Canadian Journal on Electrical and Electronics Engineering Vol. 3, No.

6, July 2012

Modeling and Analysis of a 6 DOF Robotic Arm Manipulator

Abstract The behavior of physical systems in many situations may better be expressed with an analytical model. Robot modeling and analysis essentially involve its kinematics. For robotic manipulators having high Degrees Of Freedom (DOF) with multiple degrees in one or more joints, an analytical solution to the inverse kinematics is probably the most important topic in robot modeling. This paper develops the kinematic models a 6 DOF robotic arm and analyzes its workspace. The proposed model makes it possible to control the manipulator to achieve any reachable position and orientation in an unstructured environment. The forward kinematic model is predicated on Denavit Hartenberg (DH) parametric scheme of robot arm position placement. Given the desired position and orientation of the robot end-effector, the realized inverse kinematics model provides the required corresponding joint angles. The forward kinematic model has been validated using Robotics Toolbox for MATLAB while the inverse kinematic model has been implemented on a real robotic arm. Experimental results demonstrate that using the developed model, the end-effector of robotic arm can point to the desired coordinates within precision of 0.5cm. The approach presented in this work can also be applicable to solve the kinematics problem of other similar kinds of robot manipulators.1 Key Words Modeling of robot; Robot kinematics; Robotic system simulation; Analysis of Serial mechanism.

I. INTRODUCTION Analytical prediction of the behavior of physical systems in many key situations is either extremely complicated or even impossible. Driven with the constraints to prototype a physical system, modeling finds enormous motivations to study and investigate the performance of a system. Modeling a robot involves study of its kinematic behavior. A kinematic model is concerned with the robots motion without considering forces producing the motions. The kinematics of a robotic arm deals with the study of the
1 J. Iqbal is with Robotics and Control Research (RCR) Group, COMSATS Institute of Information Technology (CIIT), Islamabad, Pakistan (e-mail: [email protected]). R. ul Islam is with Robotics and Control Research (RCR) Group, COMSATS Institute of Information Technology (CIIT), Islamabad, Pakistan (e-mail: [email protected]). H. Khan is with Department of Advanced Robotics, Istituto Italiano di Tecnologia (IIT), Via Morego 30, 16163 Genova, Italy (e-mail: [email protected]).

geometric and time based properties of the motion and in particular how various links of a robot move with respect to one another and with time. It provides an analytical description of the spatial movements of a robot i.e. a relationship between position and orientation of robot endeffector and its joint variables. The problem of kinematic modeling is usually categorized into two sub-problems. First is the forward or direct kinematics, which is the problem of solving the Cartesian position and orientation of a mechanism, given the knowledge of the kinematic structure and the joint coordinates. The second sub-problem is Inverse Kinematics (IK), which computes the joint variables using the given information of a robots end-effector position and orientation. In case of serial robotic arms, IK problem is more complex than direct kinematic problem [1]. Kinematic modeling of robots also benefits the industrial automation processes by making them semi-autonomous or even fully autonomous. Because of the task nature and operational environment, the industrial robots are usually composed up of series of rigid links mounted on a base. They operate in a manner similar to that of the human arm. A 6 Degree Of Freedom (DOF) robotic arm manipulator is widely used in the industry. The most common applications of industrial robots include Spot welding, Spraying, Assembling and Manufacturing. Many of these applications actually require accomplishment of pick and place task. Implementation of this task essentially necessitates having the kinematic model of the robotic arm being employed. In the area of robot modeling and simulation, kinematics is a rigorously researched topic. Scientific community reports various robot modeling and analysis techniques. These are usually based either on line transformation or point transformation, the later being used more commonly for robot modeling. Clothier et al. [2] proposed a geometric model to solve the unknown joint angles required for autonomous positioning of a robotic system. A new method, quaternion algebra, for solution of forward kinematic problem was derived by Sahu et al. [3]. Popovic et al. [4] developed a strategy to analyze the upper extremity movement of the arm while complete body kinematics of a radial symmetrical sixlegged robot was presented by Wang et al. [5]. A mathematical approach to analyze kinematics of a humanoid robot was reported in [6]. Cubero [7] proposed an IK model to solve all the joint variables of a serial arm manipulator of any type. This model was based on forward kinematic solution. A 300

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012 virtual model robot having forward and inverse kinematic solutions was reported by Kuma in [8]. This paper first presents kinematic model of the robot in Section II. The robot has been modeled for its forward kinematics as well as IK. Section III discusses validation of forward kinematic model using MATLAB toolbox for robotics while Section IV presents the results of workspace analysis of the robotic arm. Implementation of IK model on the real robot is discussed in Section V. Finally Section VI comments on conclusion. The overall system (Figure 2) consists of the robot, its controller interfaced with a standard PC and a teaching pendant. The function of the controller (ED-Mark (ED IV) is to provide ports for encoded motors to optically drive the robot. The controller has more than 100 higher level kernel commands that make the platform versatile. The controller can be operated in either of the two modes: host controlled or through teaching pendant. Pendant is the instructional pad for manual commanding the robotic arm. This is used to let the robot learn about any reachable coordinates. The taught points can be saved in the controller for later retrieval ret and command execution. However, the robot has to be taught every time provided the location of the object to be manipulated is changed. In an attempt to make this robotic arm autonomous, an image-guided guided robotic system has been conceived and presented in [9].

II. KINEMATIC MODEL The robotic tic platform used in the present work is a 6 DOF robotic arm manipulator ED7220C developed by ED Corporation, Korea. The robotic arm has been extensively used in research, development and teaching. It is basically a serial manipulator having all joints as revolute. The arm geometrical configuration is made up of waist, shoulder, elbow and wrist in correspondence with the human arm joints (Figure 1). Each of these joints except the wrist has a single DOF. Wrist can move in two planes (roll and pitch), thus making the end-effector effector more flexible in terms of object manipulation. Constructed in a vertical articulated fashion, the robot offers visual observation of the mechanical behavior of each joint at a glance. The arm is fully-actuated actuated with each DOF achieved by a precise servo motor (DME 38B50G 38B50G-115) equipped with an optical encoder. The end-effector effector is a two twostate gripper having rubber pads. The built in mechanical safety limits restrict the joint motion in case something in the control algorithm goes wrong. TABLE I lists salient features of the robotic arm ED7220C.

Fig. 2. Arm shown with other system components.

Fig. 1. Joints configurations in ED7220C. TABLE I. SALIENT FEATURES OF ED7220C Feature Description Position precision 0.5mm (approx.) Movement speed 100mm/s (max.) Load capacity 1 Kg Weight 33 Kg Wrist - Pitch: 260, Roll:360 Range Of Motion Elbow: 172 (ROM) Shoulder: 90 Waist: 310

A. Forward Kinematic Model The study of kinematic problem of a robot can be carried out by different methods. Two commonly used methods are based on Denavit-Hartenberg Hartenberg (DH) parameters pa and successive screw displacements. Both methods are systematic in nature and more suitable for modeling serial manipulators. Also geometric methods are frequently used by some researchers for the serial manipulators of relatively simple geometry [10]. DH method has been used to develop the kinematic model of the robot in this work because of its versatility and acceptability for modeling of any number of joints and links of a serial manipulator regardless of complexity. Figure 3 illustrates the simplified implified kinematic model of the robotic arm in an inverted L pose. The first three joints are used to move the tool point to its desired position, while the last two joints adjust the orientation of the end-effector. end Link lengths mentioned in Figure 3 are re tabulated in TABLE II.

301

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012 each link (expressing joint i in its previous neighboring joint i-1) derived in [11], the corresponding transformation matrices for each link of the robotic arm have been written. Based on the compound transformation property, these individual transformation matrices when multiplied yield the overall matrix representing the end-effector end of the robot in terms of its base (1).
0

Where

Fig. 3. ED7220C Kinematic model.

TABLE II. ED 7220C LINK LENGTHS

In (1), the 3X3 matrix comprising of first three rows and first three columns is the rotation while the last column represents the position (x, y, z) of the end-effector end w.r.t. base.
Wrist L4 155

B = C2 C23 S234

A = S2+ S23+ C234

( 1) 1

Joint Symbol Link Length [mm]

Waist L1 385

Shoulder L2 220

Elbow L3 220

DH works with quadruple {i-1, ai-1, di, i} which represents twist angle, link length, link offset and joint angle respectively. Following DH convention, an orthonormal coordinate system has been attached to each link of the manipulator (Figure 4). TABLE III lists DH parameters for the robotic arm ED7220C.

B. Inverse Kinematic Model IK model finds more potential applications in practical robotic systems. IK model computes the joint angles required to achieve the given position and orientation. Not only in robotics, IK finds its importance in other fields like for example 3D games. In contrast to the forward kinematics, IK does not have a unique solution. The solutions which ensure collision-free free operation and minimum joint motion are considered more optimum. Analytical approach has been followed to develop IK model of ED7220C. This approach ensures that for any object within robotic arm workspace (Section IV), the model determines correct joint angles. The first four joint angles i.e. waist (1), shoulder (2), elbow (3) and tool pitch (4) are calculated using this approach while tool roll (5) is directly given by the desired orientation for object manipulation. Since transformation involves rotation as well as translation, so the general form of the transformation matrix from tool to base is given by (2).

Fig. 4. ED7220C Coordinate assignment. TABLE III. ED 7220C DH PARAMETERS Symbol Joints (i) 1 2 0 -90 0 0 0 L1 1 2 - 90 3 0 4 0 5 -90 0 0 5 6 0 0

L2
0 3

L3
0 4

L4
0

Where the first 3x3 matrix and (px, py and pz) representing the rotation and the translation of end-effector end w.r.t base of the robot in an IK problem are known. The developed analytical IK model after intensive mathematical computations yields equations (3), (9), (12) and (13) for the joint angles 1, 3, 2 and 4 respectively. These equations express the required joint angles angl in terms of given coefficients of (2). 2 , (3)

(2)

Using the general form of the transformation matrix for 302

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012 (4) (5) (6)
Fig. 5. Plot for joint angle configuration [0 0 0 0].

(7) 2

2 , 1

Considering another joint configuration [t1 t2 t3 t4] as [90 90 -90 -90], the position and orientation of the end-effector expressed in the base coordinates, as computed from model as well as MATLAB Toolbox has been found to be 0 1 0 0 1 0 0 0 0 0 1 0 0 220 760 1

2 ,

(8)

(9) (10)

Figure 6 shows MATLAB plot for this joint configuration.

800 600

y x

Z [mm]

(11) (12) (13)

400 200 0 -200 -400 400 200 0 Y [mm] -200 -400 -500 500 0 X [mm]
ED7220C

III. VALIDATION OF FORWARD KINEMATIC MODEL

2 ,

Fig. 6. Plot for joint angle configuration [90 90 -90 -90].

Z [mm]

The forward kinematic model has been validated using Robotics Toolbox for MATLAB. Numerical results together with visual plot of position and orientation of a robot in MATLAB environment gives clear insight of the kinematic behaviour of a robot. Given various angle set as input to the developed forward kinematics model (1) and MATLAB toolbox, corresponding results have been compared and plotted. Considering joint angle configuration [t1 t2 t3 t4] as [0 0 0 0]; the position and orientation of the end-effector expressed in the base coordinates, as computed from (1) is 0 0 1 0 0 1 0 0 1 0 0 0 155 0 825 1

Figure 7 shows the plot for joint angle configuration [t1 t2 t3 t4] as [0 90 90 0] with the forward transformation as

0 0 1 0

0 1 0 0

1 0 0 0

65 0 165 1

400 300 200 100 0 -100 -200 500 0 Y [mm] -500 -300 -200 -100 0 100 X [mm] 200 300 400
z x y

ED7220C

Fig.7. Plot for joint angle configuration [0 90 90 0].

Using the command fkine in MATLAB Toolbox for Robotics, the same result has been obtained. The corresponding MATLAB plot for this joint configuration is illustrated in Figure 5.
800 600 400 200 Z [mm] 0 -200 -400 -600 -800
ED7220C x y z

IV. WORKSPACE ANALYSIS With technological advancements in robotics and rapid increase in the trend to employ robots in various applications, the workspace of a robot has also become a primary performance parameter in addition to its speed, accuracy and weight. The workspace of a robot, also termed as its work envelope, actually expresses a robots ability to reach specific area. Given the information about Range Of Motion (ROM) of joints of a robot and length of its links, workspace can be determined. The ROM of ED7220C joints are mentioned in TABLE I while the link lengths are given in TABLE II. With these link lengths and ROM information of each joint 303

500 Y [mm] 0 -500 -300 -200 -100 0 X [mm] 100 200 300

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012 of the robotic arm, the robot workspace has been found mathematically using equation (1). Figures 8 and 9 illustrate the robot workspace ace in XY and XZ coordinates respectively. As shown in Figure 8, the arm has a manipulation ability inside a circular radius of 580 mm. ROM constraints in the body joint restricts the robot functionality in the region shown as V in the Figure. outside the work envelope, the algorithm terminates after prompting user. Otherwise IK model computes the required joint angles pointing the end-effector effector as per given position pos and orientation. These joint angles are then mapped to lowlow level encoder ticks. Finally the Kernel based instructions in the program execute the command by moving the motors as per mapped encoder ticks. The flow chart of model implementation is presented nted in Figure 11.

580 mm

Fig. 8. Workspace in XY.

Fig. 11. IK model implementation.

.
Fig. 9. Workspace in XZ.

The corresponding 3D MATLAB plot is illustrated in Figure 10.

Fig. 10. 3D Workspace.

V. IMPLEMENTATION OF IK MODEL ON REAL ARM Write IK model implementation here. The IK model has been implemented on the real robotic arm manipulator. An object has been placed at a known position and orientation. With this known information from a user, the developed algorithm first checks if the object bject lies inside the robot workspace as demonstrated in Section IV. If the object is 304

To implement the developed IK model on the robotic arm, it has been ensured that the object (a car key with a key-chain) key lies within the robotic arm workspace. For this purpose, the platform on which the object has been placed is raised (in height z) by placing two blocks as shown in Figure 12. The task chosen is basically picking an object from one location and placing it onto another. Both the source and destination position and d orientation have been given as an input. Figure 12a illustrates the arm at its home position (with all encoder values as zero). Based on the object coordinates given by the user, the robot moves as per computed joint angles (by IK model). Figure 12(b-d) d) shows the motion of the robot toward the object. After reaching the target location, the gripper of the robot closes ultimately grasping the object. Sequence of pick-up up of the object is shown in Figure 13(a-b). 13(a The robot then moves (Figure 14a-b) 14a toward destination point, whose coordinates have also been taught by the user. The destination point should also lie inside the operational workspace of the robot. After reaching that location, the robot drops the object (Figure 15a-d) d) and then finds its way back ba to the home position (Figure 16a-b).

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012

(a)

(b)

(a)

(b)

(c) (d) Fig. 12. Moving from Home position to the object.

(c) Fig. 15. Placing the object.

(d)

(a) Fig. 13. Picking the object.

(b)

(a) Fig. 16. Moving back to Home position.

(b)

VI. CONCLUSION A widely used 6 DOF robotic arm manipulator, ED7220C has been kinematically modeled followed by the analysis of its workspace. Forward kinematic model has been validated using MATLAB. The results from the derived forward kinematic model match exactly with that of MATLAB. The IK model of the robot has also provided correct joint angles to move the arm gripper to any position and orientation within its workspace. The IK model has been implemented on the real robotic platform. Results obtained from the model have been compared with the actual performance of the robot in accomplishing a task e.g. pick and place. It has been found that with the joint angles computed by IK model, the robot achieves position precision within 0.5cm. This little 305

(a) Fig. 14. Moving to destination position.

(b)

Canadian Journal on Electrical and Electronics Engineering Vol. 3, No. 6, July 2012 deviation is because of many reasons namely, reported platform precision (0.5mm as mentioned in TABLE I), mechanical coupling of the joints, non-linearity in mapping angles to low-level encoder ticks. The strategy presented in this paper may also be used to model and analyze other 6 DOF robotic arms.
Computational Intelligence, Control and Computer Vision in Robotics & Automation, pp. 76-82, 2008. [4] Popovic N., Williams S, Schmitz-Rode T., Rau G., Disselhorst-Klug C., Robot-based methodology for a kinematic and kinetic analysis of unconstrained, but reproducible upper extremity movement, Journal of Biomechanics, Vol. 42 No. 10, 2009. [5] Wang Z., Xilun D., Alberto R. and Alessandro G., Mobility analysis of the typical gait of a radial symmetrical six-legged robot, Mechatronics, Vol. 21, No. 7, pp. 1133-1146, 2011. [6] Man C., Fan X., Li C. and Zhao Z., Kinematics analysis based on screw theory of a humanoid robot, Journal of China University of Mining & Technology, Vol. 17 No. 1, pp. 4952, 2007. [7] Cubero S., An inverse kinematics method for controlling all types of serial-link robot arms, Mechatronics and Machine Vision in Practice, Springer-Verlag, Berlin, pp. 217-232. [8] Kuma R., Kalra P and Prakash N., A virtual RV-M1 robot system, Robotics and Computer-Integrated Manufacturing, Vol. 26 No. 6, pp. 994-1000, 2011. [9] Raza ul Islam, J. Iqbal, S. Manzoor, A. Khalid and S. Khan, An autonomous image-guided robotic system simulating industrial applications, 7th IEEE International Conference on System of Systems Engineering (SoSE), Genova, Italy, pp. 314-319, 2012. [10] Tsai L.W., Robot Analysis: The mechanics of serial and parallel manipulators, John Wiley & Sons, Feb. 1999. [11] J. J. Craig, Introduction to robotics: Mechanics and control, Prentice Hall, 2nd edition, pp.84.

ACKNOWLEDGMENT The authors will like to appreciate and personally thank Sarah Manzoor, Aayman Khalid and Sana Khan for their significant contribution in developing kinematic model of the robotic arm. REFERENCES
[1] [2] Mark S., Seth H. and Vidyasagar M., "Robot modeling and control", John Wiley & Sons, 2006. Clothier K.E. and Shang Y, A geometric approach for robotic arm kinematics with hardware design, electrical design and implementation, Journal of Robotics, Article ID 984823, Vol. 2010. Sahu S., Biswal B. and Subudhi B., A novel method for representing robot kinematics using quaternion theory, IEEE Conference on

[3]

BIOGRAPHIES

Jamshed Iqbal is working as an Assistant Professor and head of Robotics and Control Research (RCR) group in Electrical Engineering Department of COMSATS, Islamabad, Pakistan. Prior to joining COMSATS, he was working as a Researcher in Italian Institute of Technology (IIT), Italy. He holds PhD in Robotics from IIT, Italy and three Master degrees in various fields of Engineering from Finland, Sweden and Pakistan. With more than 8 years of multi-disciplinary experience of industry and academia, Dr. Iqbal has a strong record of reputed publications and is a reviewer of many international conferences and journals. His research interests include robot analysis and design of rehabilitation robots and industrial manipulators. Raza ul Islam is a Master student at Electrical Engineering Department of COMSATS, Islamabad, Pakistan. He holds Bachelor in Electronics from the same institute. His research interests include robotic vision and Human Robot Interaction (HRI). Hamza Khan is a Doctoral candidate at Department of Advanced Robotics, Italian Institute of Technology (IIT), Genova, Italy. He holds two Master degrees in Robotics from Italy and Poland. His interests include dynamic walking and mechanism design for quadruped robots.

306