Conflict Avoidance Decision-Making for Unmanned Vehicles at Intersections under the Mixed Traffic

Due to the high uncertainty of human driving behaviors, unmanned vehicles only slow down or stop to avoid conflict leading to low traffic efficiency. Aiming to solve the interaction problem, this paper integrates an intention prediction algorithm into the decision-making and an optimization model to maximize the expected benefit. Firstly, the collision at intersections is judged by PET parameter. Secondly, when the unmanned vehicle faces conflicts with other vehicles at intersections, the dynamic decision-making mechanism is established based on the support vector machine and then selects an action. Finally, the validity of conflict avoidance model was verified by PreScan and MATLAB/Simulink simulation platforms. The results show unmanned vehicle could make an optimal response based on other parties driving strategies. The model not only ensures safety but improves the traffic efficiency at intersections. Compared to other decision systems, this method considers all possible optimization schemes, and the possible real-time prediction results.