Research on pedestrian trajectory prediction method based on informer

Since the pedestrian trajectory has certain correlation with the behavioral actions of pedestrians, in order to consider the influence of pedestrian behavioral features on pedestrian movement and improve the prediction performance of pedestrian trajectory, this paper proposes a pedestrian trajectory prediction framework that integrates pedestrian behavior. Firstly, the pedestrian behavior recognition network is used to discriminate the behavioral features of pedestrians and obtain the pedestrian behavior recognition results; then, the historical trajectory of pedestrian movement is used as the input feature of the trajectory prediction network, the recognized pedestrian behavior results are fused, and the time coding is added to generate the future trajectory of pedestrians. Finally, the effectiveness of the proposed pedestrian trajectory prediction method by fusing behavioral features is verified through experiments.