Efficient manipulator collision avoidance by dynamic programming

This paper describes a research effort in the area of collision avoidance path planning for robotic manipulators. A robotic arm with known geometry is to perform a spatial manipulation in the presence of known obstacles in its workspace. The task is to generate a series of waypoints for its path to pass through which will guarantee a safe, collision-free trajectory from its predefined starting point to its predefined goal position. This is an important topic in the area of automated manufacturing. Automated factories are becoming increasingly important for the goals of greater manufacturing efficiency, better equipment utilization, and lower overall manufacturing costs. Robotic devices, including manipulator arms and assembly devices, are finding more uses in these factories. The approach is to discretize the robot's workspace into a transition network. The optimal path through this network, in terms of angular displacement of the manipulator's joints, is generated by dynamic programming. While this approach has been used previously, this paper adds the innovation of variable-node spacing, with the node density in various parts of the network reflecting the need for precise position control in each local area of the workspace. In this way, precise motion control is possible without an undue computational burden.