SV-WAM is a surround‑view world‑action model that keeps all six camera views for autonomous driving while enabling efficient inference by discarding the video branch during deployment. It uses future‑video prediction as dense training supervision and introduces an action‑centered causal mask to prevent future‑video tokens from influencing action tokens during joint denoising. A differentiable drivable‑area compliance regularizer further improves safety by penalizing vehicle‑footprint corners that approach or cross drivable boundaries. Experiments on NAVSIMv2 and nuScenes show state‑of‑the‑art planning performance with low latency and strong zero‑shot transfer.
SV-WAM is a surround‑view world‑action model that keeps all six camera views while enabling efficient inference by discarding the video branch at deployment. It uses future‑video prediction as dense training supervision and an action‑centered causal mask to prevent action tokens from attending to future‑video tokens during joint denoising. A differentiable drivable‑area compliance regularizer penalizes vehicle‑footprint corners near or crossing drivable boundaries, improving safety and boundary awareness. Experiments on NAVSIMv2 and nuScenes show state‑of‑the‑art planning performance with low latency and strong zero‑shot transfer.
By Jinyang Wang, Shiwei Li, Junjian Wang, Zhiqiang Deng, Jianbin Gao, Yihang Zhao, Liu Liu, Yongjia Zhao, Jinlong Chen, Huirui Xu, Yifeng Pan, Kangwei Liu, Fan Ren, Ji Tao, Minghao Yang
arXiv:2609.36851v1 Announce Type: new
Abstract: End-to-end autonomous driving policies are commonly trained via imitation learning on logged demonstrations without observing the consequences of their...
By Hongbin Lin, Chaoda Zheng, Yiming Yang, Xiangyu Li, Shijia Chen, Jinhao Deng, Kangjie Chen, Dongbin Zhang, Jie Feng, Yu Zhang, Xianming Liu, Shuguang Cui, Boyang Wang, Zhen Li
arXiv:2606.27504v2 Announce Type: replace
Abstract: World Action Models (WAMs) unify future environment prediction with action generation for autonomous driving, yet existing approaches optimize only...
By Tianze Xia, Lijun Zhou, Kaixin Xiong, Jingfeng Yao, Zhenxin Zhu, Haiyang Sun, Bing Wang, Guang Chen, Wenyu Liu, Hangjun Ye, Xinggang Wang
arXiv:2606. 03159v1 Announce Type: cross Abstract: As autonomous vehicle capabilities advance, the safe evaluation of driving policies in long-tail scenarios remains a critical bottleneck.
By NVIDIA, :, Aarti Basant, Amlan Kar, Despoina Paschalidou, Fangyin Wei, Francesco Ferroni, Guillermo Garcia Cobo, Haithem Turki, Huan Ling, Jaewoo Seo, James Lucas, Jay Zhangjie Wu, Jialiang Wang, Jonathan Lorraine, Jun Gao, Kai He, Katarina Tothova, Kevin Xie, Micha{\l} Tyszkiewicz, Qi Wu, Riccardo de Lutio, Ruilong Li, Sanja Fidler, Seung Wook Kim, Tianchang Shen, Tianshi Cao, Tobias Pfaff, William Lew, Xindi Wu, Xuanchi Ren, Yifan Lu, Yuxuan Zhang, Zan Gojcic, Zian Wang
arXiv:2608. 10403v1 Announce Type: new Abstract: Reinforcement learning (RL) has shown promising performance in autonomous driving, yet ensuring the safety of online RL policies remains challenging due to insufficient exposure to safety-critical driving scenes.
By Xincong Hu (Nanjing University), Lei Ou (Nanjing University), Maosen Li (Yinwang Intelligent Technology Co., Ltd), Jingtao Zhang (Yinwang Intelligent Technology Co., Ltd), Liguo Hou (Yinwang Intelligent Technology Co., Ltd), Zongzhang Zhang (Nanjing University)
OPTED is a method for on‑policy fine‑tuning of end‑to‑end driving models that separates reinforcement learning from the policy update. A privileged teacher trained with RL on vectorized inputs (HD‑maps and bounding boxes) supervises the pre‑trained student during closed‑loop post‑training. Applied to the camera‑based models TransFuser and VaVAM in AlpaSim, OPTED boosts driving scores by 1.6× and 9.5×, respectively, while requiring roughly three orders of magnitude fewer simulator interactions than direct RL post‑training.
By Damiano Da Col, Maximilian Igl, Peter Karkus, Kashyap Chitta, Boris Ivanovic, Marco Pavone, Konrad Schindler, Christos Sakaridis
MILER is an end‑to‑end reinforcement learning framework that achieves zero‑shot sim‑to‑real transfer for autonomous driving in unstructured environments. It uses a custom semantic mid‑level representation (MLR) simulator for offline training, and during deployment it processes real camera and LiDAR data with BEVFusion to produce a compatible bird’s‑eye‑view representation. The policy’s actions are applied via a trajectory‑alignment strategy, allowing the system to drive 17.3 km on a 3.0 km test track without human intervention, all running on a Jetson AGX Orin.
By Thomas Steinecker, Denis Trescher, Alexander Bienemann, Thorsten Luettel, Mirko Maehlisch
As autonomous vehicle capabilities advance, the safe evaluation of driving policies in long-tail scenarios remains a critical bottleneck. In closed-loop simulation, the driving policy model actively interacts with the environment, where its actions dynamically update the simulator state and directly influence the next set of generated sensor observations.
The paper introduces CoLT-Drive, a 3,536-sample counterfactual long‑tail benchmark for evaluating decision‑level driving affordance prediction, which tests whether models can infer how rare objects affect an ego vehicle’s high‑level actions. It also proposes KPA, a knowledge‑preserving adaptation framework that combines structured prompting, expert merging, and a regime‑aware LoRA mixture‑of‑experts module to improve small VLMs on driving tasks. Experiments show KPA achieves 60.8% pair accuracy on CoLT‑Drive, outperforming the Qwen3‑VL‑2B baseline and LoRA SFT while keeping competitive in‑domain performance.
By Zhengxu Tang, Guofeng Cui, Ziyu Gong, Xiaozhou Zhang, Ruifeng Deng, Chengzhi Qi, Ke Chen, Sachin Patil, Tianjun Xiao, Langechuan Liu, Pichao Wang
arXiv:2609.26792v1 Announce Type: cross
Abstract: Faithfully evaluating end-to-end driving policies in simulation requires observations that are not merely photo-realistic, but preserve the scene fea...
By Ziyang Leng, Sicheng Mo, Seth Z. Zhao, Haoyuan Cai, Yu Zeng, Rowan McAllister, Bolei Zhou
The paper introduces a compact visual navigation system that decomposes the task into three analytically‑computed geometric interfaces and three small learned modules: an egress predictor, a navigation predictor, and an endpoint‑pinned residual diffusion generator. Only 0.58 M of the 23 M parameters are trained on 44 k frames, achieving competitive success rates and the lowest collision rate among evaluated methods across 6 060 point‑goal episodes in 60 environments. The design allows further parameter reduction by replacing the frozen image encoder with a 0.54 M MobileNetV2, supports zero‑shot deployment on a Jetson Orin Nano UGV, and enables transparent failure analysis under sensor corruption.
By Edward Beng Wai Tan, Siew-Kei Lam