Inspired by a tragic accident in Barrhaven where a young scooter rider was struck by a truck, our team of four from uOttawa and Carleton built an aftermarket safety co-pilot that retrofits onto any youth e-bike to protect commuters where city infrastructure falls short. To handle life-critical functions without cloud latency or standard OS jitter, we deployed QNX 8 on a Raspberry Pi to act as our core brain. We locally compiled OpenCV and TensorFlow Lite to run an edge-AI vision pipeline that calculates real-time Time-to-Collision metrics; if a collision vector is detected, the hard real-time microkernel instantly preempts tasks to trigger an Automated Emergency Braking system.
The system supports Tensorflow Lite compiled for QNX, OpenCV, and we use a BLE chip with a reverse-engineering of Apple's Find My system to locate your chip via GPS even with no Internet connection.
Watch our video on the 'Try it out' link :)
Log in or sign up for Devpost to join the conversation.