IJCAS

The development of hydraulically driven heavy legs that can withstand external interference for realizing the high-velocity dynamic walking of bipedal robots with eight degrees-of-freedom is challenging. Therefore, in this study, a cascade antidisturbance algorithm was proposed for highly dynamic trajectory tracking based on model prediction and task hierarchical optimization. First, in the upper layer, the time-sharing control framework of underactuated robots based on the single rigid body model ignoring the legs was designed. Linear model predictive control (MPC) was designed to calculate the contact force spin to control the posture and height of floating base in the stand phase. The desired foot location principle was used to control the forward and lateral velocity in the swing phase. Next, in the lower layer, task hierarchical optimization control (THOC) was designed to track the contact force spin predicted by MPC. The relaxation variable of the force spin was designed in the optimized variable and subsequently used to compensate for the contact force between single rigid body and whole-body dynamic models. Thus, the tie relationship was developed between the upper MPC and lower THOC. The control robustness of the proposed model under high-velocity locomotion and disturbance was verified by performing simulation experiments investigating high-velocity walking and external impact, and the fast walking velocity was increased from 2.15 m/s of nonlinear MPC to 2.5 m/s with accurate velocity tracking.

Keywords: Bipedal robot, cascade anti-disturbance control, high dynamic antidisturbance locomotion, model predictive control, task hierarchical optimization control.