Road Condition Risk Prediction Model of Self-Driving Vehicle based on Machine Vision

There are some problems to be solved in self-driving vehicles, such as complex traffic scene processing, dynamic obstacle identification, real-time prediction and so on. In order to improve the accuracy of risk prediction, a road condition risk prediction model based on machine vision is proposed. In order to solve this problem, the vehicle motion data and potential collision risk are effectively analyzed by using machine vision technology, SimAM attention mechanism, StrongSORT algorithm, SS-LSTM algorithm and so on. The machine vision algorithm and risk prediction method in this study achieve image quality improvement, accurate vehicle behavior monitoring and risk early warning. The simulation results show that the model mAP is improved to 56.17% and the risk prediction accuracy is improved by 34%. The results show that the technical method is feasible.