RoadAware: Real-Time Hazard Detection and Driver Awareness System for Mobile Devices
TL;DR
An enhanced road safety system that combines motion sensor-based detection with computer vision algorithms to create a more comprehensive hazard alert system and demonstrates significant improvements in detection accuracy.
Abstract
: Road safety remains a critical global concern, with road hazards and sudden braking incidents contributing significantly to accidents. This research introduces an enhanced road safety system that combines motion sensor-based detection with computer vision algorithms to create a more comprehensive hazard alert system. The system employs smartphone sensors including accelerometers, gyroscopes, GPS, and the device camera to detect road hazards, provide lane departure warnings, and alert drivers to potential collision risks. Our application, 'RoadAware' aims to reduce accidents by offering real-time hazard alerts, mapping dangerous road conditions, and providing visual driving assistance through multiple modes including dash-mount computer vision and heads-up display reflection. This paper details the implementation of computer vision models for lane detection and forward collision warning, performance optimization techniques across various devices, and integration with existing motion-based hazard detection. Testing results demonstrate significant improvements in detection accuracy, with pothole identification reaching 93% accuracy and false positive rates for sudden braking detection reduced to 0.5%. The multi-modal approach addresses various driving contexts and environmental conditions, enhancing the system's versatility and effectiveness.