Development and verification of human-robot collision simulation

Abstract

In recent years, researchers and industry strive to achieve close, physical collaboration between robots and humans in order to make workcells more compact and flexible. In this context, many robots and con- trol schemes have been developed which allow an intuitive, "hands-on" human-robot interaction (HRI). When robots and humans share the same workspace, collisions can generally not be avoided under all circumstances or contact might even be part of the task. Therefore, one must take appropriate coun- termeasures in control and/or mechanical design of the robot to guarantee safe HRI. Safety in robotics is ongoing topic in research, large efforts are currently taken to understand the mechanisms of human injury for different robotic contact scenarios. This thesis aims at developing a better understanding of human head injury during dynamic collisions. The general impact behavior and tolerances of facial and cranial bones are derived from both robotic and biomechanical experiments provided in literature. A simple collision model is selected and parametrized for simulating impacts to the human head and verified with real experimental data. Finally, simulated and experimental injury data is represented such that robot control schemes can easily access this data for ensuring human safety.