Using motion capture data to regenerate operator's motions in a simulator in real time

Abstract

During the human-robot cooperation assembly process, detecting an operator's position not only protects the operator, but can also predict the operator's intention according to his position. This paper describes an image-based operator monitoring system based on the direct linear transformation (DLT) method. To overcome the changing light conditions in actual factories, a special color filter method was used to detect color marks attached to the operator's body. However, it is impossible to determine whether the detected data is correct based only on the detected 3D position data of the color marks. Hence, a kinematic human body simulator was built to regenerate human operator's motions in real time. According to the experimental results, the operator monitoring system meets the measurement accuracy requirement. Furthermore, this system can easily be adjusted to satisfy the changing light conditions. Thus, human operator's motions can be conveniently regenerated in real time by importing the detected joint angle data into the kinematic human body simulator. This system provides a solid foundation to protect an operator during the human-robot cooperation assembly process.