ROBOTIC INVERSE DYNAMICS FOR SIMULATION OF VIRTUAL HUMANS IN VIRTUAL MANUFACTURING AND DESIGN

摘要

In this paper we consider the issue of inverse dynamics as part of a Physically Based Modeling and Simulation framework for 3D-motion modeling of "virtual humans" with its application to analysis of human posturing and motion. We describe and compare two main approaches: Lagrangian and Newton-Euler formulations. We also describe the main differences between a human model and robotic model. The efficient implementation of the Newton-Euler approach is described and illustrated by the design of an Inverse Dynamics Engine in the Jack software. The issues of the object-oriented design of the engine and the fulfilment of high-end interpretive language on the top of it are addressed. Our computational model assumes that, in addition to the forces caused by gravity, other external forces could be applied to the Human Figure Model. We also address problems of involving the Inverse Dynamics Engine in the motion simulation initiated by Jack's animation system. The Inverse Dynamics Engine is employed by the Human Force and Torque Analysis application which allows users to compute forces and torques for a full human figure or for an optional joint combination chosen by the user from the list of available options. The typical customer problems, which can be facilitated by using this application, are presented.