Error Compensation for a Parallel Robot Using Back Propagation Neural Networks

Abstract

This paper presents an error compensation method for a novel 6-DOF parallel robot using back propagation neural networks. The main error sources of the parallel robot are discussed. In order to improve the measuring accuracy, a specific target measured by a coordinate measuring machine is designed. The relationship of error map is established by the coordinate transformation. Experimental results showed that the pose accuracy of the parallel robot was improved by more than 60% after error compensation. A precise optical assembly was accomplished by the parallel robot after error compensation reliably and rapidly.