Enhancing Human-Machine Collaboration Safety Through Personnel Behavior Detection and Separate Speed Monitoring

Human-robot collaboration is rapidly advancing, especially in manufacturing, where robots assist with heavy, hazardous, or repetitive tasks. The goal is to enhance productivity while ensuring a safe and efficient work environment. This study explores safety technologies for detecting personnel behavior and monitoring speed and separation, focusing on identifying workers' motion intentions and controlling robots' safety distance with variable speed regulation. This allows robots to manage acceleration and deceleration safely, improving work efficiency and overall production system performance. Using 3D cameras and human posture detection with 34 key points, this research applies the DLOW model for human motion prediction. The DLOW model accurately predicts human movements within 2.5 seconds with an 81.34% accuracy, and these predictions are fed into to the robotic system, which adjusts its movement speed to avoid collisions. A distance-speed relationship model ensures the robot operates at a safe speed based on the worker's position and velocity, reducing harm and resuming normal speed outside the safe zone.