Collision avoidance using asynchronous teams

Abstract

Robotic manufacturing tasks frequently require off-line planning to find feasible manipulator trajectories. This is difficult for redundant manipulators operating in complex working environments. In this paper a collision avoidance algorithm based on asynchronous teams, which consist of cooperating autonomous software operators (agents), is presented. Both heuristic-based and gradient-based agents are introduced to reduce joint motions, smooth joint trajectories, and to avoid joint limits or collisions. An application example is shown through simulation of a 6-DOF PUMA robot performing a welding task.