Real-time Collision Detection and Traffic Congestion Control for Vehicular Networks

Living in a world in touch with technology, one is plagued by various ambiguities. These problems can all find a home within computer vision and intelligent communication systems. One such difficulty is the recognition of traffic accidents on roads and mitigating the accumulation of semi-autonomous vehicles at the site via quick vehicle-to-vehicle communication. This paper presents a real-time collision detection and traffic congestion control algorithm for vehicular networks. YOLO $\mathbf{v}$ 8 (You Only Look Once - version - 8), the computer vision technology that plays its strings in capturing and detecting collision between two approaching vehicles works on a precision scale of achieving an accuracy of up to $\mathbf{9 0 \%}$. On the other hand, after the accident has been detected, the Path Optimization Algorithm (POA) assists in neutralizing traffic with a negligible delay at the points of congestion and directs them toward the best possible path through the concepts of graph theory. Via these two techniques upon its belt, the proposed technique deals with the presented problem in an efficient way. The efficacy of the proposed scheme is verified by comparing it with shortest path Dijkstra method and random path allocation algorithm using extensive simulation results.