Lifting techniques for the humanoid robots: insights from human movements

Abstract

As a possible future task for humanoid robots, lifting movement of human-like object is experimentally studied. Our aim is to understand what is the knack, which is a important information to improve the skill level. Using a motion capture device, lifting of a human-like dummy (about 30 [kg] weigh) is measured for both experienced and inexperienced subjects. We found that three distinctive feature of experienced subjects: start with the legs open, no arm lift, and no lying back. Also, one of the participant used leverage for initiating the motion. These features are translated to the knacks and we suggest full extension of the arms, no arm lift, open strategy and use of leverage as the candidates of the knacks. For humanoid robotics, we believe these knacks are transferable to humanoid robots and expected to useful for designing behavior. Some of the candidates are about preparation of lift and reduces complexity of parameter tuning or searching space and the last one helps energetic efficiency.