Fast Prediction of Vehicle Driving Intentions and Trajectories Based on Lightweight Methods

This paper proposes a lightweight model that can quickly predict the intention of other vehicles and their driving trajectories based on the intention predictions. This model is based on the dynamic states of the vehicles and their related road information. This lightweight model uses low-dimensional features, and the lightgbm model is used for training, which greatly reduces the training and inference time of the model. In comparison to most existing prediction algorithms, this model has a low computation complexity for real-time inference and transferability to other types of driving behaviors and road users. The numerical examples on the real-world data demonstrate that the model is practicably efficient with good accuracy and fast inference.