On-Body BLE + IMU Localization System

Problem

Accurate body segment localization is critical for rehabilitation and sports biomechanics, but commercial motion capture systems are expensive and require controlled environments. We needed a low-cost, portable solution that could achieve comparable accuracy.

Constraints

• Budget limited to $500 for hardware
• System must be wireless and portable
• Latency under 50ms for real-time feedback
• Must validate against gold-standard Xsens system

My Contribution

• 1Designed custom sensor module with ESP32 microcontrollers, BLE modules, and IMUs
• 2Implemented extended Kalman filter for sensor fusion of RSSI-based positioning and IMU orientation
• 3Developed Python data acquisition pipeline with real-time visualization
• 4Created validation protocol comparing against Xsens MTw Awinda system

What I Built

• Designed and 3D printed compact sensor housings for on-body placement
• Programmed ESP32 firmware for synchronized BLE advertising and IMU data streaming
• Developed custom calibration routine for IMU drift compensation
• Developed machine learning model to improve RSSI distance estimation.

Testing & Validation

• Static positioning tests at known distances (0.5m - 1m range)
• Dynamic motion trials with controlled arm movements
• Long-duration drift analysis (30-minute continuous operation)
• Side-by-side comparison with Xsens ground truth

Results

• Achieved position accuracy within 4cm RMS error at distances up to 1m
• System cost reduced by 95% compared to commercial alternatives

Gallery

Lessons Learned

Links