Dynamic Load Adaptive Control for Quadruped Robots Based on Model Prediction

To enhance the reliability of quadruped robots in practical applications and their ability to adapt to changes in non-structured dynamic environmental parameters, a sliding mode adaptive control algorithm based on model predictive control (SMAC-MPC) is proposed. The algorithm incorporates sliding mode adaptive control into quadratic programming to determine the ground reaction forces for torque control of a quadruped robot, adapting to system uncertainties and variations to improve performance during robot interaction with unknown objects. The simulation simulated three tasks of loading, transportation, and unloading of an unknown load by a quadruped robot, and conducted static variable load testing and dynamic motion load testing on the quadruped robot. Within the allowed range of body tracking error, the robot still maintained good robustness, demonstrating the effectiveness of the proposed method.