Congestion-aware Multi-agent Trajectory Prediction for Collision Avoidance

Xie, Xu; Zhang, Chi; Zhu, Yixin; Wu, Ying; Zhu, Song‐Chun

doi:10.1109/icra48506.2021.9560994

Cited by 20 publications

(10 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The performance of LSTM based model in [15] is better than most of the best in class trajectory predicting algorithms. However, graph neural networks (GNN) with unique learning ability and have been able to produce higher accuracy results when applied to path prediction [7,16].…”

Section: Related Workmentioning

confidence: 99%

“…The work by Xie. et al [7] proposes a GNN [17] based teacher-student model that predicts higher accuracy trajectories than the previous models. The teacher model accepts frame-wise graph input built to reflect the positions of all the agents in the input frame.…”

Section: Related Workmentioning

confidence: 99%

“…We have successfully attained error rates comparable to state-of-the-art algorithms with lower model complexity and smaller model size. We use ATCN based novel encoder to grasp the positions of the vehicles in the scene for the past three seconds as compared to LSTM encoders in modern models [7,15]. The output of encoders is used to condense the details of vehicular interaction in the past using 2D convolution.…”

Section: Related Workmentioning

confidence: 99%

“…ATCN encoders embed vehicles' path histories, both neighbors and ego vehicle, into higher dimensions in order to capture their past trajectory. As opposed to previous works [7,4] DeepTrack uses two different ATCN encoders. There is one ATCN shared by all neighbors, shown by the shaded box in Fig.…”

Section: Atcn Encodermentioning

confidence: 99%

“…Compared to CS-LSTM and its other variant CS-LSTM (M) [5], DeepTrack reduces prediction error by 9.09% and 11.56%, respectively, and reduces the number of operations and model size by about 10.49% and 18.5%, respectively. Concerning CF-LSTM [7], DeepTrack error raised for 0.87%; however, it can reduce the number of operations and model size by 10.49% and model size by 18.5%. This article is organized as follows: Section 2 reviews the literature.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

DeepTrack: Lightweight Deep Learning for Vehicle Path Prediction in Highways

Katariya,

Baharani,

Morris

et al. 2021

Preprint

View full text Add to dashboard Cite

Vehicle trajectory prediction is an essential task for enabling many intelligent transportation systems. While there have been some promising advances in the field, there is a need for new agile algorithms with smaller model sizes and lower computational requirements. This article presents DeepTrack, a novel deep learning algorithm customized for real-time vehicle trajectory prediction in highways. In contrast to previous methods, the vehicle dynamics are encoded using Agile Temporal Convolutional Networks (ATCNs) to provide more robust time prediction with less computation. ATCN also uses depthwise convolution, which reduces the complexity of models compared to existing approaches in terms of model size and operations. Overall, our experimental results demonstrate that DeepTrack achieves comparable accuracy to state-of-the-art trajectory prediction models but with smaller model sizes and lower computational complexity, making it more suitable for real-world deployment.Predicting multiple possible trajectories for an active subject in the scene [4, 5] is a common practice. These trajectories are ranked based on the probability distribution of the prediction model, which may not be helpful in a real-time scenario. Hence, a deep learning algorithm for vehicle trajectory forecasting in a real-time safety-critical application must provide a single trajectory with high precision. It is imperative to consider the interactions of the surrounding

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Atcn Encodermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations