Vehicle Control System Based on Dynamic Traffic Gesture Recognition

In view of the lack of information processing speed and road condition recognition ability suitable for complex road conditions in the current autonomous driving technology and to solve the problem of rapid and accurate dynamic recognition of "traffic police gestures" in the current unmanned driving technology, a real-time interactive vehicle control system combining hardware and software is designed in this paper. The car takes Arduino UNO micro-controller as the core, realizes one-way data transmission from the camera to the computer through RTSP. After being processed by gesture recognition model based on YOLOv5 network structure, the corresponding instructions are sent back to the car to control the car's movement. Through the design and debugging, the information interaction between the camera, the computer and the car is completed. The car can move according to the designated traffic police gesture, which improves the degree of car automation and has certain significance to promote the further development of the auto automation industry.