InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer

End-to-end autonomous driving, with its holistic optimization capabilities, has gained increasing traction in academia and industry. Vectorized representations, which preserve instance-level topological information while reducing computational overhead, have emerged as promising paradigms. However,...

Full description

Saved in:
Bibliographic Details
Main Authors: Bo Zhang, Heye Huang, Chunyang Liu, Yaqin Zhang, Zhenhua Xu
Format: Article
Language:English
Published: Tsinghua University Press 2025-06-01
Series:Journal of Intelligent and Connected Vehicles
Subjects:
Online Access:https://www.sciopen.com/article/10.26599/JICV.2025.9210060
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849393985118797824
author Bo Zhang
Heye Huang
Chunyang Liu
Yaqin Zhang
Zhenhua Xu
author_facet Bo Zhang
Heye Huang
Chunyang Liu
Yaqin Zhang
Zhenhua Xu
author_sort Bo Zhang
collection DOAJ
description End-to-end autonomous driving, with its holistic optimization capabilities, has gained increasing traction in academia and industry. Vectorized representations, which preserve instance-level topological information while reducing computational overhead, have emerged as promising paradigms. However, existing vectorized query-based frameworks often overlook the inherent spatial correlations among intra-instance points, resulting in geometrically inconsistent outputs (e.g., fragmented HD map elements or oscillatory trajectories). To address these limitations, we propose intra-instance vectorized driving transformer (InVDriver), a novel vectorized query-based system that systematically models intra-instance spatial dependencies through masked self-attention layers, thereby enhancing planning accuracy and trajectory smoothness. Across all core modules, i.e., perception, prediction, and planning, InVDriver incorporates masked self-attention mechanisms that restrict attention to intra-instance point interactions, enabling coordinated refinement of structural elements while suppressing irrelevant inter-instance noise. The experimental results on the nuScenes benchmark demonstrate that InVDriver achieves state-of-the-art performance, surpassing prior methods in both accuracy and safety, while maintaining high computational efficiency.
format Article
id doaj-art-00907a95763e4a86b0e3d53bffa57cd0
institution Kabale University
issn 2399-9802
language English
publishDate 2025-06-01
publisher Tsinghua University Press
record_format Article
series Journal of Intelligent and Connected Vehicles
spelling doaj-art-00907a95763e4a86b0e3d53bffa57cd02025-08-20T03:40:14ZengTsinghua University PressJournal of Intelligent and Connected Vehicles2399-98022025-06-0182921006010.26599/JICV.2025.9210060InVDriver: Intra-instance aware vectorized query-based autonomous driving transformerBo Zhang0Heye Huang1Chunyang Liu2Yaqin Zhang3Zhenhua Xu4Institute for AI Industry Research (AIR), Tsinghua University, Beijing 100084, ChinaUniversity of Wisconsin‒Madison, Madison 53707, USADiDi, Beijing 100081, ChinaInstitute for AI Industry Research (AIR), Tsinghua University, Beijing 100084, ChinaSchool of Vehicle and Mobility, Tsinghua University, Beijing 100084, ChinaEnd-to-end autonomous driving, with its holistic optimization capabilities, has gained increasing traction in academia and industry. Vectorized representations, which preserve instance-level topological information while reducing computational overhead, have emerged as promising paradigms. However, existing vectorized query-based frameworks often overlook the inherent spatial correlations among intra-instance points, resulting in geometrically inconsistent outputs (e.g., fragmented HD map elements or oscillatory trajectories). To address these limitations, we propose intra-instance vectorized driving transformer (InVDriver), a novel vectorized query-based system that systematically models intra-instance spatial dependencies through masked self-attention layers, thereby enhancing planning accuracy and trajectory smoothness. Across all core modules, i.e., perception, prediction, and planning, InVDriver incorporates masked self-attention mechanisms that restrict attention to intra-instance point interactions, enabling coordinated refinement of structural elements while suppressing irrelevant inter-instance noise. The experimental results on the nuScenes benchmark demonstrate that InVDriver achieves state-of-the-art performance, surpassing prior methods in both accuracy and safety, while maintaining high computational efficiency.https://www.sciopen.com/article/10.26599/JICV.2025.9210060end-to-end autonomous drivingdecision makingmotion predictiontransformerartificial intelligence
spellingShingle Bo Zhang
Heye Huang
Chunyang Liu
Yaqin Zhang
Zhenhua Xu
InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
Journal of Intelligent and Connected Vehicles
end-to-end autonomous driving
decision making
motion prediction
transformer
artificial intelligence
title InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
title_full InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
title_fullStr InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
title_full_unstemmed InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
title_short InVDriver: Intra-instance aware vectorized query-based autonomous driving transformer
title_sort invdriver intra instance aware vectorized query based autonomous driving transformer
topic end-to-end autonomous driving
decision making
motion prediction
transformer
artificial intelligence
url https://www.sciopen.com/article/10.26599/JICV.2025.9210060
work_keys_str_mv AT bozhang invdriverintrainstanceawarevectorizedquerybasedautonomousdrivingtransformer
AT heyehuang invdriverintrainstanceawarevectorizedquerybasedautonomousdrivingtransformer
AT chunyangliu invdriverintrainstanceawarevectorizedquerybasedautonomousdrivingtransformer
AT yaqinzhang invdriverintrainstanceawarevectorizedquerybasedautonomousdrivingtransformer
AT zhenhuaxu invdriverintrainstanceawarevectorizedquerybasedautonomousdrivingtransformer