DESIGN AND DEVELOPMENT OF A UNIVERSAL ACCIDENT SEVERITY DETECTION AND SMART EMERGENCY RESPONSE SYSTEM FOR VEHICLES
K.G.H. Priyanath1* and L.D.R.P. Liyanage2
1Advanced Technological Institute, Colombo 15, Sri Lanka, 2Institute of Technology, University of Moratuwa, Sri Lanka
Session: Technical Session C
Abstract
This project presents the development of a universal, independent accident severity detection and emergency response system for vehicles, aiming at addressing critical delays in post-accident interventions. The system integrates multi-sensor data and artificial intelligence to detect vehicular accidents, assesses the condition of passengers, and initiates real-time communication with emergency services such as hospitals, fire brigades, and rescue teams. It uses a Raspberry Pi micro-controller to communicate with sensors, modules and cameras. The research involved analysing existing accident detection technologies, designing AI-based algorithms for passenger monitoring, and integrating these into a unified detection unit. The system was tested and validated under different lighting conditions and angles to assess its accuracy and response efficiency aiming to enhance passenger safety and emergency response times. The accuracy of camera coverage, passenger count and passenger consciousness was calculated using processing images and videos under controlled conditions. In 100 tests conducted for each factor, the percentage findings were: 78% accuracy for camera coverage, 88% for passenger count detection, 72% or consciousness detection. A buzzer and LED provided a brief manual response window to cancel false alerts. Manual trigger buttons were also made available for passengers to notify services directly informing them with the location in the event of an emergence or sensor failure. The system usually responds within 5-15 seconds depending on mobile network strength. While testing and developing the system, challenges such as camera installation, low light conditions, network issues had to be addressed. Finally, the outcome was a privacy protected modular system which is compatible with any vehicle model and can be easily installed providing a scalable and reliable solution to enhance road safety, reduce response time, and improve emergency decision making.
Keywords: condition of passengers, emergency decision-making, independent accident severity detection
DOI: 10.64752/VBAA9687