Real-Time Identification of Wrist Kinematics via Sparsity-Promoting Extended Kalman Filter Based on Ellipsoidal Joint...

Area of Science:

• Biomechanics
• Robotics
• Control Systems

Background:

• Accurate real-time wrist kinematics identification is crucial for advanced prosthetic and exoskeleton control.
• Existing methods often struggle with noisy data and complex coupled joint movements.

Purpose of the Study:

• To propose a novel method for real-time wrist kinematics identification.
• To enhance the robustness and accuracy of wrist motion modeling.

Main Methods:

• Developed a novel ellipsoidal joint formulation with quaternion-based rotation constraints and 2D Fourier Linear Combiners (FLC).
• Implemented a sparsity-promoting Extended Kalman Filter (EKF) using smooth l1-minimization for robust parameter identification.
• Validated the approach using simulations with multiple reference models and real-world experimental data.

Main Results:

• The proposed method demonstrated robust real-time identification of wrist kinematics in both simulations and experiments.
• The sparsity-promoting EKF effectively improved regression robustness under noisy conditions.
• Accurate approximation of coupled rotations and translational displacements was achieved.

Conclusions:

• The developed regression model combined with the sparsity-promoting EKF provides a reliable solution for real-time wrist kinematics modeling.
• The method shows significant potential for improving the control systems of wearable wrist exoskeletons.
• The framework is adaptable for real-time identification of other joint kinematics in exoskeleton applications.