REACTIVE MOTION PLANNING - A MULTIAGENT APPROACH

We present an efficient approach to reactive robot motion planning and collision avoidance. Unlike the traditional methods, there is no cent ralized control; instead the links and the joints of the robot are aut onomous agents. This is a completely new approach. A set of dynamic eq uations of motion for an arbitrary robot is derived Artificial forces are introduced to express and combine multiple, possibly conflicting o bjectives, such as avoiding obstacles while approaching a goal. The jo int agents impose forces of constraint between the link agents, and th ese forces establish a flow of information among the agents. The emerg ent behavior of the multiagent system gives an impression of surprisin gly intelligent overall control. The developed method is used in actua l industrial applications to control welding robot installations for s hip building with up to II degrees of freedom (DOF). Experimental resu lts from the simulation of a 25-DOF snakelike robot operating in a com plex three-dimensional structure are given. It is demonstrated that th e time complexity is O(n(3)) for branched n-DOF robots, while for seri al robots such as standard manipulators and the snake, it is O(n).