Bipedal walking control of humanoid robots by arm-swing

This work shows a method to improve stability and energy efficiency by arm-swing for a bipedal walking of humanoid robot. For stabilization, the robot calculates the moment of a swing leg and current ZMP, and swings both arms in order to cancel the moment and compensate ZMP error. For improvement of the efficiency, we propose a twist walking of a coxae using inertial force by arm-swing. Inertial force generated by swinging the leg and both arms reduces the load of yaw-axis actuators on the coxae. As a result, the humanoid can increase the stride and energy efficiency. Simulation verified these arm-swing effects. Furthermore, experiment result of walking on a rough terrain confirmed the effect of stabilization by the arm-swing.