Self-balanced motion planning for humanoid robot based on dynamic multitasking mechanism

Abstract

Maintaining Balance plays an essential part in humanoid robot's motion planning. In this paper, a dynamic priority-based multitasking mechanism is put forward and null space is used to solve the redundant problem. Self-balancing constraints are introduced into task space to merge with end-effector tracking tasks. A pickup behavior is divided into walking, arm swinging and bending forward tasks, which utilizes linear inverse pendulum and momentum control method. Also, self-collision avoidance is treated as a self-balancing constraint. The proposed algorithms are carried out on Nao robot in simulation and the results show that the robot is able to keep balance in real time while executing the pre-planned motion.