Vehicle-road Collaborative Recognition Method Based on ResNet and Improved SSD

This paper proposes a new method of automatic driving. First of all, A* algorithm is used to determine the next step forward direction of the automatic vehicle. At the same time, ResNet network and SSD network improved by focal loss are used to avoid right lights and obstacles respectively. ResNet network is used to detect whether there is a red light in front. The SSD network improved by focal loss is used to detect whether there is an obstacle in front of the vehicle and judge whether the obstacle can affect the vehicle forward. The judgment results of the two networks are combined to decide whether the vehicle stops or not. This paper will be tested in the simulation environment, and accuracy of the combination of ResNet network and SSD network is 77.2%.