CAE‐GAN: A hybrid model for vehicle trajectory prediction

Trajectory prediction of surrounding vehicles is a crucial capability intelligent driving vehicles. In scene, vehicle and its constitute an integral system, the vehicle's future motion trajectory affected by actions The influencing mode degree are hidden in relevant historical information neighbour vehicle. existing methods either do not consider confidence predicted trajectory, or accuracy requirement ignored when considering trajectory. order to address this problem, mixed Conditional AutoEncoder Generative Adversarial Network (CAE-GAN) model based on multi-loss function proposed. proposed uses encoder–decoder structure with convolutional social pool extract general features. generative adversarial networks (GANs) used features generated trajectories, which enables generate trajectories that close real trajectories. addition, classifier LSTM network added output probability belongs particular lateral maneuver so consistent maneuver. evaluated using publicly available NGSIM US-101 I-80 datasets, results show higher than methods. achieves average improvement 16.34% comparison most advanced models.