LPCによるアクロバットロボットの最適着地運動と未知パラメーター推定の同時決定法(<特集>通信と制御)

概要

論文の詳細を見る
C.O.E acrobat robot is considered as a model of a gymnast with a horizontal bar. Our final objective is to perform skillful motions such as swinging from pendant state, forward upward circling, handstand on the horizontal bar, giant swing, fling off, somersault and landing. Such controls which requires high speeds and continuous constraints switching is the one of the most important topic for realizing pragmatic humanoid robots. In this paper, we consider to realize the landing in order to solve Linear Complementarity Problem (LCP) which minimize kinetic energy just after landing with various constraints such as slipping conditions, impact conditions and so on. We also consider a way of estimating a physical parameter using a feature of LCP that gives the Lagrange multipliers of various constraints simultaneously, and also using the relationship between Lagrange multipliers and sensitivities of the constrained system. In this paper, we choose a friction coefficient, that appears in slipping conditions, as the physical parameter and estimate it. Using numerical simulations, the effectiveness of the proposed method are shown by that it can generate the optimal landing motion and estimate an unkown friction coefficient simultaneously.
日本シミュレーション学会の論文
2005-06-15