AI-PaaS: Towards the Development of an AI-Powered Accident Alert System

Abstract

The development of an accident detection system is a crucial step towards improving emergency response times, saving lives and achieving the ambitious projection of the United Nations General Assembly to drastically reduce the global fatality rate of road traffic crashes by half by the year 2030. It is also cardinal to the attainment of the United Nation's SDG 11 goal of making cities and human settlements inclusive, safe, resilient and sustainable. In this study we present a preliminary development of an AI-powered Accident Alert System (AI-PaaS). The system has four modules namely, sensors module, detection module, registration module and messaging module. The detection module is powered by sensing technology and the Hidden Markov Model to intelligently and correctly detect that an accidents sound. The MPU 6050 containing both accelerometer and gyroscope is also integrated to detect any sharp variation in the acceleration and angular vis-a-vis a predefined threshold value. Once an accident has been detected, the messaging module is triggered to communicate first responders and the victims' pre-registered kin. Preliminary results are presented. The system can potentially reduce road accident fatality by providing accurate and timely location-based information to emergency service providers.