Model of the humanoid body for self collision detection based on elliptical capsules

Abstract

This paper presents a self collision detection scheme for humanoid robots using elliptical and circular capsules as bounding volumes. A capsule is defined as an elliptical or circular cylinder capped with ellipsoids or spheres respectively. The humanoid body is modeled using elliptical capsules, while the moving segments, i.e. arms and legs, of the humanoid are modeled using circular capsules. This collision detection model provides a good fit to the humanoid body shape while being simple to implement. A case study of the self collision free workspace of the humanoid arm is then presented to illustrate the effectiveness of the collision detection scheme.