The uncertain behavior of surrounding vehicles in lane changing scenario is one of the most potential risk for the autonomous vehicle. Trajectory prediction plays a critical role in the condition of safe obstacle avoidance. Considering the partial observable state of the surrounding vehicles, trajectory prediction model of surrounding vehicles for autonomous vehicle based on POMDP (Partially Observable Markova Decision Process) principle have been proposed in this paper. The driver's intention recognition and the vehicle trajectory have been applied to train the prediction model parameters. The experiment has been implemented with NGSIM traffic data set. Experimental results show that the trajectory prediction model we proposed can predict the trajectory accurately. Results of preview time ahead have dramatic advancement with performance surpassing classic method.
Trajectory Prediction of Surrounding Vehicles for Autonomous Vehicle Using POMDP Method
2022-09-23
4185790 byte
Conference paper
Electronic Resource
English
Parametric trajectory prediction of surrounding vehicles
IEEE | 2017
|TRAJECTORY PREDICTION OF SURROUNDING VEHICLES USING PREDEFINED ROUTES
European Patent Office | 2021
|Trajectory prediction of surrounding vehicles using predefined routes
European Patent Office | 2022
|British Library Conference Proceedings | 2021
|