A Real-Time Identification and Tracking Method for the Musculoskeletal Model of Human Arm

This paper aims at the development of a unified method for online identification and tracking of a kinematic musculoskeletal model of human arm to pave the way for related realtime applications, such as human-robot interaction, teleoperation and biomedical analysis. In order to decouple the identification of the joint angles of human arm kinematic model from a variety of motion capture (MoCap) setups, a generalized human arm triangle, which can be easily calculated by raw motion data, is introduced as an intermediate unified expression interface of human arm posture. An analytical solution to the Inverse Kinematics (IK) problem from the proposed human arm triangle to the joint angles of a commonly used OpenSim human right arm model is derived in detail. Once the human arm kinematic model is reconstructed, the involved muscles can be located correspondingly for related applications. Comparative simulation and experiment are conducted to validate the performance of the proposed IK and the whole tracking method. The results manifest that the calculation efficiency of the proposed IK can achieve an enormous speedup of 400-600 times with respect to the OpenSim built-in IK while maintaining comparable accuracy. Therefore, the proposed method can be an important tool to enable many online applications using human arm musculoskeletal model.