APPLICATION OF DIFFUSION THEORY-BASED LEARNING METHOD IN COMPUTING INVERSE MAP OF REDUNDANT ROBOTS

This paper introduces a diffusion theory-based learning method for the inverse map of redundant robots. First a partial differential diffusion equation is applied to solve the inverse map of robots which can keep the topology conserving performance during mapping.Then,the feedback error learning is applied to reduce the learning errors. Based on the above,a parallel decentralized computation architecture is put forward to accomplish the inverse map.The results of analysis and simulation show that a continuous inverse map can be attained with high speed and high precision by the method which needs only a few of trial motions.