ingenious and
inspiring legged robots. For example, see Berns,
(2006), Kar (2003) and for many examples. The Leg

The kinematic modeling of the robot is carried out

to realize robot automatic control. Generally, there are
two main problems in robot kinematics research. The
ISSN : 2319 3182, Volume-2, Issue-4, 2013

21

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

first one is to assign joint angles to robot and then

calculate the position and orientation, which is called
forward kinematic modeling. The other is to calculate all
the corresponding joint angles of robot given position
and orientation of the robot, is called inverse kinematic
modeling. Forward kinematics is done initially by
assigning the co-ordinate frame to each link. D-H
algorithm is used to assign frames to the links. The
single line diagram is as follows,

As the four legs of the robot have same symmetry,

we can calculate the forward and inverse kinematic
model for one leg and then apply it to remaining 3 legs.
Each leg has 2 degrees of freedom, hence total there are
total 8 degrees of freedom for four legs.
III. INVERSE KINEMATIC MODEL
The analysis of forward kinematic problem can be
used to determine whether the motion of the robot meets
with the requirements or not, it can also be used as
preparation for the analysis of inverse kinematic
problem. The inverse kinematic model verifies whether
the robot can achieve the required position and
orientation or not. The inverse kinematic model finds
out the rotation angle with which the robot can move
[10]. The solution to the inverse kinematic problem is
found out with the help of Tool Configuration Vector
(TCV) [10].TCV (Tool configuration vector) is the
compact representation of the foot tip position and foot
orientation. The direction in which the foot is oriented
or pointed is given by the approach vector or the last
column of rotation matrix specifies the roll, but does not
give any information about yaw and pitch.

Fig 1: Single line diagram of each link

The tool configuration vector is given by,

Fig 2: Co-ordinate system of each linkage

The final homogeneous transformation matrix of
co-ordinate system is given by,

ISSN : 2319 3182, Volume-2, Issue-4, 2013

22

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

foot. A smooth function is one that is continuous and

has a continuous first derivative [11]. The smooth
motion has the important advantage of reducing the
vibrations and wear of the mechanical system. Here we
consider the leg encounter a semicircular path. Some
assumptions are made for this kind of trajectory which
is:
a)

Some data has been assumed from the Bigdog and

HyQ robots, which are rough terrain mechanical
mule. The body length of the robot should be 1m
long and is at near about 0.8 m from the ground.

b) The legs of the robot are placed at a distance of 0.6

m from each other and all the legs have the same
symmetry.
c)

The alternate pair of legs moves at a time i.e leg 1

and leg 3 move simultaneously. This is called as
trotting. The distance between two legs i.e front and
rear one is 0.6m and the distance covers in one
stance is 0.3m.The speed of the robot is 2 m/sec.

Firstly the 3D model of the robot has been

developed in UG-NX, some of the design parameters are
L1=L2=0.5m.

IV. TRAJECTORY PLANNING

The goal of trajectory planning is to describe the
requisite motion of the manipulator as a time sequence
of joints or links locations and derivatives of locations,
which are generated
by interpolating or
approximating the desired path by the polynomial
function. These time base sequence locations, also
called time law or time history, obtained from the
trajectory planning serve as reference input or control
set points to the manipulators control system. The
control system, in turn, assures that the manipulator
executes the planned trajectories.

Fig 3: 3D CAD model of quadruped robot

Secondly, with the help of semicircular trajectory,
parametric equation for this kind of quadruped robot is
deduced.
x=r-r*cos (*t)
y=r*sin (*t)

Every robot leg has at least capability to move from

an initial location (position and orientation) to a final
desired location. The kinematics and dynamics of the
manipulator govern the transition of posture with
actuators exerting the desired torques. As no limits must
be violated and no unmodelled behavior should be
executed, it is necessary to devise planning algorithms
that generate smooth trajectories.

From these equations, the x and y co-ordinates of

the robot feet is derived. By using these co-ordinates
and with the help of inverse kinematic model, the hip
and knee angles of each leg can be found out. Angles for
alternate pair of legs are calculated, considering leg1leg3 simultaneously moves first and leg2-leg4 will
follow [4].

One of the main objectives of any trajectory

planning algorithm is to achieve a smooth motion of the

The trajectory of the robots leg is set with the help

of parametric equations and the values of angles are
found out from the inverse kinematics. One step should
ISSN : 2319 3182, Volume-2, Issue-4, 2013

23

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

take 6.67 sec with uniform speed. The graphics

developed using Matlab2011a, is shown in fig 3 to fig 7.

There is the verification of the forward kinematic model

and inverse kinematic model with the help of robotic
tool box. The robotic toolbox provides many functions
that are useful in robotics such as kinematics, dynamics
and trajectory planning. The toolbox is useful for
simulation as well as in analyzing results from
experiments with real robots. Robotics mathematics can
be programmed in MATLAB2011a and simultaneously
results can be checked with MATLAB-Robotics
toolbox.
The result that is obtained by direct and inverse
kinematics is validated using Robotic Toolbox. There
are two links L1 and L2 of length 0.5 m each.

Fig 4: Movement of Leg 1 and Leg 3 in x-direction

L1=link (0 0.5 60 0)
L2=link (0 0.5 60 0)
R=robot ({L1 L2})
T=fkine(r,q)
q=ikine(r,T)
By using robotic toolbox, the transformation matrix
obtained is given as follows :

Fig 5: Movement of Leg 2 and Leg 4 in x-direction We can

also check out how the robot moves along y-axis.

This result is verified with the results which are obtained

from direct and inverse kinematics.
q= [1.0471 1.2391];
Above are the angles of hip and knee in radians for a
specified position.
V. APPROACH FOR DIFFERENTIAL MOTION
AND STATICS
The direct and inverse kinematic model establishes
the relationship between the manipulators joint
displacement and position and orientation of its end
effector. These relationships permit the static control of
the manipulator to place the end effector at specified
location and make it traverse a specified path in space.
However, for the manipulator not only the final location
of the end-effector is of concern, but also the velocities
at which the end-effector would move to reach the final
location are an equally important concern.

Fig 6: Movement of Leg 1 and Leg 3 in y-direction

This requires the co-ordination of the instantaneous endeffector velocity and joint velocities. One way to
achieve this is to take the time derivative of kinematic
equations of the manipulator. The transformation from
the joint velocities to the end-effector velocity is
described by a matrix, called the Jacobian. The Jacobian

Fig 7: Movement of Leg 2 & leg 4 in y-direction.

ISSN : 2319 3182, Volume-2, Issue-4, 2013

24

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

matrix which is dependent on manipulator configuration

is a linear mapping from velocities in joint space to
velocities in Cartesian space. The Jacobian is one of the
most important tools for characterization of differential
motions of the manipulator. At certain locations in joint
space, the Jacobian matrix may lose rank and it may not
be possible to find its inverse. These locations are
referred to as singular Configurations [10].

Where jvi and jwi represents the component k of linear

velocity and angular velocity respectively, contributed
by joint i with k=x, y, z or z and i=1, 2 n. The
contribution of joint i to column is computed depending
on whether joint I is prismatic or rotary.

It is observed that the Cartesian velocities (linear as

well as angular) of the end effector are linearly related
to the joint velocities. The relationship between
infinitesimal (differential) joint motions with
infinitesimal (differential)

Each column of Jacobian matrix is computed separately

and all the columns are combined to form the total
Jacobian matrix. Jacobian matrix for rotary joint is given
by following equation,

changes in the end-effectors position (and orientation)

has been investigated. As these changes take place in
infinitesimal (differential) time, there is a need to map
instantaneous end-effector velocities, Ve (in Cartesian
space) to instantaneous joint velocities (in joint
space).This mapping between differential changes is
linear and can be expressed as
Ve (t) =J (q)----------------------------(1)

The Jacobian matrix column J1 for joint 1, which is a

rotary joint, is determined as follows,

Where,

The joint axis vector P0 (Pi-1 for i=1) is

Ve (t) =6*1 Cartesian velocity vector (End-effector

velocity),
J (q) =6*n Manipulator Jacobian or Jacobian matrix,
q=n*1 vector of n joint velocities.
Ve (t) = [J1 (q) J2 (q)Jn (q)]------------------ (2)
In the above equation, Ji(q) is the ith column of the
Jacobian matrix.
We can write the end-effector velocity as,

Equation (3) represents the forward differential

motion model or differential kinematics model
presented schematically in following figure which is
similar to the forward kinematic model. Note that the J
(q) is the function of the joint variables [11].The first
three rows of Jacobian J(q) are associated with the linear
velocity of the end effector v, while last three rows
correspond to the angular velocity w of the end effector.
Each joint of the manipulator generate some linear
and/or some angular velocity of the manipulator.
Column (i) of the Jacobian matrix (q) is made up of
three linear velocity components jvi and three angular
velocity components and jwican be expressed as,

The first column of Jacobian is calculated by

substituting equation (7) and (9) in equation (5). Thus,

ISSN : 2319 3182, Volume-2, Issue-4, 2013

25

International Journal on Theoretical and Applied Research in Mechanical Engineering (IJTARME)

VII. REFERENCES

Equation (10) provides the linear and angular velocities

of the end effector i.e foot of the robot.

From the above equation, we get the linear velocity

of the foot position. By using Jacobian, we can find out
the linear and angular velocity of the final foot position.
With the help of Jacobian, we get the three linear
velocity and three angular velocity components of the
foot.

[1]

J.A.Tenreiro Machado and Manuel F.Silva , An

Overview Of Legged Robots, Department of
Electrical Engineering, Institute of Engineering
of Porto, Porto, Portugal.

[2]

Sven Bttcher, Principles of robot locomotion,

Seminar Human robot interaction.

[3]

Freyr Hardarson, Locomotion for difficult terrain,

Mechatronics Division, Department of Machine
Design Royal Institute of Technology,
Stockholm, Sweden.A Survey Study 1997-04-01.

[4]

Thanhtam Ho, Sunghac Choi, Sangyoon Lee,

Development of a Biomimetic Quadruped Robot,
Department of Mechanical Design and
Production Engineering, Konkuk University,
Seoul 143-701, Korea.

[5]

MIT Leg Lab Robots. 13 Oct 2008

<http://www.ai.mit.edu/projects/leglab/robots/rob
otsmain.html>.

[6]

Boston Dynamics: The Leader in Lifelike

Human
Simulation.
13
Oct
2008
<http://bostondynamics.com/content/sec.php?sect
ion=BigDog>

[7]

Shibendu Shekhar Roy, Ajay Kumar Singh1, and

Dilip Kumar Pratihar, Department of Mechanical
Engineering, National Institute of Technology,
Durgapur, India Analysis of Six-legged Walking
Robots, Department of Mechanical Engineering,
Indian Institute of Technology, Kharagpur, India.

[8]

Raghu Mandayam Anand, Kinematic Analysis of

a Quadruped Robot capable of functioning as a
machine tool, University of Texas at Arlington.

[9]

Filipe Miguel Teixeira Pereira da Silva,

Department
of
Mechanical
Engineering,
University of Aveiro and Jos Antnio Tenreiro
Machado, Department of Electrical and
Computer Engineering, Kinematic Analysis and
Modeling of Biped Locomotion Systems, J. Braz.
Soc. Mech. Sci. vol.21 no.3 Rio de Janeiro Sept.
1999

[10]

Modeling and Control of Robot Manipulators, By

Lorenzo Sciavicco & Bruno Siciliano.

VI. CONCLUSION
The kinematic analysis of quadruped robot is the
basic of space planning, motion control and optimal
design. Here the direct and inverse kinematic models for
the robot are calculated. Direct kinematics gives exact
position and orientation of robot feet and the inverse
kinematics gives the angles of the joints of the legs. The
results are verified with robotic toolbox of Matlab2011a.
It provides the real time analysis of the joint variables,
which lays foundation for following up motion control.

[11] Robotics and Control, By R.K.Mittal and

I.J.Nagrath.

ISSN : 2319 3182, Volume-2, Issue-4, 2013

26