Automatic trajectory generation for multi-joint robot system

This paper presents a new approach to an automatic trajectory generation in multi-joint robot systems. The developed method consists of two parts. During the off-line part a 3D-trajectory-net inside the robot's workspace is generated under consideration of the fixed obstacles of the environment. By testing the robot's motion along this net a set of trajectory segments defining safe TCP locations is computed. The second part, the on-line part, derives a trajectory-graph from these off-line generated segments. Considering the current position of the robot in its environment and the state of possible nonfixed obstacles a sequence of trajectory segments is computed to realize a desired task. By using this method for the generation of necessary trajectories the time for the teach-in programming is drastically reduced.