Detection of Human-Robot Collision using Kinetic

Abstract

In this work, a system capable of detecting human-robot potential collisions is presented. It is intended to assist a robotic arm on its movement around the working area. By detecting risk zones, it prevents the robot from interfering with the surgeons work. To accomplish this, it is necessary to place both the robot and the surgeon on a coordinate system. The tracking of the robot is achieved by detecting color labels that were placed on the robot's joints. On the other hand, the physician's position is obtained using Kinect's depth stream. The output of the system will be used by the robot to define risk zones and to avoid them by modifying its trajectory.