ML Workout Wearable Firmware

This repository contains the Arduino firmware for a TinyML-powered workout wearable based on the Arduino Nano 33 BLE.
It reads 9-axis IMU and heart-rate data, classifies the exercise (bench, squat, deadlift) with a TFLite model, counts repetitions with a second TFLite model, and streams live heart rate and workout summaries over BLE to a companion app.

---

📂 Repository Structure

/
├──serial_logger.py # UART Capture Script
├──README.md # (this file)
├──imu_printer/imu_printer.ino # Main Sketch
├──imu_printer/model.h # Header for embedded TFLite models
├──imu_printer/model.c # Embedded TFLite model data

---

🔧 Hardware

• Board: Arduino Nano 33 BLE
• IMU: Bosch BMI270 + BMM150 (9-axis: accel, gyro, magnetometer)
• Heart-Rate Sensor: MAX30102 (via SparkFun MAX30105 library)
• Connections:
  • IMU & MAX30102 → I²C (SDA, SCL), IMU is implicitly connected
  • 3.3 V and GND → Sensors
• Power & Programming: USB

---

📦 Software Dependencies

Install the following libraries via the Arduino Library Manager:

1. Arduino mbed-enabled Boards (to support Ticker)
2. ArduinoBLE
3. Arduino_BMI270_BMM150
4. SparkFun MAX30105 (and its heartRate helper)
5. TensorFlowLite Micro by Chirale (or TensorFlowLite_ESP32 fork that includes all_ops_resolver)
6. Wire (core)

---

⚙️ BLE Services & Characteristics

• Service UUID: 181C (Fitness Machine)
• Characteristics:
  • Heart Rate 2AB4 (Read + Notify) – live heart‐rate buffer
  • Workout Summary 2AC8 (Read + Notify) – JSON string:

    {"exercise":"<bench|squat|deadlift>","reps":<float>}

  • Command 2A3A (Write) – (reserved for future control commands)

---

🚀 Installation & Usage

1. Clone this repo into your Arduino sketchbook folder.
2. Open imu_printer.ino in Arduino IDE.
3. Install all required libraries via Sketch → Include Library → Manage Libraries…
4. Select board Arduino Nano 33 BLE and correct COM port.
5. Upload the sketch to your board.
6. Pair with your smartphone app (implements BLE TFT Fitness Machine central).
   • Live heart rate → updates on the app.
   • When the phone disconnects, the device sends a final workout summary.

---

🔍 How It Works

1. Sensor Setup:

   • BMI270_BMM150 IMU sampled in bursts for motion data.
   • MAX30102 streamed into a small circular buffer for heart rate.

2. TinyML Inference:

   • classification_model
     • Inputs sliding-window IMU data for exercise classification.
   • rep_count_model
     • Inputs processed sensor features to increment a reps counter.

3. BLE Communication:

   • Periodically notify live heart rate.
   • On workout completion or disconnect, notify workout summary JSON.

---

🔄 Extending & Troubleshooting

• Model Updates: Replace model.h/model.c with your own .tflite arrays.
• Buffer Sizes: Tweak CL_BURST_LEN, RC_BURST_LEN in imu_printer.ino to adjust window lengths.
• Battery & Power: Ensure 3.3 V supply to sensors; the Nano 33 BLE’s regulator handles USB power or LiPo input.

---