This repository contains the code for our project: Aggressive Driving Detection and Alert System Leveraging Current Signal Light Infrastructure by Emma Li and Brian Li.
According to the National Highway Traffic Safety Administration, aggressive driving behaviors are responsible for 54 percent of annual traffic accidents from 2007 to 2019. These accidents cost our society billions of dollars economically. Additionally, hundreds of thousands of innocent lives are taken away and a profound psychological impact is placed on survivors. Existing research on aggressive driving detection mostly relies on devices mounted on vehicles for collecting the data when driving. Such approaches, however, will be hard to enforce in real life because of consumer privacy concerns. Also, not all existing vehicles can have the monitoring system, thus the overall effectiveness is limited. The approach that we propose here is to leverage the existing infrastructure of traffic lights to enforce the detection of aggressive driving behaviors for all vehicles at all times. The undertaken approach introduces centralized detection and alert systems that can work with an existing traffic light control system and a real digital map system as a real-world application. The detection system utilizes both traffic lights and speed sensors mounted alongside to detect aggressive driving behaviors at intersections. The alert system adopts an efficient specialized graph traversal algorithm to pinpoint impacted traffic lights and alert only the drivers nearby those lights. To demonstrate the approach, we prototyped both a detection system and an alert system running on an Arduino board. We verified their consistent high accuracy and real-time response by applying them to an experimental setup that consists of streets and intersections with traffic lights and infrared sensors.