Real-Time Gait Analysis Sync

Python · C · STM32 · Bluetooth (BLE) · Embedded Systems · Signal Processing · Data Acquisition

Overview

This EPFL semester project focused on creating a robust system to synchronize data from four distinct acquisition technologies (custom STM32-based pressure insoles, PEDAR, motion capture, EMGs) for research on smart footwear aimed at diabetic foot ulcer prevention. Accurate temporal alignment of this data is critical for comprehensive biomechanical analysis.

My Role

Designed and implemented the end-to-end synchronization methodology.
Improved the sampling consistency of custom wearable insoles to 50Hz using interrupt-driven sampling on STM32 microcontrollers.
Developed Python scripts and embedded C code for efficient data acquisition, utilizing bit-packing to reduce Bluetooth transmission load.
Engineered a hybrid synchronization solution using hardware triggers and software-based latency compensation achieving <1ms sync accuracy between insoles.

Challenges

Managing variable Bluetooth latency and interference.
Integrating diverse systems with different sampling rates and communication protocols.

Outcomes

Delivered a system enabling single-command triggering for all sensors, streamlining experimental workflows.
Successfully aligned data streams from wearable insoles and clinical standard systems during walking trials, crucial for comparative analysis.
Identified and quantified sources of residual timing offsets (0.03-0.1s) for future refinement.

Figures

Fig. 1: System overview diagram showing the integration of multiple sensor systems and their synchronization

Fig. 2: Hardware implementation of the synchronized data acquisition system

Fig. 3: Visualization of synchronized data from wearable insoles and PEDAR system during walking trials