Neural network adaptive admittance control for improving position/force tracking accuracy of fracture reduction robot

Abstract

The accuracy of fracture reduction robot is critical factor for clinical application. In this study, a neural network adaptive admittance control (NNAAC) system is proposed to improve the position/force tracking accuracy of the robot under large variable reduction forces and uncertain robot dynamic models. The NNAAC consists of an inner loop and an outer loop. The inner loop combines inverse dynamics control (IDC) and radial basis function neural network (RBFNN) to compensate for inaccuracies in the dynamic model, and an admittance parameter adaptive controller based on the RBFNN in the outer loop improves force tracking accuracy and stability. The performance of the NNAAC is evaluated by fracture model reduction experiments, comparing it with the inverse dynamics control and constant admittance control. The experimental results show that the NNAAC achieves better control effectiveness and high accuracy with a maximum reduction force of 190 N. Specifically, the proposed controller exhibits a maximum position error of less than 1 mm and a maximum orientation error of less than 0.5°.