VISION-BASED SAFETY SYSTEM FOR INDUSTRIAL COLLABORATIVE ROBOT

This paper proposes a new method to improve the safety operation of collaborative robots (cobots) by combining two existing safety features: distance-based speed control and obstacle avoidance. The system utilizes a Kinect V2 camera to capture depth data of the robot's workspace. This data is used to determine the distance between humans and the robot, allowing for speed regulation to ensure safety. Additionally, the depth data is converted into a map to enable obstacle avoidance by the robot. The proposed method was evaluated in a simulated environment using various pick-and-place scenarios. The results demonstrate the effectiveness of the system in ensuring safe human-robot collaboration. The paper also identifies limitations, such as the limited depth perception due to a single camera. Future work will explore using multiple cameras and implementing the system on a physical cobot.