This week, we focused on both hardware stabilization to enhance sensor reliability and prepare for machine learning. We hope to begin testing on the bike soon.
Hardware Improvements:
- Hall Effect Sensors (KY-003 on GPIO 15 & 16):
- Supply voltage was increased from 3.3V to 5V to address signal instability and eliminate phantom speed readings. This change significantly improved the reliability of wheel speed measurements.
- Startup Stability Enhancement:
- A capacitor was added across the main power input to stabilize voltage during startup, especially under high current draw from the radio module. This has improved boot consistency and prevented ESP32 resets when running on battery power.
Software Updates:
- FMOD Integration:
- The app now supports a new FMOD project file with three interactive parameters—wheel speed, pitch, and jump/drop events—enabling real-time audio adaptation based on live sensor input.
- IMU Data Recording for ML:
- Integrated recording of IMU angle data within the app to support future machine learning efforts.
- RTOS Task Optimization:
- Increased the frequency of relevant RTOS tasks to improve sensor data transfer accuracy, specifically in preparation for ML-based music modulation.
Next Steps:
- Testing on bike to achieve sensor-driven music:
- Implement live synchronization between IMU/wheel speed data and FMOD parameters for dynamic, sensor-driven music modulation by testing on bike.
- ML experimenting
- Train a ML model to classify events on the bike like jumps and drops.
- FMOD
- Potentially expand to allow user-uploaded audio and assign them to custom tracks to ehance the apaptive audio experience.
Music-Bike 
Leave a comment