A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot

Transcription

1 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot Sergiu-Dan Stan 1, Milos Manic 2, Vistrian Mătieş 1, and Radu Bălan 1 1 Department of Mechatronics, Technical University of Cluj-Napoca, C. Daicoviciu no. 15, Cluj-Napoca, Romania 2 Department of Computer Science, University of Idaho, USA sergiustan@ieee.org, misko@uidaho.edu, {matiesvistrian+radubalan}@yahoo.com Abstract. This paper presents an IG3PR-VRI, a novel Virtual Reality (VR) interface for the control of the Isoglide3 Parallel Robot (IG3PR), an effective robotic device with three degrees of freedom manipulation. The IG3PR manipulator introduced in [24], offers the superior characteristics with regards to the other parallel manipulators, such as the light weight construction on one hand, while on the other hand alleviating some of the typical shortcomings of parallel manipulators such as the extensive spherical joint use and the platform orientation and position coupling. The innovative user interface for high-level control of the Isoglide3 parallel robot presented in this paper IG3PR-VRI was verified and tested on the examples of path tracking performance, and results were presented in MATLAB, Simulink, and SimMechanics environment. Keywords: virtual reality, parallel robots, ISOGLIDE3, simulation. 1 Introduction Many researchers realized the tremendous potential of the virtual reality tools in design, development, and manufacturing of the robotic industry. Virtual reality simulation is done in the artificial environment created with computer hardware and software, presented to the user in such a way that it appears and feels realistic. The definition of virtual reality involves three components (see Fig. 1): 1. real-time interaction, 2. real-time simulation, 3. immersion and direct interaction by novel I/O devices. By using virtual reality simulation with a virtual robot, one can experience a three dimensional design that is still relatively new and be able to perform testing and experience possible outcomes without the expense of having to actually program the real robot. Using of the virtual reality therefore saves resources (time and money), and allows for the work in a controlled, safe environment. C. Xiong et al. (Eds.): ICIRA 2008, Part I, LNAI 5314, pp , Springer-Verlag Berlin Heidelberg 2008

2 1266 S.-D. Stan et al. Fig. 1. The main three pillars of Virtual Reality 2 Kinematics of the 3 DOF Isoglide3 Parallel Robot (IG3PR) The kinematic structure of the IG3PR, the 3 DOF Isoglide3 parallel robot is shown in Fig. 3, where a mobile platform is coupled with the fixed base by the three PRRR (Prismatic Revolute Revolute Revolute) links [24]. Fixed coordinate frame originates at the point O. Here the reference frame XYZ is attached to the fixed base (Fig. 2.). The mobile platform can be visualized as a square whose side length 2L is defined by B 1, B 2, and B 3 points. The fixed base of this square is defined by three guide rods that passing through A 1, A 2, and A 3 points, respectively. Fig. 2. Kinematic scheme of 3 DOF Isoglide3 parallel robot The three revolute joint axes at each of these links are parallel to the ground connected prismatic joint axis, and are located at points A i, M i, and B i, respectively.also, the three prismatic joint axes passing through points A i, for i = 1, 2, 3, are parallel to the X, Y, and Z axes, respectively [24]. The first prismatic joint axis lies on the X-axis; the second prismatic joint axis lies on the Y axis; while the third prismatic joint axis is parallel to the Z axis.

3 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot 1267 Consequently, the location of point P is determined by the intersection of three planes. The forward and inverse kinematic analysis is trivial. A simple kinematic relation can be written as (1) [24]. x d y = d z d 2.1 Dynamic Model of the Mechanical Structure In order to simulate the behaviour of the robot, a dynamic model of the robot has to be built. This is a complex subject and different methods were developed in order to solve it. A classical approach to closed-chain dynamic modeling is to first consider an equivalent tree-structure, and then to consider system constraints via Lagrange multipliers or d Alembert s principle [2]. Other approaches include the use of virtual work, Lagrange formalism, Hamilton s principle, and Newton-Euler equations. In this article the dynamic model of the robot is built using the SimMechanics toolbox from Simulink. The toolbox uses the standard Newtonian dynamics of forces and torques in order to solve both the direct and the inverse problem. The model was built from Simulink blocks that represent the kinematic elements and joints of the robot. These blocks allow modelling of mechanical systems consisting of any number of rigid bodies, connected by joints representing translational and rotational degrees of freedom. In order to build a SimMechanics model, one has to specify the inertial properties of the body such are degrees of freedom and constraints, along with coordinate systems that is attached to each body of the structure. This procedure can be very difficult for bodies with complex geometric forms, however, the process can be simplified by use of a SolidWorks tool. The model of the mechanical structure is presented in Fig. 3. All three kinematic chains are defined by body and joint blocks. The inertia properties and the coordinates of the joints for each body were determined automatically when the CAD model was imported in MATLAB/Simulink environment. The connection of the mechanical model of the robot to the rest of the robot model was realized via actors and joint blocks. Inputs of the model can be one of the following: the generalized force, the position, speed, or the acceleration of the motor joints. In this paper, the inputs chosen were the speed of all three motor joints of the robot. As outputs, the angles of each motor elements were chosen. In addition, sensors were used to determine the position of the end-effector (1)

4 1268 S.-D. Stan et al. Fig. 3. Dynamic model of the robot s mechanical structure 3 Control of the 3 DOF Isoglide3 Parallel Robot The control of the robot is implemented using a joint-based control scheme. In such a scheme, the end effecter is positioned by finding the difference between the desired quantities and the actual ones expressed in the joint space [1]. The command of the robot is expressed in Cartesian coordinates of the end effecter. Using the inverse kinematic problem, these coordinates become displacements. These displacements further become the reference points for the control algorithm. Fig. 4 presents the model of the robot. For the control of the robot, one motor element was used for each actuator of a PI control algorithm. Fig. 4. Simulink model of the Isoglide3 parallel robot

5 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot 1269 The inputs of the algorithm were the differences between the angles computed (via inverse kinematic problem equations), and the values from sensors. The control signal was applied on three dc motors which were actuating the robot structure. The controller parameters kp and ki were optimized for a given trajectory and a maximum error, using the block Signal Constraint from the Simulink Response Optimization toolbox. 4 Virtual Reality Robot Interface for Isoglide3 Parallel Robot (IG3PR-VRI) This section of a paper presents an innovative user interface for high-level control of robot manipulators. The interface is based on a virtual reality approach in order to provide the user with an interactive 3D graphical representation of the parallel robot. The interface was designed to give a novice user an intuitive tool to control any kind of mechanical structure (serial, parallel or hybrid), requiring no programming skills. Computer based simulation allows mimicking of a real life or potential situations. SimMechanics models, however, can be interfaced seamlessly with ordinary Simulink block diagrams. For example, this enables user to design the mechanical and the control system in one common environment. Simulation in Virtual Reality should be used whenever real systems are expensive, dangerous, or for any other reason difficult or impossible to run. With the simulation, one can: - gain better understanding of how the system works; - identify problems prior to their implementation; - test potential effects of changes; - identify areas for resource deployment; - design an efficient and cost-effective system. In addition, Virtual Reality Toolbox for MATLAB makes more realistic renderings of bodies possible. Arbitrary virtual worlds can be designed with Virtual Reality Modeling Language (VRML), and interfaced to the SimMechanics model. The user simply describes the geometric properties of the robot first. Then, in order to move any part of the robot through 3D input devices, the inverse kinematics is Fig. 5. Overview of the virtual reality robot interface

6 1270 S.-D. Stan et al. Fig. 6. Isoglide3 virtual reality robot interface automatically calculated in real time. The interface was also designed to provide the user decision capabilities when problems such are singularities are encountered. VR interface enables users to interact with the robot in an intuitive way. This means that the operator can pick and choose any part of the robot and move it using translation and rotation using 3D sensors, as easily as a drag and drop operation is. Thus, trajectories can be easily defined, optimized, and stored. Not only that, but the virtual world can be accessed and controlled via the Internet, too. The use of a VR interface to simulate robots drastically improves the feeling for the robot. In particular, the interface allows user to understand the behavior of an existing robot, and to investigate the performance of a newly designed structures without the need and the cost associated with the hardware implementation. 4.1 Control Using Joystick With the help of the VR interface, the user can study whether a specific structure is well adapted to the task or not, and can further adjust the structure parameters to optimize the specific task. Various MATLAB tools were used to build upon the previous virtual reality model of the experimental system. For example, Simulink and SimMechanics sub modules were combined to construct the simulation model. The Genius MaxFighter F-16U Joystick will be used to enter position and velocity information into the system. In this paper, instead of a physical implementation, real-time simulations with visualizations that relieve the designer from the costs and risks associated with actual hardware (see Fig. 15) are presented. Because the code was based on MATLAB, the users can easily and rapidly modify the code and experiment with a variety of controllers. Users can also pursue these efforts at home. In contrast to the typical commercial system, the simulated systems can be controlled manually through a joystick interface. The simulation program gives the option of manual or automatic control. Manual control is directly available through a joystick interface, while the automatic control requires that the user provide the appropriate m-files. Manual control is a great opportunity

7 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot 1271 Fig. 7. Simulink model of the Isoglide3 parallel robot Fig. 8. Controlling Isoglide3 parallel robot with a joystick to, through a fun way, get a better insight into the challenges of control of a robot structure. The simulation programs can also easily be modified to incorporate more complex models, or more elaborate systems. In contrast to the usual experiments, users can experiment with the system through a joystick interface, enabling users to gain an appreciation for the challenges of control. The experiments may be used to illustrate fundamental issues in control systems, including stabilization of unstable systems, pole placement, frequency-domain compensation, effect of actuator saturation, discrete time implementation, and others.some drawbacks have been identified typical for the normal human joystick operation. This refers to the imprecision due to the human imperfection in precise movement of the joysticks. 4.2 Collision Detection In general, collision detection is integrated into simulation loops in a following way. Collision detection within a simulation loop loop forever move objects by simulation check collisions take appropriate actions based on collision reports (e.g., highlight objects, do back-tracking, or calculate forces,...) end of loop forever

8 1272 S.-D. Stan et al. Fig. 9. Collision detection with a spherical object Fig. 9. demonstrates how 3 translation actuators (2), (5) and (6) were mounted on the fixed frame (1) that controls the position of the mobile platform. This platform also acts as an end-effector (4). 5 Experimental Test 5.1 Path Tracking Performance and Visualization The first tests on the prototype encourage the direction of the research: the chosen control algorithms emphasize the peculiar characteristics of the parallel architecture and, in particular, the good dynamic performance due to the limited moving masses, and advantageous robot behaviour. For the evaluation of the Isoglide3 parallel robot model performance, a trajectory in 3 dimensional space was used, as shown by Figs. 10 & 11. Various simple simulations were also executed as a way to better understand and test the sensors, movement, and assembly before attempting to implement them in the Fig. 10. A 3D trajectory within workspace

9 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot 1273 Fig. 11. Input data to obtain the 3D trajectory [x, y, z] Table 1. What is to be done in the use of Isoglide3 parallel robot What is to be done Necessary actions Automated calibration algorithms and routines to realize positional accuracy Active compensation for the disturbing forces to be taken up in loaded guiding chains Interfaces for compensation models in the control complex simulation. In the following table is presented what is to be done in the use of Isoglide3 parallel robot. 6 Conclusions The paper presents an IG3PR-VRI, a novel Virtual Reality Interface for the 3 DOF Isoglide3 parallel robot (IG3PR) control. an evaluation model from the Matlab/SimMechanics environment was used for the simulation. An interactive tool for dynamics system modeling and analysis was presented and exemplified on the control in Virtual Reality environment of this Isoglide3 parallel robot. The main advantages of this parallel manipulator are that all of the actuators can be attached directly to the base, that closed-form solutions are available for the forward and inverse kinematics, and that the moving platform maintains the same orientation throughout the entire workspace. By means of SimMechanics, the authors considered robotic system as a block of functional diagrams. Besides, such software packages allow visualizing the motion of mechanical system in 3D virtual space. Especially non-experts will benefit from the proposed visualization tools, as they facilitate the modeling and the interpretation of results.

10 1274 S.-D. Stan et al. References 1. Gosselin, C.: Determination of the workspace of 6-d of parallel manipulators. ASME Journal of Mechanical Design 112, (1990) 2. Merlet, J.P.: Determination of the orientation workspace of parallel manipulators. Journal of intelligent and robotic systems 13, (1995) 3. Kumar, A., Waldron, K.J.: The workspace of mechanical manipulators. ASME J. Mech. Des. 103, (1981) 4. Tsai, Y.C., Soni, A.H.: Accessible region and synthesis of robot arm. ASME J. Mech. Des. 103, (1981) 5. Gupta, K.G., Roth, B.: Design considerations for manipulator workspace. ASME J. Mech. Des. 104(4), (1982) 6. Sugimoto, K., Duffy, J., Hunt, K.H.: Special configurations of spatial mechanisms and robot arms. Mech. Mach. Theory 117(2), (1982) 7. Gupta, K.C.: On the nature of robot workspaces. Int. J. Rob. Res. 5(2), (1986) 8. Davidson, J.K., Hunt, K.H.: Rigid body location and robot workspace: some alternative manipulator forms. ASME J. Mech. Transmissions Automat. Des. 109(2), (1987) 9. Agrawal, S.K.: Workspace boundaries of in-parallel manipulator systems. Int. J. Robotics Automat. 6(3), (1990) 10. Gosselin, C., Angeles, J.: Singularities analysis of closed loop kinematic chains. IEEE-T. Robotics Automat. 6(3), (1990) 11. Cecarelli, M.: A synthesis algorithm for three-revolute manipulators by using an algebraic formulation of workspace boundary. ASME J. Mech. Des. 117(2(A), (1995) 12. Agrawal, S.K.: Workspace boundaries of in-parallel manipulator systems. IEEE Transactions on Robotics and Automation 7(2), (1991) 13. Pernkopf, F., Husty, M.: Reachable Workspace and Manufacturing Errors of Stewart- Gough Manipulators. In: Proc. of MUSME 2005, the Int. Sym. on Multibody Systems and Mechatronics Brazil, pp (2005) 14. Stan, S.: Diplomarbeit, Analyse und Optimierung der strukturellen Abmessungen von Werkzeugmaschinen MIT Parallelstruktur. IWF-TU Braunschweig, Germany (2003) 15. Cleary, K., Arai, T.: A prototype parallel manipulator: Kinematics, construction, software, workspace results, and singularity analysis. In: Proceedings of International Conference on Robotics and Automation, Sacramento, California, pp (April 1991) 16. Ferraresi, C., Montacchini, G., Sorli, M.: Workspace and dexterity evaluation of 6d of spatial mechanisms. In: Proceedings of the ninth World Congress on the theory of Machines and Mechanism, Milan, pp (August 1995) 17. Ceccarelli, M., Carbone, G., Ottaviano, E.: An Optimization Problem Approach For Designing Both Serial And Parallel Manipulators. In: Proc. of MUSME 2005, the Int. Sym. on Multibody Systems and Mechatronics Uberlandia, Brazil, March 6-9 (2005) 18. Ceccarelli, M.: Fundamentals of Mechanics of Robotic Manipulation. Kluwer/Springer, Dordrecht (2004) 19. Company, O.: Machines-outils rapides à structure parallèle. Méthodologie de conception, applications et nouveaux concepts. Dissertation, Universite Montpellier II, Sciences et Techniques du Languedoc (2000) 20. Du Plessis, L.J., Snyman, J.A.: A numerical method for the determination of dextrous workspaces of Gough-Stewart platforms. Methods in Engineering 52, (2001) 21. Liu, X.J.: Optimal kinematic design of a three translational DoFs parallel manipulator. Robotica 24(2), (2006)

11 A Novel Virtual Reality Robot Interface for Isoglide3 Parallel Robot Hesselbach, J., Kerle, H., Krefft, M., Plitea, N.: The Assesment of Parallel Mechanical Structures for Machines Taking Account of their Operational Purposes. In: Proc. of the 11th World Congress in Mechanism and Machine Science-IFToMM 11, Tianjin, China (2004) 23. Stan, S.-D., Manic, M., Mătieş, M., Bălan, R.: Evolutionary Approach to Optimal Design of 3 DOF Translation Exoskeleton and Medical Parallel Robots. In: HSI 2008, IEEE Conference on Human System Interaction, Krakow, Poland, May (2008) 24. Stan, S.-D., Manic, M., Mătieş, M., Bălan, R.: Kinematics Analysis, Design, and Control of an Isoglide3 Parallel Robot (IG3PR). In: IECON 2008, The 34th Annual Conference of the IEEE Industrial Electronics Society, Orlando, USA, November (2008)