Plenary Lecture

Obstacle Avoidance for Kinematically Redundant Manipulators Based on an Improved Problem Formulation and Two Recurrent Neural Networks

Professor Jun Wang
Department of Mechanical and Automation Engineering
The Chinese University of Hong Kong
Shatin, N.T., Hong Kong

Abstract: With the wide deployment of kinematically redundant manipulators in industrial applications, obstacle avoidance emerges as an important issue to be addressed in robotic motion planning. In this talk, we show the formulation of the inverse kinematic control of redundant manipulators with obstacle avoidance task as a convex quadratic programming problem with both equality and inequality constraints. Compared with our previous formulation, the new problem formulation is more favorable with better solutions or bigger solution set to the problem. To solve this time-varying quadratic programming problem in real time, two recurrent neural networks are applied to compute inverse-kinematic solutions with obstacle avoidance capability in real time. The effectiveness of the proposed approach is demonstrated by using simulation results based on the Mitsubishi PA10-7C manipulator.