Gait coordination control of bipedal modular reconfigurable robot based on adaptive neural network

This paper proposed an adaptive neural network (NN) control method to track the desired trajectory of a bipedal modular reconfigurable robot (MRR), which can solve the gait coordination problem of bipedal MRR. The leg dynamic model and the body dynamic model of bipedal MRR are established based on the Newton-Euler iterative method, and the global dynamic model is subsequently established. Aiming at the gait coordination problem between double support phase (DSP) and single support phase (SSP), the desired trajectory is generated based on the zero-moment point (ZMP) method. The adaptive NN controller is designed to track the generated desired trajectory, which also compensates interconnected dynamic coupling (IDC) effects of the bipedal MRR. The stability of bipedal MRR system is proved by Lyapunov theory. In the end, the effectiveness of the control method is verified by comparative simulation. The simulation results show that the proposed adaptive NN method reduces the position tracking error by similar to 10 % and the control torque by similar to 15 % compared with the existing control methods.