Dynamic postural control method for biped in unknown environment

摘要

We propose a new way to solve the dynamic equilibrium problem of a biped robot under unknown external perturbations. For that, a multichain mechanical model of a human body (called BIPMAN) is proposed with a general architecture. The principle of the dynamic postural control is based on the correction of the trunk center of mass acceleration and on the force distribution exerted by the limbs on the trunk. To ensure a real time dynamic compensation faced to an unknown external perturbation, we have introduced a new approach called real time criteria and constraints adaptation. This is based on a linear programming technique with a general criterion which optimizes the force distribution. The ponderation coefficients and the task constraints notions are introduced in order to specify criteria and constraints adapted to specific task classes in presence of an unknown external perturbation. Many results are presented to demonstrate criteria and constraints influences on the dynamic equilibrium.