Analyzing the “knacks” of transferring human and developing a two-body integrated model as a basis for humanoid control

Abstract

In order to help with care work and rescue operations, humanoid robots should have the ability to lift and carry humans gently and steadily. This paper is a step towards the ultimate goal of giving humanoid robots the ability to lift and carry humans, a task which involves very complex physical interactions. We measure several such lifting and carrying motions using a motion capture system, paying special attention to "knacks" which are motion patterns which are critical to success in the task. The knack movements are basically simple, so if they can be found, they can be used to obtain control methods enabling human-lifting motions for humanoid robots. In the experimental results, we found two common knacks between three subjects, and that falling momentum plays a very important role in the motion. Moreover we discovered that when a subject uses the two special knacks effectively, along with a rotation movement, the load on the waist is decreased. Thus, we propose a motion model for humanoid robot control that includes translation movement and rotation movement, and we expect that this model will have the ability to express various human handling motions.