Estimating net joint torques from kinesiological data using optimal linear system theory

Area of Science:

• Biomechanics
• Human Movement Analysis
• Motor Control

Background:

• Net joint torques (NJT) are crucial for understanding motor control in dynamic biomechanical systems.
• Traditional inverse dynamics methods for computing NJT rely on kinematic and kinetic measurements, often introducing errors through numerical differentiation.
• These errors can lead to unstable forward simulations, hindering studies of muscle coordination.

Purpose of the Study:

• To develop an alternative approach for computing NJT that ensures stable forward dynamic simulations.
• To improve the accuracy of NJT estimation by minimizing reliance on explicit kinematic differentiation.
• To enhance the replication of observed movements and measurements in dynamic biomechanical models.

Main Methods:

• Developed and applied the linear quadratic follower (LQF) algorithm to compute NJT.
• The LQF algorithm utilizes a cost function based on data confidence, avoiding explicit differentiation of kinematic data.
• Tested the LQF approach by estimating NJT in standing humans subjected to support-surface perturbations.

Main Results:

• The LQF algorithm successfully computes NJT that result in stable forward simulations.
• The method effectively replicates observed movements and measurements.
• Demonstrated that insufficient kinematic and force data reduce the confidence in NJT estimates, regardless of the method used.

Conclusions:

• The LQF algorithm offers a robust alternative to inverse dynamics for estimating NJT in biomechanical analyses.
• Accurate NJT estimation and stable simulations are achievable with the LQF approach.
• Emphasized the critical need for sufficient kinematic and force data for reliable NJT estimation in any method.