Trajectory Unified Transformer for Pedestrian Trajectory Prediction

被引:38
作者
Shi, Liushuai [1 ]
Wang, Le [1 ]
Zhou, Sanping [1 ]
Hua, Gang [2 ]
机构
[1] Xi An Jiao Tong Univ, Inst Artificial Intelligence & Robot, Natl Engn Res Ctr Visual Informat & Applicat, Natl Key Lab Human Machine Hybrid Augmented Intel, Xian, Peoples R China
[2] Wormpex AI Res, Bellevue, WA USA
来源
2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2023) | 2023年
基金
国家重点研发计划;
关键词
D O I
10.1109/ICCV51070.2023.00887
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Pedestrian trajectory prediction is an essential link to understanding human behavior. Recent work achieves state-of-the-art performance gained from hand-designed post-processing, e.g., clustering. However, this post-processing suffers from expensive inference time and neglects the probability that the predicted trajectory disturbs downstream safety decisions. In this paper, we present Trajectory Unified TRansformer, called TUTR, which unifies the trajectory prediction components, social interaction, and multimodal trajectory prediction, into a transformer encoder-decoder architecture to effectively remove the need for post-processing. Specifically, TUTR parses the relationships across various motion modes using an explicit global prediction and an implicit mode-level transformer encoder. Then, TUTR attends to the social interactions with neighbors by a social-level transformer decoder. Finally, a dual prediction forecasts diverse trajectories and corresponding probabilities in parallel without post-processing. TUTR achieves state-of-the-art accuracy performance and improvements in inference speed of about 10x - 40x compared to previous well-tuned state-of-the-art methods using post-processing.
引用
收藏
页码:9641 / 9650
页数:10
相关论文
共 50 条
[41]   Joint Intention and Trajectory Prediction Based on Transformer [J].
Sui, Ze ;
Zhou, Yue ;
Zhao, Xu ;
Chen, Ao ;
Ni, Yiyang .
2021 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2021, :7082-7088
[42]   MSTF: Multiscale Transformer for Incomplete Trajectory Prediction [J].
Liu, Zhanwen ;
Li, Chao ;
Yang, Nan ;
Wang, Yang ;
Ma, Jiaqi ;
Cheng, Guangliang ;
Zhao, Xiangmo .
2024 35TH IEEE INTELLIGENT VEHICLES SYMPOSIUM, IEEE IV 2024, 2024, :573-580
[43]   Monocular camera and IMU integration for Pedestrian Trajectory Prediction [J].
Chen, Yu-Jou ;
Chiu, Chui-Hong ;
Lin, Yu-Chen ;
Wang, Pao-Kai .
2024 IEEE SENSORS, 2024,
[44]   Physics constrained pedestrian trajectory prediction with probability quantification [J].
Sang, Haifeng ;
Wang, Jinyu ;
Liu, Quankai ;
Chen, Wangxing ;
Zhao, Zishan .
EXPERT SYSTEMS WITH APPLICATIONS, 2024, 255
[45]   PEDESTRIAN TRAJECTORY PREDICTION WITH ATTENTION AND SOCIAL BEHAVIOR INTEGRATION [J].
Li, Zhengming ;
Chen, Guoqiang ;
Chen, Keqian ;
Li, Ganchuang .
International Journal of Mechatronics and Applied Mechanics, 2024, 2024 (17) :198-207
[46]   Pedestrian trajectory prediction method based on group perception [J].
Wang, Ruyan ;
Zhou, Yudie ;
Wu, Dapeng ;
Duan, Ang ;
Cui, Yaping ;
He, Peng .
Tongxin Xuebao/Journal on Communications, 2024, 45 (12) :44-56
[47]   Modeling social interaction and intention for pedestrian trajectory prediction [J].
Chen, Kai ;
Song, Xiao ;
Ren, Xiaoxiang .
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS, 2021, 570 (570)
[48]   Pedestrian trajectory prediction based on dynamic evolving graph [J].
Mi, Jing ;
Zhang, Xu-Xiu ;
Yan, Han .
Kongzhi yu Juece/Control and Decision, 2024, 39 (07) :2345-2353
[49]   A Set of Control Points Conditioned Pedestrian Trajectory Prediction [J].
Bae, Inhwan ;
Jeon, Hae-Gon .
THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 5, 2023, :6155-6165
[50]   Pedestrian trajectory prediction model with social features and attention [J].
Zhang Z. ;
Diao Y. .
Xi'an Dianzi Keji Daxue Xuebao/Journal of Xidian University, 2020, 47 (01) :10-17and79