Driving behavior recognition method based on trajectory data detected by millimeter wave radar

In this paper, a multi-step vehicle driving behavior recognition method based on roadside millimeter-wave radar detecting trajectory data is proposed. The time and vehicle radial speed in trajectory data are selected as characteristic parameters. By analyzing the characteristic parameters of vehicles, different vehicle driving behaviors are classified. The proposed method marks and judges single driving behaviors, continuous driving behaviors, and complete driving behaviors in the state of RA (rapid acceleration), RD (rapid deceleration), GA (general acceleration), GD (general deceleration), and CS (constant speed) by calculating acceleration, interval time, and duration. The identification of vehicle driving behavior is completed finally. Using the vehicle trajectory data of continuous traffic flow scenarios at urban signal intersections as a sample, the established recognition method is applied for recognition and the results are compared with the actual driving behaviors of the sample. The identification results are consistent with the driving behavior reflected by the sample time-speed variation curves. It shows that the identification method proposed in this paper can effectively identify five types of driving behavior, and the accurate identification results of vehicle driving behavior have significance for traffic safety and traffic congestion improvement decision-making.