Abstract: Focusing on the ball double-screw hydraulic knee joint as the research object, this paper analyzes the load driving performance of the hydraulic knee joint. Taking the posture data of the human body such as walking, squatting and landing buffer as initial learning objects, motion features are extracted. By simplifying the trajectories of different motion actions into key feature control points and flexible trajectory fitting, the trajectory of joint actions is optimized. This method can realize the adaptability of the hydraulic robot knee joint in different movements, and take the flexible action as the optimization goal under the condition of ensuring the movement performance, so as to reduce the damage to the knee joint caused by the foot impact in motion. The simulation model was built by Adams and Matlab to complete the performance analysis and motion optimization experiment of the knee joint. The simulation results show that the foot impact force of the experimental model decreases gradually through optimization. Finally, the method is applied to the hydraulic joint experimental prototype to prove its load capacity and flexible motion control performance.

Key words:  , Bionic robot , · Ball double-screw hydraulic joint , · Robot knee joint , · Performance analysis , · Humanoid motion control , · Flexible control