Real-time inverse kinematics for the upper limb: a model-based algorithm using segment orientations

Area of Science:

• Biomechanics
• Human Movement Analysis
• Robotics

Background:

• Model-based analysis of human upper limb movements is crucial for understanding motor control.
• Traditional inverse kinematics calculations are computationally intensive and typically performed offline.
• Inertial motion sensing offers a mobile, cost-effective alternative to optical systems for lab-free measurements.

Purpose of the Study:

• To develop an algorithm for accurate, real-time estimation of anatomical joint angles.
• To integrate inertial sensor data with the OpenSim upper limb model for movement analysis.
• To extend the workflow for lab-free measurement and analysis of human arm movements.

Main Methods:

• Analyzed the internal structure of an OpenSim upper limb model to develop a custom marker set.
• Formulated a reconstruction algorithm utilizing orientation data from inertial measurement systems.
• Validated the algorithm's performance across various platforms, including embedded systems.

Main Results:

• Achieved significant execution performance improvements on all tested platforms.
• Demonstrated a 50-15,000x speedup compared to OpenSim's Inverse Kinematics tool.
• Maintained numerical accuracy while drastically reducing computation time.

Conclusions:

• The algorithm enables real-time reconstruction of standardized anatomical joint angles.
• Facilitates complex applications requiring fast, accurate model-based inverse kinematics.
• Establishes a new approach for biomechanical analysis using inertial sensors in diverse environments.