Model-based trajectory prediction approach using an improved dynamic window approach and an interactive behaviour model

The capacity of predicting trajectories of surrounding vehicles for autonomous vehicles is significant as it would improve safety and smoothness. We propose a novel approach which combined an improved dynamic window approach (IDWA) and an interactive behaviour model for vehicle trajectory prediction. The advantages of this approach are its adaptive ability to different scenarios and its high accuracy. Firstly, an improved approach is proposed to take into account the Ackermann steering constraint and thus the approach would be adaptive to predict vehicle trajectories in different scenarios. Secondly, to improve the accuracy of predicted results, an interactive behaviour model of vehicles is constructed. The proposed approach was tested on a real road dataset NGSIM I-80 (I) and an experimental platform PreScan. The results show that our proposed approach performs better than the baseline approach of the constant velocity model approach in NGSIM I-80 (I) and has an excellently adaptive ability in different scenarios validated in PreScan.