Skate Sensor

The Project

Figure skating is a beautiful but highly demanding sport, especially for beginners. More skaters are entering the sport, but many of them quickly discover that figure skating requires constant practice, access to coaching, and the ability to repeat skills over and over again with proper technique. Beginner skaters often do not have enough affordable and accessible support while training. Figure skating is expensive because families must pay for private lessons, ice time, equipment, competition fees, and off ice training. This makes it difficult for many skaters to receive frequent, personalized feedback. Skate Sensor helps with this problem. Skate Sensor is a mobile app with a hardware companion device designed to help skaters better analyze and improve their techniques on the rink. Using a combination of accelerometer sensors connected to ESP32 boards, the user’s movements can be transmitted via Bluetooth to the app for analysis. The sensors themselves are to be strapped to the user's shins, facing outward to allow for accurate recordings with minimal interference. The app itself is designed to have a combination of algorithmic and machine learning models to help determine the success of a technique being performed. Skate Sensor provides users with quick, personalized feedback on specific techniques. A major challenge was keeping the left and right skate sensors accurately synchronized. If the two devices send data at slightly different times, the system could misinterpret the timing between leg movements, which is critical in figure skating. Delays or unequal sampling rates reduce the accuracy of technique detection. To address this, I designed the system to timestamp every reading as it arrives and align both streams onto a shared timeline. I also averaged repeated readings within short intervals and fill small gaps with nearby values so each model input remains complete and consistent.

View this project on YouTube