A Multimodal States Based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle Trajectory Prediction

Precise trajectory prediction of surrounding vehicles is critical for decision-making of autonomous vehicles, and learning-based approaches are well recognized for the robustness. However, state-of-the-art learning-based methods ignore (1) the feasibility of the vehicle’s multimodal state information for prediction and (2) the mutually exclusive relationship between the global traffic scene receptive fields and the local position resolution when modeling vehicles’ interactions, which may influence prediction accuracy. Therefore, we propose a “vehicle descriptor”-based long short-term memory (LSTM) model with the dilated convolutional social pooling (VD+DCS-LSTM) to cope with the above issues. Firstly, each vehicle’s multimodal state information is employed as our model’s input, and a new vehicle descriptor (VD) encoded by stacked sparse autoencoders is proposed to reflect the deep interactive relationships between various states, achieving the optimal feature extraction and effective use of multimodal inputs. Secondly, the LSTM encoder is used to encode each vehicle’s historical sequences composed of VDs, and a novel dilated convolutional social pooling (DCS) is proposed to improve modeling vehicles’ spatial interactions. Thirdly, the LSTM decoder is used to predict the probability distribution of future trajectories based on maneuvers. The validity of the overall model was verified over the NGSIM US-101 and I-80 datasets, and our method outperforms the latest benchmark.