Skip to content

OpenDriveLab/HDGT

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HDGT: Modeling the Driving Scene with Heterogenity and Relativity

HDGT: Heterogeneous Driving Graph Transformer for Multi-Agent Trajectory Prediction via Scene Encoding [IEEE TPAMI 2023] pipeline

Introduction

HDGT is an unified heterogeneous transformer-based graph neural network for driving scene encoding. It is a SOTA method on INTERACTION and Waymo Motion Prediction Chanllege.

By time of release in April 2022, the proposed method achieves new state-of-the-art on INTERACTION Prediction Challenge and Waymo Open Motion Challenge, in which we rank the first and second respectively in terms of the minADE/minFDE metric.

Getting Started

License

All assets and code are under the Apache 2.0 license unless specified otherwise.

Bibtex

If this work is helpful for your research, please consider citing the following BibTeX entry.

@article{jia2023hdgt,
  title={HDGT: Heterogeneous Driving Graph Transformer for Multi-Agent Trajectory Prediction via Scene Encoding},
  author={Jia, Xiaosong and Wu, Penghao and Chen, Li and  Liu, Yu  and Li, Hongyang and Yan, Junchi},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)},
  year = {2023},
}  
@inproceedings{jia2022temporal,
  title={Towards Capturing the Temporal Dynamics for Trajectory Prediction: a Coarse-to-Fine Approach},
  author={Jia, Xiaosong and Chen, Li and Wu, Penghao and Zeng, Jia and  Yan, Junchi and Li, Hongyang and Qiao, Yu},
  booktitle={CoRL},
  year={2022}
} 

About

[IEEE T-PAMI 2023] Unified heterogeneous transformer-based graph neural network for motion prediction

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Sponsor this project

 

Packages

No packages published

Languages