Autonomous Warning System To Detect Road Blockers And Collisions While Notifying Users About Potential Risks

Usually, only time and distance are considered when choosing a route, not the risk involved. Due to not taking the danger into account before the trip, we might have to experience numerous problems and risks. As a result, there may be dangerous situations for the driver and passengers and time and fuel wastage. Thus, the issue is how to evaluate the road while choosing the route before using it and What elements are crucial to take into account while evaluating a route’s issues. This project aims to develop a system to evaluate risk before a vehicle enters a dangerous area and predict the best path by evaluating aspects of roads including accidents, road repairs, and road blockers. Image Classification is used to detect road repairs and blockers. Federated Learning will be used to improve the accuracy of the Road blocker detector. Here, Clients will train locally using private data, send their optimized parameters to the central server, and then aggregate those trained parameters to produce a global model. This process will be continued until the system has a trained model that is ideal for predicting roadblocks. YOLOV7 object detector will be used to detect vehicle collisions and accidents in real-time. Finally, Federated Learning will be used to notify users about detected incidents and event details will be marked in a Google map which is running on a Flask application. Using the Blocker map, drivers can decide which route is more suitable based on the proposed system’s detections and blocker details.