Fuzzy Logic-based Risk Estimation for Safe Collaborative Robots

Abstract

The increasing trend of using industrial robots due to their robotic strength and endurance made the manufacturing processes more efficient. Humans, however, are still an integral part in this industry due to processes that robots could not automate, therefore, the need for human-robot collaboration (HRC) arises. The matter of safety arises during HRC since industrial robots are typically enclosed around a work fence. The current effective implementations for HRC consider humans as a foreign obstacle in the workspace and adapts based on its distance and velocity with respect to the robot manipulator. These parameters, however, do not fully translate the risks associated with each situation. To overcome this, a risk estimate must first be produced. This paper introduces a Fuzzy Logic-based risk estimation which takes inputs from the human such as head orientation and upper body orientation together with the pairwise distance of the human and robot, the speed of the robot manipulator, and speed of the human. This was implemented using a Kinect V2 depth sensor, and a KUKA KR6 R700 6-DoF robot manipulator.