Think Twice before Driving: Towards Scalable Decoders for End-to-End Autonomous Driving
Xiaosong Jia, Penghao Wu, Li Chen, Jiangwei Xie, Conghui He, Junchi Yan, Hongyang Li
Abstract
Throttle 0.05 Steer 0.05 × 𝑵𝑵 Adjust Driving Plan Figure 1. We propose ThinkTwice, a scalable decoder paradigm that generates the future trajectory and action of the ego vehiclefor endto-end autonomous driving. Conditioned on the coarse action/trajectory, we propose the Look Module to retrieve information from critical regions and the Prediction Module to anticipate the outcome of the ego behavior. Taking features from the two modules as input, the coarse prediction is refined by predicting its offset from ground-truth. We could stack multiple such layers and scale up the capacity of the decoder with dense supervision and spatial-temporal prior.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
64 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving cites BEVDepth: Acquisition of Reliable Depth for Multi-View 3D Object Detection
- ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous Driving cites DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving
- Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detection cites STD: Sparse-to-Dense 3D Object Detector for Point Cloud
- AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning cites DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving
- Vista: A Generalizable Driving World Model with High Fidelity and Versatile Controllability cites DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving
- VADv2: End-to-End Vectorized Autonomous Driving via Probabilistic Planning cites BEVDepth: Acquisition of Reliable Depth for Multi-View 3D Object Detection
- VADv2: End-to-End Vectorized Autonomous Driving via Probabilistic Planning cites DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving
Cited by top-tier papers64
- Vista: A Generalizable Driving World Model with High Fidelity and Versatile ControllabilityShenyuan Gao, Jiazhi Yang, Li Chen, Kashyap Chitta et al.NeurIPS 2024 · 403 citations
- AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-TuningZewei Zhou, Tianhui Cai, Seth Z. Zhao, Yun Zhang et al.NeurIPS 2025 · 310 citations
- VADv2: End-to-End Vectorized Autonomous Driving via Probabilistic PlanningBo Jiang, Shaoyu Chen, Hao Gao, Bencheng Liao et al.ICLR 2026 · 259 citations
- ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous DrivingYongkang Li, Kaixin Xiong, Xiangyu Guo, Fang Li et al.ICLR 2026 · 196 citations
- DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous DrivingXiaosong Jia, Yulu Gao, Li Chen, Junchi Yan et al.ICCV 2023 · 154 citations
Builds on31
- Deformable DETR: Deformable Transformers for End-to-End Object DetectionXizhou Zhu, Weijie Su, Lewei Lu, Bin Li et al.ICLR 2021 · 7,353 citations
- A ConvNet for the 2020sZhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer et al.CVPR 2022 · 6,782 citations
- Voxel R-CNN: Towards High Performance Voxel-based 3D Object DetectionJiajun Deng, Shaoshuai Shi, Peiwei Li, Wengang Zhou et al.AAAI 2021 · 1,128 citations
- BEVDepth: Acquisition of Reliable Depth for Multi-View 3D Object DetectionYinhao Li, Zheng Ge, Guanyi Yu, Jinrong Yang et al.AAAI 2023 · 954 citations
- STD: Sparse-to-Dense 3D Object Detector for Point CloudZetong Yang, Yanan Sun, Shu Liu, Xiaoyong Shen et al.ICCV 2019 · 840 citations
Related papers
- Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous DrivingYu Yang, Jianbiao Mei, Yukai Ma, Siliang Du et al.AAAI 2025 · 53 citations
- OpenDriveVLA: Towards End-to-end Autonomous Driving with Large Vision Language Action ModelXingcheng Zhou, Xuyuan Han, Feng Yang, Yunpu Ma et al.AAAI 2026 · 119 citations
- DiffRefiner: Coarse to Fine Trajectory Planning via Diffusion Refinement with Semantic Interaction for End to End Autonomous DrivingLiuhan Yin, Runkun Ju, Guodong Guo, Erkang ChengAAAI 2026 · 4 citations
- HiP-AD: Hierarchical and Multi-Granularity Planning with Deformable Attention for Autonomous Driving in a Single DecoderYingqi Tang, Zhuoran Xu, Zhaotie Meng, Erkang ChengICCV 2025 · 4 citations
- See Tomorrow, Act Today: Foresight-Driven Autonomous DrivingBozhou Zhang, Nan Song, Yuang Wang, Jiankang Deng, Xiatian Zhu, Li ZhangCVPR Findings 2026
