Trajectory Prediction for Autonomous Driving Based on Structural Informer Method

Accurate and efficient prediction of the future trajectories of surrounding vehicles is of utmost importance in motion planning for autonomous driving. The ability to predict longer-term trajectories provides valuable information for effective motion planning. Numerous studies have contributed to the prediction of long-term vehicle trajectories. However, it is important to note that longer-term predictions can potentially lead to a trade-off between accuracy and computational complexity. In this work, we propose a structural Informer method, which can achieve accurate and efficient long-term trajectory prediction of the target vehicle. Specifically, the proposed method considers not only the temporal and spatial features of the interaction vehicle trajectory, but also the impact of vehicle state changes on the trajectory. To reduce computational redundancy and complexity while improving memory usage and prediction accuracy, the ProbSparse self-attention mechanisms and attention distillation operations are employed. The method is validated and evaluated using the NGSIM dataset, and the results demonstrate that the proposed structural Informer achieves satisfactory accuracy and time cost in long-term prediction of the TV compared with state-of-the art methods Note to Practitioners —The motivation of this research is to address the impact of future trajectories of surrounding vehicles on the motion planning of autonomous vehicles. The method proposed applies advanced deep learning methods, and its strongpoint is that the proposed network can achieve higher efficiency and accuracy of trajectory prediction compared with state-of-the art methods. The specific implementation method is to use structured embedding methods and networks to extract more valuable features of the target vehicle, such as spatiotemporal features and vehicle state features. The novel attention mechanism is designed to solve the problem of exponential growth in computational complexity of traditional attention mechanisms in long-term prediction. The advancement of the method proposed confirmed by verification on naturalistic driving dataset.