A biological humanoid joint controller based on muscle model

It is challenging to design a humanoid robot control framework that maintains high-level motion performance in an unknown environment as this requires complicated sensing capabilities and time-consuming process. On the other hand, basic autonomous motion control in joint space is relatively easier and faster to fulfil dynamic tasks. Based on modeling of human muscles, this paper proposes a biological joint controller developed from biological patterns of muscle activity and dynamics of internal forces. Combining autonomous and synergic control, the controller has the capability of fast active response of environment and autonomous recovery from abnormality. The experiment in Waseda Kyotokagaku Elbow Robot verified the effectiveness of the proposed biological joint controller.