Greedy Trajectory Prediction of Autonomous Vehicles Based on DTDGP-IE Framework and Hazard Index Graph

In urban road traffic environments, intention estimation and trajectory prediction of surrounding vehicles are necessary prerequisites for autonomous vehicles to plan their own trajectories reasonably. In this paper, a greedy trajectory prediction method of autonomous vehicles based on Gaussian process-information entropy probability framework in discretetime dimension (DTDGP-IE) and hazard index graph is proposed. Firstly, a greedy road grid search method based on urban road traffic rules is employed for intention estimation of surrounding vehicles. Secondly, to further improve the accuracy of prediction and realize the uncertainty estimation, a hazard index graph is utilized based on DTDGP-IE, focusing attention on vehicles with higher hazard index. Eventually, trajectory prediction of vehicles with higher hazard index selected by hazard index graph is carried out utilizing the fusion of greedy road grid sequence and state of surrounding vehicles predicted by DTDGP-IE. In addition, the lane keeping and lane changing scenarios are tested on a public autonomous driving dataset. By comparing with the classical trajectory prediction algorithms of LSTM and GRU, and analyzing the computational complexity of the proposed algorithm, it can be proved that the proposed algorithm is possessed with better prediction ability and faster calculation time under the same computational cost, which verifies the effectiveness of the proposed method.