Inverse optimal control with time-varying objectives: application to human jumping movement analysis

Area of Science:

• Biomechanics
• Robotics
• Motor Control

Background:

• Complex human movements require intricate coordination between the central nervous system and the musculoskeletal system.
• Understanding these movements is crucial for rehabilitation, sports science, and human-robot interaction.
• Existing methods often assume constant control objectives, which may not reflect real-world scenarios.

Purpose of the Study:

• To propose a novel inverse optimal control approach for analyzing complex human movements.
• To develop a method that accommodates time-varying control objectives during movement.
• To extract and understand the underlying control strategies influencing movement performance and learning.

Main Methods:

• An inverse optimal control framework is presented, assuming movement trajectories are optimal with respect to a time-varying cost function.
• The cost function is a sum of weighted basis cost functions, with weights estimated using a sliding window approach.
• A dataset of standing broad jump movements was collected to demonstrate the method's application.

Main Results:

• The method successfully extracts time-varying control objectives, such as center-of-mass takeoff vector and foot placement, crucial for task success.
• Identified control objectives correlate with ensuring participants land on target during the broad jump.
• The approach can differentiate between motion strategies and track changes in control strategy during motor learning.

Conclusions:

• The proposed inverse optimal control method provides a powerful tool for analyzing complex human movements with dynamic objectives.
• It offers insights into motor learning, strategy adaptation, and individual movement styles.
• The publicly available dataset facilitates further research in human movement analysis and robotics.