FormWings
Completed

FormWings

On-device running form analysis powered by IMU and machine learning

An on-device running form analyser built on IMU sensing and machine learning. A Nano 33 BLE (LSM9DS1) streams raw accelerometer and gyroscope data over UART; the UNO Q Linux side runs an XGBoost classifier (with a fp16 TFLite Transformer as fallback) to assess running form per window. Metrics tracked: cadence (spm), vertical oscillation (cm), ground contact time (ms), trunk forward lean (deg), heel-strike likelihood, and impact loading rate (BW/s). Modulino Thermo adds environment context — temperature, humidity, heat index, and heat-risk scoring. Predictions are broadcast over BLE using a compact MsgPack JSON format and visualised on a React + Vite web dashboard with live posture arc and metric strip.

Poster

FormWings poster

System architecture

FormWings architecture diagram

Built by

Built with

Arduino Nano 33 BLE UNO Q Modulino LSM9DS1 IMU BLE XGBoost TFLite Edge AI Python React Vite MsgPack IoT