An Upper Limb Fall Impediment Strategy for Humanoid Robots

Falling is an unavoidable problem for humanoid robots due to the inherent instability of bipedal locomotion. In this paper, we present a novel strategy for humanoid fall prevention by using environmental contacts. Humans favour to contact using the upper limbs with the proximate environmental object to prevent falling and subliminally or consciously select a pose that can generate suitable Cartesian stiffness of the arm end-effector. Inspired by this intuitive human interaction, we design a configuration optimization method to choose a well thought pose of the arm as it approaches the long axis of the stiffness ellipsoid, with the displacement direction of the end-effector to utilize the joint torques. In order to validate the proposed strategy, we perform several simulations in MATLAB & Simulink, in which this strategy proves to be effective and feasible.