Research on Obstacle-Avoidance Control Algorithm of Rehabilitation Robot Combined with Neural Network

Abstract

This paper proposes an obstacle-avoidance algorithm which can be applied to omnidirectional Lower Limbs Rehabilitation Robot. The method generates the attribute of obstacle scene using detected angle information of obstacle boundary, and fuzzy matching with the obstacle scene knowledgebase to realize the obstacle-avoidance control of robot. The implementation of fuzzy matching depends on fuzzy neural network. The neural network enhances self learning ability, greatly simplifies data calculation. Obstacle-avoidance algorithm makes the rehabilitation robot more safety for human using and reduces environmental using demand. This paper proves the feasibility and effectiveness of the obstacle-avoidance algorithm through theoretical derivation and simulation results.