In this paper, we present a new computational framework for biodynamic modeling of human movement. The framework decouples the conventional dynamic modeling process into two stages: in the first stage, two-component agonistantagonist torque actuators under hypothesized and testable parametric control drive the forward dynamics, and parameters are identified by tracking both kinematics and kinetics; the second stage completes the mapping from the muscletendon forces to the predicted joint torques. An empirical test using multifinger grasping movement data was conducted to illustrate the application of the proposed framework and showed that the model reproduced the measurement accurately in both kinematics and kinetics. The torque components exhibited consistent spatialtemporal patterns across joints, digits, and subjects. The muscletendon forces computed based on the model-predicted kinematics and kinetics had the peak values within the same order of magnitude as in vivo data reported in the literature. The potential to predict was also demonstrated as we applied the control parameters of one subject to another and achieved close matches.
All Science Journal Classification (ASJC) codes
- Biomedical Engineering
- Computational tractability
- Finger model
- Forward dynamics
- Inverse dynamics
- Musculoskeletal dynamic modeling