Two-Stream Networks for Lane-Change Prediction of Surrounding Vehicles

Llorca, David Fernández; Biparva, Mahdi; Izquierdo-Gonzalo, Rubén; Tsotsos, John K.

doi:10.1109/itsc45102.2020.9294326

Cited by 18 publications

(20 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Trajectory prediction methods based on machine learning mainly include Gaussian process, hidden Markov model, or Bayesian network [13][14][15]. Tran et al [16][17][18] learned the model parameters of vehicle trajectory through a Gaussian process, and Patterson et al [19,20] used a mixed Gaussian model to learn the vehicle trajectory generation model. Streubel et al [21,22] independently predicted the discrete action of each object using a hidden Markov model.…”

Section: Introductionmentioning

confidence: 99%

Vehicle Motion State Prediction Method Integrating Point Cloud Time Series Multiview Features and Multitarget Interactive Information

Ruibin

Guo

Long

et al. 2022

Journal of Advanced Transportation

View full text Add to dashboard Cite

A vehicle motion state prediction algorithm integrating point cloud timing multiview features and multitarget interaction information is proposed in this work to effectively predict the motion states of traffic participants around intelligent vehicles in complex scenes. The algorithm analyzes the characteristics of object motion that are affected by the surrounding environment and the interaction of nearby objects and is based on the complex traffic environment perception dual multiline light detection and ranging (LiDAR) technology. The time sequence aerial view map and time sequence front view depth map are obtained using real-time point cloud information perceived by the LiDAR. Time sequence high-level abstract combination features in the multiview scene are then extracted by an improved VGG19 network model and are fused with the potential spatiotemporal interaction of the multitarget operation state data extraction features detected by the laser radar by using a one-dimensional convolution neural network. A temporal feature vector is constructed as the input data of the bidirectional long-term and short-term memory (BiLSTM) network, and the desired input-output mapping relationship is trained to predict the motion state of traffic participants. According to the test results, the proposed BiLSTM model based on point cloud multiview and vehicle interaction information is better than other methods in predicting the state of target vehicles. The results can provide support for the research to evaluate the risk of intelligent vehicle operation environment.

show abstract

Section: Introductionmentioning

confidence: 99%

Vehicle Motion State Prediction Method Integrating Point Cloud Time Series Multiview Features and Multitarget Interactive Information

Ruibin

Guo

Long

et al. 2022

Journal of Advanced Transportation

View full text Add to dashboard Cite

show abstract

“…The novel vision data is used to predict modified lane change prediction with object detector in multi-layer perceptron algorithms [10]. Based on the video information, the Two-Stream Convolutional Networks, the Two-Stream Inflated 3D Convolutional Networks, the Spatiotemporal Multiplier Networks, and the SlowFast Networks are used to identify and predict interactive behaviors by analyzing different size areas around the vehicle [11]. The above method has insufficient advance time in predicting vehicle right-of-way change, which poses a significant challenge to implementing corresponding feedback for the ego-vehicle.…”

Section: Related Workmentioning

confidence: 99%

“…Adaptive Momentum Estimation uses momentum and root mean square based on gradient descent to improve the performance of sparse gradient problems and adjusts it according to the average value of the weight gradient. The back-propagation gradient dt at time step t is shown in Equation (11).…”

Section: Model Optimizationmentioning

confidence: 99%

Vehicle Interaction Behavior Prediction with Self-Attention

Xin

Lian

et al. 2022

Sensors

View full text Add to dashboard Cite

The structured road is a scene with high interaction between vehicles, but due to the high uncertainty of behavior, the prediction of vehicle interaction behavior is still a challenge. This prediction is significant for controlling the ego-vehicle. We propose an interaction behavior prediction model based on vehicle cluster (VC) by self-attention (VC-Attention) to improve the prediction performance. Firstly, a five-vehicle based cluster structure is designed to extract the interactive features between ego-vehicle and target vehicle, such as Deceleration Rate to Avoid a Crash (DRAC) and the lane gap. In addition, the proposed model utilizes the sliding window algorithm to extract VC behavior information. Then the temporal characteristics of the three interactive features mentioned above will be caught by two layers of self-attention encoder with six heads respectively. Finally, target vehicle’s future behavior will be predicted by a sub-network consists of a fully connected layer and SoftMax module. The experimental results show that this method has achieved accuracy, precision, recall, and F1 score of more than 92% and time to event of 2.9 s on a Next Generation Simulation (NGSIM) dataset. It accurately predicts the interactive behaviors in class-imbalance prediction and adapts to various driving scenarios.

show abstract

“…Selecting a list of hand-crafted features requires the expert's knowledge and might not result in optimal choices for the prediction model. In contrast to hand-crafted feature selection, some of the existing studies utilize Deep Neural Networks (DNNs) to learn relevant features from raw sensor data [5], [9]. Although such a strategy leads to no information loss, large computational resources are required to learn relevant features from high-dimensional raw sensor data, which is challenging due to the limited computational resources of an automated vehicle.…”

Section: A Input Representationmentioning

confidence: 99%

“…In [7], a model containing several Gated Recurrent Unit (GRU) is proposed to model pairwise interaction between the TV and each of the SVs. Convolutional Neural Networks (CNNs) have been used in some of the existing studies mainly to model spatio-temporal dependencies in image-like input data such as simplified BEV representation [4], [12], [25] or raw sensor data [5], [9]. In [19], the input features are categorized into TV's motion, Right lane SVs, Left Lane SVs, Same lane SVs and Street features based on which part of the environment they represent.…”

Section: B Prediction Modelmentioning

confidence: 99%

Early Lane Change Prediction for Automated Driving Systems Using Multi-Task Attention-based Convolutional Neural Networks

Mozaffari,

Arnold,

Dianati

et al. 2021

Preprint

View full text Add to dashboard Cite

Two-Stream Networks for Lane-Change Prediction of Surrounding Vehicles

Cited by 18 publications

References 24 publications

Vehicle Motion State Prediction Method Integrating Point Cloud Time Series Multiview Features and Multitarget Interactive Information

Vehicle Motion State Prediction Method Integrating Point Cloud Time Series Multiview Features and Multitarget Interactive Information

Vehicle Interaction Behavior Prediction with Self-Attention

Early Lane Change Prediction for Automated Driving Systems Using Multi-Task Attention-based Convolutional Neural Networks

Contact Info

Product

Resources

About