GRNN-based error-compensating algorithms in feeding beam of tunnel Rock-drilling robot

Tests showed that, at the feeding beam´s end, there are large position and pose error during the boring orientation of Rock-drilling robot. And the error are relate to both the roll angle and the extended length of the feeding beam. With analyzing the flexible deformation about different fed length, it is showed that flexible deformation is not the main reason of the error. The error is also caused by many reasons which are difficult to establish the mathematical model. In this paper, a GRNN (General Regression Neural Network) method is introduced into predicting and compensating orientation error, and satisfactory results have been obtained.