fastSW: Efficient Piecewise Linear Approximation of Quaternion-Based Orientation Sensor Signals for Motion Capturing...

Area of Science:

• Wearable Technology
• Sensor Data Compression
• Human Motion Analysis

Background:

• Inertial measurement sensors in wearables are crucial for applications like motion capturing.
• High data transmission rates for motion tracking deplete battery life and limit device use.
• Efficient data reduction is needed to overcome communication limitations in wearable sensor networks.

Purpose of the Study:

• To introduce fastSW, a novel piecewise linear approximation technique for efficient data reduction in wearable inertial measurement sensors.
• To reduce the amount of data transmitted from wearable devices without compromising orientation estimation accuracy.
• To enhance the usability of wearable devices by minimizing battery consumption associated with wireless communication.

Main Methods:

• Developed fastSW, a piecewise linear approximation algorithm tailored for wearable sensor data.
• Implemented a data transmission strategy that prioritizes devices actively moving or exceeding an approximation error threshold.
• Optimized the algorithm for embedded platforms, achieving low computational complexity (e.g., 210 instructions on ARM Cortex-M4).

Main Results:

• fastSW compresses data to 10% of its original size in experiments on a public dataset.
• The technique maintains orientation estimates without deviating from unit quaternions.
• Achieved an average angular deviation of approximately 2° and a maximum deviation below 9°.

Conclusions:

• fastSW offers an efficient method for compressing inertial sensor data from wearable devices.
• The technique effectively reduces data transmission, leading to lower battery consumption and extended device usage.
• fastSW provides a viable solution for enhancing the practicality of motion capturing and other sensor-based applications in wearables.