Improvement of Obstacle and Singularity Avoidance Path Planning Algorithm for Redundant Manipulators

Abstract

A collaborative optimization scheme of obstacle avoidance and singularity avoidance path planning method is presented for redundant robot. An improved real-time minimum distance calculated method is presented and search the connect rod which easy to collision based on this minimum distance. Complete the obstacle avoidance based on the self-motion of the redundant manipulator on a null space and two obstacle avoidance parameters related to real-time minimum distance are introduced to improve optimization of obstacle avoidance. Adopt the DLS method to solve the problem that very high joint velocities in the vicinity of singular configuration. At last, through simulation of planar 3R redundant manipulator, the algorithm proves to be feasible and effective. Introduction The assembling work is more complicated as the miniaturization of 3C products and then the high speed manipulator was introduced by manufactures as auxiliary equipment of the assemble work to improve the assembly efficient and quality. The risk of man-machine collaboration is increased significantly because the collaborators and the robot working space overlap in large areas while assembling product in man-machine collaboration. Moreover, the usual inverse kinematics solutions based on Jacobian pseudo-inverse cause very high joint velocities in the vicinity of singular configurations, and it will lead to the manipulator deviate from expected trajectory and will have a certain influence on assembly quality. So if we can design a path planning method which can make the robot complete obstacle and singularity avoidance automatically, then it will make the man-machine collaboration system large-scale application in the field of 3C product assemble. Compared with the traditional manipulator the redundant manipulator has the additional degrees of freedom, makes it has the advantages of flexible operation and can complete obstacle and singularity avoidance. In the field of obstacle-avoiding of the manipulator, the main research method including the artificial potential field method,fuzzy method, neural network method, genetic algorithm, Probabilistic Roadmaps method,Rapidly-exploring Random Tree method.However, these methods have some shortcomings and the contradiction between the optimal path the planning time and the complexity of algorithm is very difficult to solve. In the area of singularity-avoiding of the manipulator, the main research methods such as the Damped least square method, SICQP method, normal forms method. But these method is difficult to solve the contradiction of the tracking accuracy and the complexity of algorithm. An improved calculation method is proposed in this paper to compute the minimum distance in order to improve the real-time performance of the system. Two obstacle avoidance parameters related to minimum distance are introduced and the self-motion in null space of the redundant manipulator is utilized to accomplish obstacle avoidance. The Damped Least Square method is adopted to optimize manipulator performance in the vicinity of singular configurations, complete the optimization between singularity avoidance and obstacle avoidance for redundant manipulator. At last through simulation of 3-DOF redundant manipulator, the scheme proves to be feasible and effective. 5th International Conference on Computer Sciences and Automation Engineering (ICCSAE 2015) © 2016. The authors Published by Atlantis Press 959 Traditional distance calculation method For the traditional minimum distance calculation method, the first step is calculating the minimum distance between obstacles and each manipulator connecting bar of the manipulator and then taking the minimum value. Take a 3-DOF, for example, as shown in Figure 1(a), the real-time minimum distance can be described as dmin=min{d1,d2,d3}. While the manipulator motion to some special position such as shown in Figure 1(b), the pedal of the obstacle to the manipulator connecting bar may fell on the extension of the bar, because the extension of the rod is not part of the manipulator, so it need to selected some mark point from the connecting rod and compute the distance between the mark point and the obstacle, and then taking the minimum value. But this minimum distance is not accurate, and this method will lead to huge calculation, low efficiency and longtime of computation.