Blunt thoracic injury: Unifying experimental knowledge from biomechanics and robotics for estimating human injury probability

Abstract

Physical human-robot interactions are more ubiquitous in the past few years than ever before. This is due to the development of more sophisticated light mechanical structures and advanced control systems. These direct interactions eliminate classical safety barriers between the robot and the operator. Nevertheless, the safety of humans working closely with robots should be guaranteed despite the absence of such barriers. Recommended speed motion ranges can be very limiting and are not knowledgebased. As a consequence, applications that require high robot mobility and performance are restricted by the imposed motion constraints in order not to inflict injury while in direct contact with a human. The approach in this thesis is to gain knowledge from biomechanics on the injury mechanisms for thoracic and abdominal blunt impacts and estimate human injury probability by means of the so called severity indices. By doing so, one can parametrize injury probability in terms of the robot characteristics such as reflected mass and speed of motion. Therefore, this parametrization allows the robot to a) move in an agile and non-restrained manner while b) guaranteeing that in the case of an impact, the injury probability does not exceed certain thresholds. The implementation of such parametrization, in impact simulators for the thorax and abdomen, allows for rapid assessment of generic robots and estimation of injury probability under various impact conditions. The simulation results are presented in the form of mass-velocity to injury mappings that can be easily used in control schemes.