FitAI Band

This project develops a compact wearable circuit board that recognizes strength-training exercises automatically. The device mounts to the wrist and combines a multi-axis motion sensor, gyroscope and magnetometer to capture fine-grained motion data. An ESP32 microcontroller aggregates the sensor streams and transmits them to a PC or smartphone, where an AI model analyzes the data and classifies the exercise (for example: biceps curls, bench presses, squats).

The goal of FitAI Band is to let users track workouts without manual logging. Early prototypes already demonstrate reliable movement recognition; however, model accuracy will improve with more labeled training data from diverse users and exercise variations. This project is open source and intended to support the fitness community by providing an easily extendable platform for automated workout tracking.

The FitAI Band requires a connected battery for stable operation. The ESP32 cannot be reliably powered by USB-C alone on this design because the charging/power module is configured to draw primary power from the battery. For example, a single-cell Li-ion battery (≈3.6 V nominal) can be used. When the battery is connected the board’s blue LED illuminates to indicate the power supply is active. USB-C still serves a useful role: it allows direct code uploads to the ESP32 and can be used for charging the battery (depending on the specific charging module).

Key features & technical notes

Wrist-mounted board with accelerometer, gyroscope, magnetometer.
ESP32 central controller for data collection and wireless transmission.
Requires an attached Li-ion battery (≈3.6 V nominal) for proper operation — blue LED indicates battery power.
USB-C is available for direct firmware uploads and may be used to charge the battery.
Prototype proven in tests; more user data improves the AI model.
Open-source hardware and firmware to encourage community contributions and reproducibility.