Human Grasping Behavior Analysis in Human-Robot Collaboration Based on Markerless Motion Capture

Abstract

In recent years, human-robot collaboration (HRC) has attracted significant attention. For robots to work efficiently and safely with humans in a shared workspace, it is necessary to collect the operator’s motion data and analyze the behavior of operators interacting with robots. Marker-based motion capture requires expensive equipment. Emerging markerless kinematics motion capture offers a solution to this problem. With correctly calibrated scenes and cameras, high-precision 3D human motion data can be obtained after triangulation processing. However, the skeletal point data collected by markerless kinematics should be transformed to the operator’s joint angle to support further human behavior analysis. This study utilizes OpenSim to construct a personalized skeletal model for the operator based on the collected skeletal point data and converted the skeletal point data into joint angle data, based on which the operator’s behavior, such as the use of the dominant hand in grasping actions, is studied.