Carv | AI ski coach

As a personal project combining my passion for skiing and technology, I set out to create my own version of Carv's digital ski coach. The goal was to build a system that could capture and analyze skiing technique data in real-time, similar to the commercial product but designed and built entirely by myself.

The project presented several technical challenges:

I built a custom hardware solution using an Adafruit Feather nRF52840 Express board combined with a BMI270 IMU 6-axis sensor, powered by a LiPo battery. This compact setup could be attached to ski boots to capture detailed motion data.

For data processing, I implemented sensor fusion and a calibration step to stabilize the sensor data and align with the ski axes, following similar principles to the commercial Carv system. The mobile application was developed in Flutter, with both local SQLite storage and Supabase backend integration for when internet connectivity was available.

The sensor system captures data at 400Hz and transmits it via Bluetooth to the mobile app, which processes the information in real-time to provide immediate feedback on skiing technique.

The completed system tracks several key skiing performance metrics in real-time:

The Flutter app provided real-time feedback and stored historical data for later analysis, with automatic synchronization to Supabase when internet connectivity was available. The hardware design proved durable enough to withstand the physical demands of skiing, while the battery life extended to multiple full days of skiing on a single charge.

The technical implementation involved:

Key technical challenges were solved by optimizing the BLE protocol for reliable high-frequency data transmission and implementing efficient algorithms that could process the IMU data in real-time on a mobile device.

Although this was primarily a fun personal project that wrapped up just as the ski season ended (limiting extensive testing on the slopes), it successfully demonstrated the feasibility of creating a DIY version of sophisticated commercial sports technology using accessible components and frameworks.

The project showcased skills in embedded systems programming, sensor data processing, mobile app development, and cloud integration - all valuable experience that translates to more complex IoT and wearable technology applications.