Abstract:Prediction of the future location of vehicles and other mobile targets is instrumental in intelligent transportation system applications. In fact, networking schemes and protocols based on machine learning can benefit from the results of such accurate trajectory predictions. This is because routing decisions always need to be made for the future scenario due to the inevitable latency caused by the processing and propagation of the routing request and response. Thus, to predict the high-precision trajectory beyond the state of the art, we propose a generative adversarial network (GAN)-based vehicle trajectory prediction method, GAN-VEEP, for urban roads. The proposed method consists of three components: 1) vehicle coordinate transformation for data set preparation; 2) neural network prediction model trained by GAN; and 3) vehicle turning model to adjust the prediction process. The vehicle coordinate transformation model is introduced to deal with the complex spatial dependence in the urban road topology. Then, the neural network prediction model learns from the behavior of vehicle drivers. Finally, the vehicle turning model can refine the driving path based on the driver's psychology. Compared with its counterparts, the experimental results show that GAN-VEEP exhibits higher effectiveness in terms of the average accuracy, mean absolute error, and root-mean-squared error.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to improve the accuracy of vehicle trajectory prediction in intelligent transportation systems. Specifically, due to the rapid changes in vehicle movement and the complex spatio - temporal dependence of vehicle trajectories, high - accuracy traffic prediction has always been a challenging problem. Most of the existing traffic prediction models are trained with historical data. These models perform well in macro - traffic - flow prediction, but are insufficient in micro - vehicle - trajectory prediction for individual vehicles. Therefore, this paper proposes a vehicle trajectory prediction method (GAN - VEEP) based on the generative adversarial network (GAN), aiming to achieve high - accuracy vehicle trajectory prediction by learning the behavior patterns of drivers, especially in the urban road environment. The main contributions of the paper include: 1. **Using the generative adversarial network (GAN)**: Use GAN to learn the behavior of drivers from historical trajectory data. As far as the author knows, this is the first study to apply GAN to generate accurate vehicle trajectories. Different from the traditional GAN model, the discriminant network in this paper consists of two independent networks, which receive the input and output data of the generative network respectively to improve the prediction accuracy. 2. **Coordinate transformation model**: In order to better prepare the training data, a coordinate transformation model is proposed to pre - process the vehicle trajectory data. This model can reduce the complexity of trajectory prediction caused by the spatial dependence in historical data and improve the accuracy of trajectory prediction. 3. **Extensive experimental verification**: A large number of experiments were carried out through simulation tools, and the proposed trajectory prediction method was evaluated using the vehicle trajectory data set generated by SUMO. The experimental results prove the superior performance of GAN - VEEP in vehicle trajectory prediction. The specific problem definitions in the paper are as follows: - Vehicle trajectories are the result of complex interactions between vehicles and roads. Due to the dynamics of vehicles and the diversity of driving behaviors, it is difficult to accurately predict the future movement of vehicles. - The goal of this paper is to predict future vehicle trajectories according to the current driving conditions. The driving condition data is defined as \( V_t=\{V_{ts}\cup V_{to}\} \), where \( V_{ts} \) represents the set of vehicle states, including position coordinates \((x, y)\), instantaneous speed \( s \) and destination \( des \); \( V_{to} \) represents the road environment information, including whether there are other vehicles ahead \( fh \), the distance of the vehicle ahead \( fd \), the speed of the vehicle ahead \( fs \), the distance to the next intersection \( id \) and the traffic light state of the intersection \( tl \). - In order to improve the prediction accuracy, the vehicle data is first pre - processed, and the complex spatial dependence between urban lanes is processed through the coordinate transformation model. Then, a neural network model is used to learn the behavior model of the driver and predict the future vehicle trajectory. The prediction process can be expressed as \( V_{ts} = G(V_{t - 1}) \), where \( V_{t - 1} \) is the driving condition data at time \( t-1 \), \( V_{ts} \) is the set of vehicle states at time \( t \), and \( G \) is a mapping function from \( V_{t - 1} \) to \( V_{ts} \). - In order to deal with the problem of vehicles choosing the driving direction when reaching the next road section, a vehicle turning model based on the driver's psychology is designed. The turning strategy is defined as \( C = \{c_1, c_2,\ldots, c_b\} \), where \( c \) represents the driving direction of the vehicle at the intersection, and \( b \) is the number of intersections passed by the vehicle. In summary, this paper mainly solves how to improve the accuracy of vehicle trajectory prediction in intelligent transportation systems, especially by introducing the generative adversarial network and the coordinate transformation model to deal with the complex spatio - temporal dependence in the urban road environment.

A Novel Generation-Adversarial-Network-Based Vehicle Trajectory Prediction Method for Intelligent Vehicular Networks

Deep learning-based vehicle trajectory prediction based on generative adversarial network for autonomous driving applications

CAE‐GAN: A hybrid model for vehicle trajectory prediction

Enhanced Multimodal Trajectory Prediction for Autonomous Vehicles Using Advanced Diffusion Model Techniques

Distributed Multi-Attention Generative Adversarial Network for Surrounding Vehicles Trajectories Prediction Based On Comprehensive Social Repulsion

Graph and Recurrent Neural Network-based Vehicle Trajectory Prediction For Highway Driving

Vehicle Trajectory Prediction Based on Attention Mechanism and GAN

VTGNet: A Vision-based Trajectory Generation Network for Autonomous Vehicles in Urban Environments

A Novel Deep Neural Network for Trajectory Prediction in Automated Vehicles Using Velocity Vector Field

A Hybrid Data-Driven and Mechanism-Based Method for Vehicle Trajectory Prediction

Multimodal Trajectory Predictions for Autonomous Driving without a Detailed Prior Map

Multiagent Multimodal Trajectory Prediction in Urban Traffic Scenarios Using a Neural Network-Based Solution

Data generation for connected and automated vehicle tests using deep learning models

Group Vehicle Trajectory Prediction With Global Spatio-Temporal Graph

Vehicle trajectory prediction and generation using LSTM models and GANs

Vehicle Interactive Dynamic Graph Neural Network Based Trajectory Prediction for Internet of Vehicles

Road Traffic Network State Prediction Based on a Generative Adversarial Network

Generative adversarial network for car following trajectory generation and anomaly detection

KI-GAN: Knowledge-Informed Generative Adversarial Networks for Enhanced Multi-Vehicle Trajectory Forecasting at Signalized Intersections

Probabilistic Crowd GAN: Multimodal Pedestrian Trajectory Prediction Using a Graph Vehicle-Pedestrian Attention Network

A Cognition‐inspired Trajectory Prediction Method for Vehicles in Interactive Scenarios