TMCnet News
The Research on Collision-free Trajectory Planning for Welding Robot Based on Fishertechnik Model [Sensors & Transducers (Canada)](Sensors & Transducers (Canada) Via Acquire Media NewsEdge) Abstract: Based on the path and collision-free that welding robot should do in the period of machining, a trajectory plan for certain collision-free movement for robot is provided in this paper. The first is searching the collision-free path through subdivision the working space by tangent plane projection in concentric cylinder coordinate system. The second is to assemble the welding robot with three degrees of freedom in order to confirm the collision-free trajectory when welding. Finally, obtain the optimized path by software simulation. Copyright © 2013 IFSA. Keywords: Welding robot, Collision-free, Trajectory, Simulation. (ProQuest: ... denotes formulae omitted.) 1. Introduction When obstacles are in the robot's movement environment, in order to reach a target position, a collision-free movement path need to be determined, this problem is called collision-free path planning. Here, the "planning" is actually intuitive meaning for solving constrained geometry problems, rather than the sequence of operations or action steps. On the other hand, if the moving object as a certain state of the research question, the obstacle as an end condition of problem, no collision-free path will meet the constraints of the solution, then, a collision-free path planning is kind of solving multi-constraint problem process. In order to facilitate the analysis, the welding robot selected is fisher welding robot model with three degrees of freedom and three motor driving [1,2]. 2. Obstacles and Collision Detection In a complex environment for a robot in the welding station, it is extremely possible collision would happen between robot and static equipment (such as work piece, positioning seat, fixtures, etc.) or collision between robots themselves because the machine station intensively distributed and machine space is relatively small. The robot working space can divided into free space and obstacles space when path planning. Robot that moves only in the free space can avoid collision problems. For the collisions between the robot and stationary equipment can be solved in advance. The key is to analyze the obstacles distribution in space [3]. 2.1. Definition of Obstacle Obstacle is all equipments and objects that hinder robots' welding operation, such as work piece, fixtures, transport device, support device, auxiliary machine equipments, and other robots. There are two kinds of obstacles, static obstacle and dynamic obstacle. 1) Static obstacle. Static obstacle refer to the position of the device is unchanged relative to the ground coordinates when the working robot move, Usually before trajectory planning for the welding robot, layout for static obstacle in space can be get. The static space layout can be defined by projection of section in concentric cylinder coordinate system, thus provide the possibility for collision-free trajectory plan for welding robot. 2) Dynamic obstacle. Dynamic obstacle refers to the objects which position would change relative to the geographical coordinates when the working robot moves, such as the moving fixture, the moving carrier, and other working robots. 2.2. Analysis of Space layout for Static Obstacles in Welding Station The analysis method of tangent projection plane in concentric cylinder coordinate is provided considering the characteristic of welding robot. It expressed the layout of obstacles on working position. The clearance between the concentric cylinders can be planned considering the welding torch used in welding to ensure the reasonable planning and simplifying the algorithm [4, 5]. 1) Subdivide the working space. Select the starting point of the robot for the initial position qn ; select the first welding spot for the end spot qo , using a series of concentric cylinders to divide the robot working space, as shown in Fig. 1. Concentric cylinders should be selected so that the axial direction parallel to the Z axis coordinate system of the entire working station, the axis go through the center of the initial position of the robot, and the end spot is at the cylindrical surface which with maximum radius. In order to further subdivide for the space, reselect a series of parallel plane which perpendicular to the cylinder axis to cut for the welding position. As shown in Fig. 2. Obtain the layout of two-dimensional projection for the equipment. Then the three-dimensional question is converted into two-dimensional question for further study. In the process of series of plane cut station space, since the initial position of the welding torch and the first welding spot may not be on the same plane which parallel to the horizontal plane, the projection of initial position and end spot on each plane is selected as the corresponding start point and end point in order to speed up plan when series of concentric cylinders is cut [6]. How to select the clearance between the concentric cylinders and how to select the clearance between the cutting planes direct affects the efficiency of path optimization algorithms when speed up plan in robot moving space. If the clearance made too large, it may result in missing some obstacles. If the clearance made too small, it may increase the computation time. For this reason, the variable clearance of concentric cylinders and variable clearance of cutting plane is used. 2) Obstacle layout. The clearance of concentric cylinders and clearance of cutting plane is expressed as Ô. According to the initial clearance value 8, select d = 108 to subdivide the robot workspace and then project on two-dimension plane so as to transform the three-dimension problem into two-dimension problem. By such subdivision, each projection after series of plane cut should be a series of circle which contains all obstacles. The clearance between the circles is determined by the selected value d (as shown in Fig. 2). Thus obtaining k-th concentric circles on plane, the expression of radius is: [7, 8]. ... Considering Fig. 3, the polar coordinate system that endpoint qo set as circle center is established, and then obtain: ... Obstacles located in the concentric circle (set of m) can be expressed as: ... According to the result of subdivision, the midpoint of the free space between the two obstacles can be select as the point that the robot must pass through, the midpoint is cascaded, and is tentatively fixed as the initial welding trajectory, and then analyze if it's a feasible moving trajectory by the collision experiment. If it is feasible, the selected trajectory is the feasible machining path. If it is not, analyze which space segment that collision may occur then re-plan. For the collision area when re-plan, select d = 50 then analyze the feasibility of machining path. If a collision is still occur, repeat the above steps and select d =20 when re-plan. If collision is still exists until d = Ô is selected, then no viable welding path for select in the current device layout. 2.3. Collision Detection The most important step of path planning for welding robot is detecting where the collision occurs in machining, and then avoids it. The path planning question is always around the collision detection and how to avoid them. 1) Collision definition. Collision is two objects have same space region and interfere with each other. Usually defines that object A is a collection of a series of Ai objects, the object B is another collection of a series of Bi objects, that is ... Then if the collision occurs between object A and B, the collision is defined as ... That is, part of the object A and part of object B have disturbed each other when collision occurs. 2) Dynamic obstacles. Research of layout of station obstacle is mainly on static obstacle. However, in the welding process, the dynamic obstacle is formed between the robots and between the jigs. As the dynamic obstacle position is not fixed, so the research will focus on moving trajectory, and put forward the robot's moving priority and moving rules in order to achieve collision-free path [9]. 3. Assemble of Welding Robots Models In order to verify the theory mentioned above, the welding robot model with three degrees of freedom was assembled based on fisher. The main components of the model are manufactured by high quality nylon plastic, with characteristics of accurate dimension, not easy to wear, compact organization and flexibility. It also can repeatedly dismantle but not affect the accuracy of the model. The patented design of the part is dovetail groove, making it's possible for part joint in six directions. The unique design feature can easily realize structure developed. The welding robot assembled by Fisher is shown in Fig. 4. The structure of the robot is consisted of three parts: the mechanical part, the sensor part and control part. Mechanical part is mainly consisted of the mechanical body, three DC motors, gears - worm drive mechanism and four bar linkage. DC motor drive the mechanical part working which shaft is connected with spur gear. The shaft of worm is driven by spur gear. Wherein the motor shaft is the output shaft and the worm shaft is the working shaft. When the worm rotates then drives three worm wheels working. The LLWin.3.0 software of Fisher is a kind of graphical programming software with characters of easy to use and real-time control. When 31002 intelligent interface boards are used to control the model, ladder programming is adopted. 18 kinds of software modules can be programmed in any combination. It can graphically display, automatically connect. Program is shown in Fig. 5 [10]. Welding robot has three degrees of freedom. The movement of the robot depends on the drive of three motors. Motor Ml is responsible for the large arm rotation of the welding robot. Motor M2 is responsible for the forearm stretch. Motor 3 is responsible for the mid-arm pitching. The function of switch El limits the rotation of large arm. Pulse counter E2 is set for large arm rotation counter to control the rotate angle for the large arm. Switch E3 is set for limit switch of forearm. Pulse counter E4 is set for stretch count of forearm to control the length of stretch. Switch E5 is set for limit switch of up and down for mid-arm. Pulse counter E6 is set for pitching angle control of mid-arm. M4 is a small bulb which represents the welding torch working or stop when it's on or off. The working principle of pulse counter is: when the pulse counter is controlled by the rotary-disk pulsar, it would switch five times during each rotation. And the rotary-disk pulsar is connected with the motor reducer, so the number of turns of the motor can be measured by the pulse counter. The mechanic arm of the welding robot shows a clearly moving path which has three degrees of freedom. Before work, the welding robot is reset by switch El, E3, and E6. Then the welding robot did the movement of rotation, pitching, stretch at the same time in the three degrees of freedom. The motor M7 is working then does the movement to the welding spot for welding on the control of pulse counter E2, E4 and E6. The counter value of E2, E4 and E6 is assigned to three different welding places. In order to achieve barrier-free movement, the welding torch can be removed or extended according to the requirement. After finished welding in the three spots, the robot is reset before next cycle. 4. Welding Trajectory Plan 4.1. Welding Trajectory Plan When Robot Machine Selection of welding path can use the method of inversed order algorithm, i.e., seek the path from the welding end to the start point. Find out the possible welding path by combine the method of subdivide the working space and the method of collision-free area selection. The possible welding path can be found from the end point to the start point on the projection of initial position to end spot on each plane. Try to avoid the possible collision area or deviated area when path plan [11]. As shown in Fig. 6, for example, start from the end point to find out the non-collision pitch point 1, 2 on the circle which is most near the end point. The principle of pitch point selected is a little away the barrier edge considering the welding torch. Connect the end point to the pitch point, and then measure its distance. Then start from this pitch point, connect other pitch point on the adjacent circle till connect with the start point. As shown in Fig. 7. Based on the theory of non-collision detection in the experiment, the LLWin.3.0 program is adjusted by right click to modify the input and output parameters after the reference value is calculated. The collision-free path for the robot should be the connected path before mentioned. The query to the obstacles distribution for the processing equipment and work stations can be made in many cases [12]. With simulation software can analyze the obstacles on every layout. The selection of robot machining path is to avoid these obstacles, select the space of collision-free road to complete welding tasks. Although a collision-free path can be selected in Fig. 8, but this is mainly focused on the twodimensional plane. This also needs change the problem of two-dimensional plane into threedimensional plane, as described below. 4.2. Simulate for Trajectory Plan In Rob CAD simulation environment, all kinds of CAD model which simulate mechanical hand, racks, shelves and parts are called into through the data conversion interface of physical model to plane model. The type of robot need select before build the simulation model. For the station layout that robot located, with the center of the base as a reference point, Build series of concentric cutting cylindrical plane in robot working space, combining the method mentioned before of subdivide working space and considering the method of tangent projection. Then speed up the algorithm of convergence for cutting working space. The path optimization problem of three-dimensional working space can be transformed into two-dimensional problem. As shown in Fig. 8. The obstacle layout of the machine equipment and station is clearly checked up [13]. Although the welding robot can select a collisionfree path in Fig. 8, it still concentrates upon twodimensional picture. Therefore, something in the two-dimension plane needs to revert to that of in three-dimension. Subdivide the working space of robot along the axial direction of concentric circle. Select and collect all the free space on corresponding two-dimensional cutting plane. Then overlay the space in two-dimensional plane and expand it in three-dimensional space. Find out a feasible welding path in space. However, the model has been simplified in the process of the working space subdivided. It may results the collision happen between the equipments when simulation. At the collision place, the collision equipment should be cut along the normal direction. Then get its projection on two-dimensional plane. By this method, the whole working path in the working space and welding place can be got. The welding place and the key point in the movement are linked in series according the path calculation result on interpolation method. Then a complete three-dimensional space machining path can be got. After running the simulation, the simulation results which include the welding time, welding place and the moving path are output respectively. Run the simulation to verify the welding path after modify the input parameters of welding robot [14]. 5. Trajectory Optimization for Welding Robot Because the overlap area is limited on the entire space station and it's more limited for essential pass point of welding path on the overlap area as well as the number of welding spot. Thus, the machine priority can be used to solve the collision problem. As shown in flow Fig. 9. Using the method above to find the possible welding path can get many different paths. The every path can become the welding path. Therefore, the optimal path should be selected from the entire possible path. After selected two or three path, verify all of them in the virtual manufacturing environment then find the optimization plan. When simulation in this program, the perfect results has achieved though the error still exists when simulate for direct and inverse kinematics solution. 6. Conclusions The trajectory selected and trajectory optimization for one welding robot have discussed in the paper. In practice, many robots usually work together to finish the welding task. It's possible that collision may occur between them. The mainly happened area is concentrate on overlap area of two robots. That is the dynamic barriers between them. The solution method is: 1) Define the welding priority before working. When collision area is detected, the robot with higher priority works first. The robot with lower priority then waited or detour the collision area. 2) Replants the welding path to solve the collision problem. The purpose of trajectory plan and analyze is to build the simulation model for verify in the future. This method is of universal applicability and transplant ability for avoiding obstacles of welding robot. Foundation item: Supported by the National Natural Science Foundation of China (51105261) References [1] . Peng Zhi, Robot collision-free path planning, Shenyang University, Shenyang, 2003, pp. 54-56. [2] . Cai Zi-Xing, Jiang Zhi-Ming, Senior robotics planning based on expert system, Journal of Electronics, 10, 2001, pp. 88-91. [3] . Wang Wei, Yang Yang, Yuan Kui, The robot modeling for C-space obstacle boundary and collision-free path planning, Robots, 20, 4, 2008, pp. 280-283. [4] . Xiong Youlun, Robotics, first edition, Mechanical Industry Press, Beijing, 2000. [5] . Liu Chengliang, Zhang Weigong, The research of Kinematics and simulation for welding robot RV 12L6R, 20, 5, 1998, pp. 333-335. [6] . Yan Junqi, Virtual manufacturing theory and practice, Shanghai Jiaotong University Press, Shanghai, 2003, pp. 191-195. [7] . Zhu Xiangyang, Zhong Binglin, Xiong Youlun, Detection for collision-free path planning of robot, Applied Science, 16, 1998, pp. 106-107. [8] . Cai Zixing, Robotics, Tsinghua University Press, 2000. [9] . Lin Changyang, Cheng Shanben, Li Chengtong, Welding robot and its application, first edition, Mechanical Industry Press, Beijing, 2000. [10] . L. L. Win 3.0 Operating Manual, German Fischertechnik. [11] . Ma Xiangfeng, Robot mechanism study, first edition, Mechanical Industry Press, Beij ing, 1991. [12] . Paul R. P., Shimano B. E., Mayer G., Kinematics control equations for simple manipulators, IEEE Traps SMC, 11, 6, 1981, pp. 449-455. [13] . Lee G. S. G., Ziegler M., Geometric approach in solving inverse kinematics of PUMA robots, IEEE Trans AES, 20, 6, 1999, pp. 695-706. [14] . Weinrab A., Bryson A. E., Optimal control of systems with hard control proc, in Proceedings of the Amer Contr Confer, 1985, pp. 1248-1252. Liu Ming Dan, Lü Xiao-Rong, Qixiang-Jun The Information & Engineering Technology College Sichuan Agricultural University 625014 Sichuan Yaan, China E-mail: [email protected] Received: 9 October 2013 /Accepted: 22 November 2013 /Published: 30 December 2013 (c) 2013 International Frequency Sensor Association |