PreTR: Spatio-Temporal Non-Autoregressive Trajectory Prediction Transformer

Achaji, Lina; Barry, Thierno; Fouqueray, Thibault; Moreau, Julien; Aioun, François; Charpillet, François

doi:10.1109/itsc55140.2022.9922451

Cited by 7 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This was achieved by flattening the agent information and temporal sequence to apply attention simultaneously. Achaji [13] pointed out the time-consuming nature of using merged attention in Agentformer [12] and proposed a method based on temporal-spatial divided attention. They also utilized a non-autoregressive model based on learnable queries to enable parallel application of Transformers.…”

Section: A Trajectory Prediction On Bird's Eye Viewmentioning

confidence: 99%

“…Furthermore, if both training and inference operate autoregressive manner applying previous predictions, it hampers the utilization of parallelization power during training, resulting in inefficient learning time. Therefore, taking inspiration from [13], we adopt a learnable query with a temporal dimension of N p . By incorporating the learnable query, parallel decoding can be performed during both training and inference, effectively addressing the aforementioned issues.…”

Section: Transformer-based Trajectory Predictionmentioning

confidence: 99%

See 1 more Smart Citation

Non-Autoregressive Transformer Based Ego-Motion Independent Pedestrian Trajectory Prediction on Egocentric View

Kim,

Seo,

Noh

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Predicting the future trajectories of surrounding pedestrians is undoubtedly one of the most essential but challenging tasks for safe urban autonomous driving. Despite this importance, there has been limited research conducted on the egocentric view from easy-to-access vehicle-mounted cameras for autonomous driving applications. This paper presents a non-autoregressive transformer based trajectory prediction methodology for pedestrian on egocentric view. Furthermore, our proposed model predicts egomotion independent future trajectories for utilization in downstream tasks such as motion planning in autonomous vehicles. This approach differs from previous research as it focuses on predicting the future position of pedestrians based on the current observed image context, rather than their future positions in future observed images. The proposed model, referred to as the TransPred network in this paper, is composed of three main modules: vehicle motion compensation, non-autoregressive transformer, and conditional variational autoencoder(CVAE). The transformer structure is employed to effectively handle raw images and the historical trajectory of the target pedestrian, enabling the generation of advanced future predictions. Additionally, the CVAE module is utilized in the final part of the overall model to predict plausible multiple future trajectories. It contributes to generating diverse and realistic future trajectory predictions. The performance of our model has been evaluated on Nuscenes and In-house dataset obtained from our vehicle equipped with sensors. We achieves the state-of-the-art performance for prioritized trajectories on both datasets. Moreover, the effectiveness of the proposed ego-motion independent trajectories is demonstrated through risk assessment experiments.

show abstract

Section: A Trajectory Prediction On Bird's Eye Viewmentioning

confidence: 99%

Section: Transformer-based Trajectory Predictionmentioning

confidence: 99%

Non-Autoregressive Transformer Based Ego-Motion Independent Pedestrian Trajectory Prediction on Egocentric View

Kim,

Seo,

Noh

et al. 2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Modeling spatial interactions can guide the model to avoid pedestrian collisions for more realistic trajectory prediction. Current temporal interaction modeling approaches for capturing historical motion factors of pedestrians typically employ LSTM [14,15,16,17,18,19] or GRU [20,21], Transformers [22,23,24,25], and graph convolution neural networks [26,27,28]. Existing spatial interaction modeling methods capture social interactions through pooling mechanisms [14,15,29], attention mechanisms [18,23,24,30,31], and graph convolutions and their variants [21,26,27,28,32].…”

Section: Introductionmentioning

confidence: 99%

“…Previous trajectory prediction methods for modeling spatial and temporal interactions have typically summed [26,27,30] or sequenced [24,28] the features of both. However, none of these methods consider the relationship between spatial and temporal interactions.…”

Section: Introductionmentioning

confidence: 99%

STIGCN: Spatial-Temporal Interaction-aware Graph Convolution Network for Pedestrian Trajectory Prediction

Chen

Sang

Wang

et al. 2023

Preprint

View full text Add to dashboard Cite

Accurately predicting the future trajectory of pedestrians is critical for tasks such as autonomous driving and robot navigation. Previous methods for pedestrian trajectory prediction dealt with social interaction and pedestrian movement factors either concurrently or sequentially, neglecting the link between them. Therefore, a Spatial-Temporal Interaction-aware Graph Convolution Network (STIGCN) is proposed for pedestrian trajectory prediction. STIGCN considers the correlation between social interaction and pedestrian movement factors to achieve more accurate interaction modeling. Specifically, we first constructed spatial and temporal graphs to model social interactions and movement factors. Then, we designed the spatial-temporal interaction-aware learning to utilize the spatial interaction features of each moment to assist the temporal interaction modeling and utilize the temporal interaction features of each pedestrian to assist the spatial interaction modeling, resulting in more accurate interaction modeling. Finally, a Time-Extrapolator Pyramid Convolution Neural Network (TEP-CNN) is designed to jointly estimate the two-dimensional Gaussian distribution parameters of future trajectories by combining the prediction features from multiple layers. Experimental results on two benchmark pedestrian trajectory prediction datasets show that our proposed method outperforms existing methods in terms of average displacement error and final displacement error and achieves more accurate predictions for pedestrian motions such as convergence and encounter.

show abstract

STIGCN: spatial–temporal interaction-aware graph convolution network for pedestrian trajectory prediction

Chen,

Sang,

Wang

et al. 2023

J Supercomput

View full text Add to dashboard Cite

PreTR: Spatio-Temporal Non-Autoregressive Trajectory Prediction Transformer

Cited by 7 publications

References 21 publications

Non-Autoregressive Transformer Based Ego-Motion Independent Pedestrian Trajectory Prediction on Egocentric View

Non-Autoregressive Transformer Based Ego-Motion Independent Pedestrian Trajectory Prediction on Egocentric View

STIGCN: Spatial-Temporal Interaction-aware Graph Convolution Network for Pedestrian Trajectory Prediction

STIGCN: spatial–temporal interaction-aware graph convolution network for pedestrian trajectory prediction

Contact Info

Product

Resources

About