Robot motion control for assistance tasks

Abstract

Service robots serve and assist human beings sharing a common environment. Therefore, fast responding and robust planners generating collision-free motions are required to guarantee a safe and convenient man-machine interaction. In most cases the working environment is partially known or perceived by a sensor system. To reduce planning time, this work presents a motion planner interacting with reactive plan execution systems. By dividing the work space into subspaces appropriate motion plans are determined. Based on an environment model, collisions are avoided by interacting with an obstacle avoidance system. Tactile sensors are used to detect collisions of the robot arm, e.g. due to an incomplete knowledge of the environment or enforced by a human touching the robot. Experimental results of our 8 degree of freedom manipulator arm mounted on a mobile platform are presented. In these experiments the robot acts as a barkeeper.