Car Route Recommendation Based on Machine Learning Intelligent Traffic Light Status Prediction

Road network models are often very complex, containing a large number of intersections, road segments, and different traffic control information. These factors create difficulty for the route recommendation algorithm to estimate the traffic cost at each intersection accurately. This usually causes a reduction in both the time of the recommendation and its accuracy. For this reason, a car traffic route recommendation method using a machine learning (ML) intelligent prediction for traffic light status is proposed. Firstly, use the iDS‐TCV900 intelligent traffic‐specific camera to obtain projected images of traffic lights. The detection area is segmented in the HSV color space based on its color characteristics. Moreover, input the segmented images into a BP neural network to achieve traffic light state prediction. Finally, based on the results predicted as the main driving factor, the SCR algorithm is used for the recommendation of automobile traffic routes, network topological mapping of the road network, calculation of the traffic cost on each intersection, and finding the best route between the current location of the vehicle and the destination through flood control measures and comparisons of weights. The research work presented in the paper introduces one kind of successful technical solution for intelligent transportation systems. The empirical results suggest that the accuracy level of approximately 97% for traffic light detection is achieved in the research with a recommended route time of 5–9 s, thus making it practicable in reality.