From Human Motion Captures To Humanoid Spatial Coordination

This paper describes a methodology translating human spatial coordination in a humanoid robots context. Once the human locomotion is captured, we highlight coordination relations describing motions. Relations and inverse kinematics are applied to a virtual humanoid, which is a tradeoff between human (anthropomorphic proportions) and robot (joints configuration). Further, we quantify all required adaptations for a real humanoid, such as scaling and equilibrium. Finally, this methodology is applied to a specific humanoid robot (called NAO) in order to illustrate and compare some preliminary results.