An approach to integrate human motion prediction into local obstacle avoidance in close human-robot collaboration

Within Human-Robot Collaboration (HRC) safety is one key-issue that has to be guaranteed at any time during joint collaboration. Collisions in a shared workspace of a Human-Robot-Team (HRT) must be prevented. In addition, the comfort of the collaboration behavior should be provided. Facing these challenges, a robot has to be able to detect critical states at an early stage on the one hand and should react to them within a very short time span on the other hand. In this paper a collision avoidance algorithm using compliance control that guarantees a fast reaction to dynamic obstacles, e.g. humans, without the need of high computational effort is outlined. To further improve the avoidance behavior of the robot, a human motion prediction algorithm based on the minimum-jerk model is integrated. In an experimental analysis of a case-study about collecting LEGO-bricks on a table with various subjects, the impact of the integration of human motion prediction on both the robot's reaction time and human's perception of the robot co-worker is studied. Finally, the comfort and acceptance of the robot colleague by the human collaborator is drawn out through an analysis of the subjective human feedback questionnaires.