AI-Powered Smart Bike Light: Enhancing Cyclist Safety Through Real-Time Hazard Detection and Alerts

Abstract

Cyclist safety is a growing concern as accidents due to limited visibility and inadequate real-time hazard detection persist. This study introduces an AI-powered smart bike light that enhances hazard detection and improves road safety for cyclists with the help of increased visibility to drivers. Built using a Raspberry Pi, TensorFlow for machine learning, and a Pi Camera, the system detects hazards—such as vehicles and pedestrians—up to 220 feet away. Alerts are provided to the cyclist with the help of a RGB strip and buzzer on the handlebar and surrounding drivers via a multi-color RGB matrix and audible buzzers. Field tests were conducted with 3 different devices: using the smart bike light, a regular bike light, and no light. When comparing the distance at which a car stopped behind the bicyclist the smart light increased vehicle stopping distances from 4.3 to 7.5 feet and reduced passing speeds from 37 mph to 28 mph. These results demonstrate the system’s potential to significantly enhance cyclist awareness and safety, reducing accidents. Future improvements should include performance testing in adverse weather conditions and low lighting to assess the robustness and reliability of the system.