基于轨迹相似度的行人轨迹预测方法研究

On the trajectory similarity and prediction of pedestrian trajectory

针对传统方法行人轨迹预测结果与行人实际行进路线偏差量较大的问题,提出了基于轨迹相似度的行人轨迹预测方法.引入社会关系影响理论,确定行人之间的最近距离,建立行人轨迹模型;结合轨迹相似度方法,选取最为相似的行人轨迹,并简化展示该轨迹预测点,得到行人的预测轨迹.实验证明,该方法得到轨迹的偏差量较小,准确度较高,适用于受多种因素影响时行人的轨迹预测.

This paper proposes the pedestrian trajectory prediction on basis of trajectory similarity with regard to the deviation between the traditional approaches. The theory of social relationship influence is introduced, and the shortest distance between pedestrians determined and a pedestrian trajectory model established. The trajectory similarity-based prediction points of the trajectory are selected and simplified. Thus, this method minimizes the deviation, which is suitable for pedestrian route prediction when affected by multiple factors.