Neural-network-based trajectory error compensation for industrial robots with milling force disturbance

Abstract

Industrial robots are increasingly used in advanced manufacturing fields such as aerospace due to their high efficiency and low cost. In robotic machining applications, deviation of the tool centre point trajectory from the desired path due to load disturbances acting on the end-effector of an industrial robot can result in poor dimensional accuracy and surface quality of products. Therefore, improving robot trajectory accuracy under external load disturbances is extremely important. This study proposes an error compensation methodology using neural networks optimized by a hybrid marine predators-grid search algorithm to compensate for robot trajectory error. Two neural networks are developed: one for predicting load disturbances and the other for predicting trajectory errors in robotic machining. The milling experiment results show that the compensated robot trajectory errors in x, y, and z directions are reduced by 65%, 76%, and 77% respectively, which proves the effectiveness of this method in improving the robotic milling accuracy.