Body sensor network-based strapdown orientation estimation: application to human locomotion.

Area of Science:

• Robotics and Biomechanics
• Sensor Fusion and Estimation Theory

Background:

• Accurate real-time orientation tracking of human limbs and robotic segments is crucial for applications like gait analysis and advanced prosthetics.
• Existing methods often face challenges with computational complexity and bias sensitivity in sensor data.

Purpose of the Study:

• To develop and evaluate novel orientation estimation filters for bias-insensitive, real-time strapdown tracking of human limb and robotic segment orientation.
• To reduce the computational cost and complexity of orientation estimation algorithms for body sensor networks.

Main Methods:

• Utilized triaxial gyrometer, accelerometer, and magnetometer data from a body sensor network.
• Implemented quaternion representation for rotation, employing the Quaternion Based Estimation Algorithm (QUEST) for preprocessing sensor data.
• Designed and tested an Extended Kalman Filter (EKF) integrated with QUEST (EKF-QUEST) and a novel complementary sliding mode observer.

Main Results:

• QUEST-based preprocessing significantly reduced complexity and calculation costs for the body sensor network.
• Both EKF-QUEST and the complementary sliding mode observer demonstrated effective bias-insensitive orientation tracking in simulations.
• Validation against a reference motion tracking system on a treadmill confirmed the filters' performance.

Conclusions:

• The developed EKF-QUEST and complementary sliding mode observer offer efficient and accurate solutions for real-time orientation tracking.
• These methods are suitable for bias-insensitive gait analysis and robotic applications utilizing body sensor networks.
• The QUEST preprocessing step is key to reducing computational load without sacrificing accuracy.