Study of Series-parallel Mechanism Used in Legs of Biped Robot

Abstract

Walking is an essential ability for biped robots, and the stability of walking is affected by both mechanical structure and control algorithm. In this work, a new design of legs with low moment of inertia was proposed. Due to the application of parallel mechanism, the inverse kinematics solution was analytically derived, while the forward kinematics solution was numerically obtained, using the neural network method. Walking at 2.0 km/h was achieved in dynamics simulation, and the results of original design and new design in the same gait planning method were compared. It's indicated that the torque requirement of new design was lower than that of original one. The torque requirements for hip yaw, hip pitch, knee pitch and ankle joints decrease with the decreasing moment of inertia of legs, and those for hip roll, knee pitch and ankle joints decrease with the decreasing mass of whole robot. Moreover, the parallel mechanism makes the actuators support the motion of joints together, leading to the decreasing torque requirements.