Research on Admittance Control of Massage Robot Based on RBF Neural Network Sliding Mode Algorithm

In addressing the issues of constant force control and trajectory tracking in a robot performing massages on the human body, a robust admittance control algorithm based on the RBF neural network controller is proposed. Firstly, the admittance control strategy is employed to establish the relationship between the robot's force and position, enhancing the robot's smooth operation during the massage. Subsequently, to achieve rapid and stable tracking of the desired trajectory, the RBF neural network is utilized to mitigate modeling errors and environmental disturbances. A sliding mode robust term is introduced to enhance system robustness. Simulation experiments to evaluate the disturbance rejection performance of the RBF neural network robust controller were conducted using MATLAB/Simulink. Furthermore, a massage robot platform was established to perform a comparative validation with traditional admittance control for constant force massage. The results indicate that the admittance control based on the RBF neural network sliding mode robust term can effectively achieve constant force tracking, with force fluctuations within approximately ±0.5N. Compared to admittance control, the force tracking error is reduced by approximately 1N, ensuring the safety of the massage process.