3 Collision Avoidance for a 7-DOF Redundant Manipulator

Abstract

Collision detection and obstacle avoidance are two features that play an important role in fully or partly autonomous operations of robotic manipulators in cluttered environments. A compact and fast collision-avoidance scheme would be particularly useful for robotic applications in space, underwater, and hazardous environments. Collision avoidance for robot manipulators can be divided into two categories: end-effector level and link level. Much of the work reported to date has dealt with obstacle avoidance as an off-line path planning problem, i.e., find a collision-free path for the end-effector [7], [28], or by mapping the obstacle into joint space, find a collision-free path in joint space [36], [11]. These methods are not applicable to environments with moving objects. Moreover, for non-redundant manipulators, tracking an end-effector trajectory while avoiding collisions with obstacles at the link level, or self-collision avoidance, is often not achievable. Kinematic redundancy has been recognized as a major characteristic for operation of a robot in a cluttered environment [33]. For redundant manipulators, a real-time collision avoidance approach has been developed recently by Seraji and Bon [70] that formulates the problem as a force-control problem so that the task of collision avoidance is solved primarily by augmenting the manipulator control strategy.