A Nonlinear Dynamics-Based Estimator for Functional Electrical Stimulation: Preliminary Results From Lower-Leg...

Area of Science:

• Biomechanics
• Sensor Technology
• Signal Processing

Background:

• Miniature inertial measurement units (IMUs) are susceptible to drift, noise, and magnetic interference, limiting accurate joint angle measurement.
• Existing methods like the extended Kalman filter (EKF) and rotation matrix methods have limitations in accuracy and sensor requirements.

Purpose of the Study:

• To introduce and validate a novel nonlinear state estimation technique, state-dependent coefficient (SDC) estimation, for improved joint angle prediction from IMU data.
• To compare the SDC estimator's performance against established methods like EKF and rotation matrix methods.

Main Methods:

• Developed a state-dependent coefficient (SDC) estimation technique utilizing nonlinear limb dynamics without Jacobian linearization.
• Formulated a nonlinear knee musculoskeletal model and identified it through experimental procedures.
• Experimentally validated the SDC estimator by measuring knee joint angles during functional electrical stimulation, comparing it with EKF and rotation matrix methods using a rotary encoder for ground truth.

Main Results:

• The SDC estimator achieved a root mean square error (RMSE) of 2.70°, outperforming the EKF (4.42° RMSE) and slightly outperforming the rotation matrix method (2.86° RMSE).
• The SDC method demonstrated the ability to measure knee angles using a single IMU, unlike the rotation matrix method requiring two IMUs.

Conclusions:

• The SDC estimation technique offers a promising advancement for accurate joint angle measurement from IMUs, particularly in dynamic movements.
• Leveraging limb dynamics provides superior performance and sensor efficiency compared to kinematic models or standard EKF approaches.