Generating human-like motions for an underactuated three-link robot based on the virtual constraints approach

In the field of robotics there is a great interest in developing strategies and algorithms to reproduce human-like behavior. In this paper, we consider motion planning and generation for humanoid robots based on the concept of virtual holonomic constraints. For this purpose, recorded kinematic data from a particular human motion are analyzed in order to extract geometric relations among various joint angles defining the instantaneous postures. The analysis of a simplified human body representation leads to dynamics of a corresponding underactuated mechanical system with parameters based on anthropometric data of an average person. The motion planning is realized by considering solutions of reduced system dynamics assuming the virtual holonomic constraints are kept invariant. The relevance of such a mathematical model in accordance to the real human motion under study is shown. An appropriate controller design procedure is presented together with simulation results of a feedback-controlled robot.