Human-Centered Geodesics for Motion Planning.

Area of Science:

• Robotics
• Biomechanics
• Rehabilitation Engineering

Background:

• Designing human-like movement trajectories for exoskeleton-aided rehabilitation is challenging due to complex human biomechanics and redundant kinematic chains.
• Current trajectory planning algorithms often fail to replicate natural human movements or perform well across diverse motions.

Purpose of the Study:

• To develop a novel geodesics-based computational method for generating human-like reference trajectories in exoskeleton rehabilitation.
• To address kinematic redundancy and incorporate clinical requirements like joint constraints, energy efficiency, and user comfort.

Main Methods:

• A geodesics-based computational approach was developed, integrating joint-level constraints, energy minimization, and comfort criteria.
• The method was validated using upper-limb movement data collected from an exoskeleton platform.
• Performance was compared against standard methods, including minimum-jerk and cubic polynomial planning.

Main Results:

• The proposed geodesics-based method significantly outperformed standard approaches in generating human-like trajectories.
• Trajectories closely matched human demonstrations at both the joint (configuration) and hand (task-space) levels.
• The method demonstrated versatility across different motion types and subjects.

Conclusions:

• The advanced geodesics-based method effectively emulates human movements for exoskeleton rehabilitation.
• This approach offers significant performance gains and shows strong potential for clinical and assisted living settings.
• The method's ability to generate human-like trajectories addresses key limitations in current exoskeleton control strategies.