Vehicle Trajectory Prediction with Lane Stream Attention-Based LSTMs and Road Geometry Linearization

Abstract

It is essential for autonomous vehicles at level 3 or higher to have the ability to predict the trajectories of surrounding vehicles to safely and effectively plan and drive along trajectories in complex traffic situations. However, predicting the future behavior of vehicles is a challenging issue because traffic vehicles each have different drivers with different driving tendencies and intentions and they interact with each other. This paper presents a Long Short-Term Memory (LSTM) encoder–decoder model that utilizes an attention mechanism that focuses on certain information to predict vehicles’ trajectories. The proposed model was trained using the Highway Drone (HighD) dataset, which is a high-precision, large-scale traffic dataset. We also compared this model to previous studies. Our model effectively predicted future trajectories by using an attention mechanism to manage the importance of the driving flow of the target and adjacent vehicles and the target vehicle’s dynamics in each driving situation. Furthermore, this study presents a method of linearizing the road geometry such that the trajectory prediction model can be used in a variety of road environments. We verified that the road geometry linearization mechanism can improve the trajectory prediction model’s performance on various road environments in a virtual test-driving simulator constructed based on actual road data.

A Comprehensive Investigation of Lane-Changing Risk Recognition Framework of Multi-Vehicle Type Considering Key Features Based on Vehicles’ Trajectory Data

“…LSTM addresses the problem of vanishing gradients in RNNs and can identify long-term dependencies in sequential data by incorporating a memory state and gate mechanism to update the cell state's information retention. Each LSTM unit consists of three gates, the forget gate, the input gate, and the output gate, which selectively control the flow of information into the unit [52], Figure 4a depicts the structure of LSTM.…”

Machine Learning-Based Lane-Changing Behavior Recognition and Information Credibility Discrimination

“…To prevent overfitting in the RNN model training, a larger data set was used, and a layer of dropout was added between the input layer and the hidden layer, and between the hidden layer and the output layer, respectively [49,50], randomly dropping some neurons during training. This allowed the model to not be too dependent on any one neuron, thus avoiding overfitting.…”

Behavior trajectory prediction research of adjacent vehicles based on LSTM

“…As shown in Figure 4, the social tensor is added to the convolution module to further improve interaction information extraction and is used to record the main vehicle. As soon as the convolution operation is performed, what is extracted from the grid far from the main body can be considered observation information, and what is extracted from the grid close to the main body can be considered interaction information [8]. In addition, to extract and fuse the comprehensive relationship between the subject and the adjacent vehicles, avg-pooling is especially added, and finally, FC is used to fuse the above information.…”