Advancing Locomotion Control of a Quadruped Robot: Harnessing Hybrid Control Algorithms and Digital Twin Technology

In order to achieve effective locomotion in complex terrain, it is crucial to implement a robust dynamics control scheme for the quadruped robot. This paper proposes a novel approach to locomotion control, utilizing both model predictive control (MPC) and sliding mode control (SMC). The proposed control scheme decomposes the whole control problem into body level and leg level. Firstly, the dynamics of the body are simplified and summarized into quadprog problem (QP), and the MPC-based controller is designed to solve the problem which is constrained by the range of ground reaction forces (GRFs) and the friction cone. Secondly, the neural network-based adaptive faster fixed-time nonsingular terminal sliding mode control (NTSMC) is utilized for the leg control. Thirdly, the proposed control scheme is verified of a trotting gait with the speed of 0.5 m/s and the simulation results indicate that the swing leg control has a tracking error below 0.01 m, and the body orientation errors are below 0.005 rad for roll, pitch, and yaw. Lastly, the proposed control scheme is validated with a digital twin in Webots.