arXiv Computer Vision

RoXDrive: Closed-Loop Reinforcement Learning for End-to-End Autonomous Driving via Action-Faithful Rollouts

arXiv AI
Jun 30

WoVR: World Models as Reliable Simulators for Post-Training VLA Policies with RL

arXiv:2602. 13977v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) promises to unlock capabilities beyond imitation learning for Vision--Language--Action (VLA) models, but its requirement for massive real-world interaction prevents direct deployment on physical robots.

By Zhennan Jiang, Shangqing Zhou, Yutong Jiang, Zefang Huang, Mingjie Wei, Yuhui Chen, Tianxing Zhou, Zhen Guo, Hao Lin, Quanlu Zhang, Yu Wang, Haoran Li, Chao Yu, Dongbin Zhao
arXiv Computer Vision
6d ago

MAPLE: Latent Multi-Agent Play for End-to-End Autonomous Driving

MAPLE is a new framework that enables reactive, multi‑agent rollout of dynamic driving scenarios directly in the latent space of vision‑language‑action models. It trains the ego vehicle and surrounding traffic agents independently over multiple steps, using supervised fine‑tuning followed by reinforcement learning with safety, progress, and interaction rewards. The approach is scalable, does not rely on external simulators, and achieves state‑of‑the‑art performance on Bench2Drive.

By Rajeev Yasarla, Deepti Hegde, Hsin-Pai Cheng, Shizhong Han, Yunxiao Shi, Meysam Sadeghigooghari, Hanno Ackermann, Litian Liu, Pranav Desai, Fatih Porikli, Mohammad Ghavamzadeh, Hong Cai
arXiv Machine Learning
Oct 23

TerraTransfer++: Demonstration-Free End-to-End Planner

TerraTransfer++ demonstrates a demonstration‑free, end‑to‑end autonomous driving framework that leverages self‑play in vectorized simulators to pretrain a policy. The method distills self‑play rollouts into a flow‑matching trajectory teacher and trains lightweight adapters on a frozen vision backbone, using action KL divergence, a batch‑relational low‑rank loss, and trajectory‑matching loss. The resulting closed‑loop policy surpasses all published imitation‑based methods on the HUGSim benchmark, while the planner achieves state‑of‑the‑art performance on NAVSIM navhard and navtest without any human demonstrations.

By Zikang Xiong, Weixin Li, Zhouchonghao Wu, Akshay Rangesh, Saarth Bonde, Grantland Hall, Chen Tang, Yihan Hu, Wei Zhan
arXiv Computer Vision
Sep 4

SV-WAM: An Efficient Surround-View World-Action Model for End-to-End Autonomous Driving

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 Computer Vision
Sep 7

Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learning

The paper introduces a reinforcement learning post‑training scheme that trains robot world models on their own autoregressive rollouts, using a contrastive RL objective adapted from diffusion models. It also proposes a training protocol that compares multiple variable‑length futures, a multi‑view visual fidelity reward, and demonstrates state‑of‑the‑art rollout fidelity on the DROID dataset, outperforming baselines on LPIPS, SSIM, and human preference tests.

By Jai Bardhan, Patrik Drozdik, Josef Sivic, Vladimir Petrik
Hugging Face Trending Papers
Sep 3

SV-WAM: An Efficient Surround-View World-Action Model for End-to-End Autonomous Driving

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.

arXiv AI
Oct 15

DriveWLA: Unifying Future Dynamics and Scene Semantics through World-Language-Action Modeling for Autonomous Driving

DriveWLA is a unified World‑Language‑Action model that integrates semantic understanding, future scene dynamics, and trajectory generation for autonomous driving. It employs a Mixture‑of‑Transformers architecture with three experts—World, Language, and Action—connected by Cross‑Expert Attention to align semantic and dynamic representations. After reinforcement learning refinement, DriveWLA outperforms existing VLM‑based and world‑model‑based planners on NAVSIM benchmarks and generalizes well to other datasets without fine‑tuning.

By Shuzheng Wang, Haoxuan Li, Shengzhen Li, Chenxi Wu, Yajing Sun, Lu Hou, Wenyu Liu, Xinggang Wang
arXiv Computer Vision
6d ago

Generative Scenario Rollouts for End-to-End Autonomous Driving

Generative Scenario Rollouts (GeRo) is a plug‑and‑play framework that augments Vision‑Language‑Action models for autonomous driving by jointly planning and generating language‑grounded future traffic scenes through an autoregressive rollout strategy. The approach trains a VLA model to encode vehicle and agent dynamics into latent tokens, then uses language‑conditioned autoregressive generation to produce future latent tokens and textual responses, with a rollout‑consistency loss to maintain temporal coherence and text‑action alignment. On the Bench2Drive benchmark, GeRo improves driving score by +15.7 and success rate by +26.2, and achieves state‑of‑the‑art closed‑loop and open‑loop performance, demonstrating strong zero‑shot robustness.

By Rajeev Yasarla, Deepti Hegde, Shizhong Han, Hsin-Pai Cheng, Yunxiao Shi, Meysam Sadeghigooghari, Shweta Mahajan, Apratim Bhattacharyya, Litian Liu, Risheek Garrepalli, Thomas Svantesson, Fatih Porikli, Mohammad Ghavamzadeh, Hong Cai
arXiv Computer Vision
Aug 26

Do Robotic World Models Really Follow Actions? Diagnosing and Aligning Action-Conditioned Generation for Policy Learning

arXiv:2608.24885v1 Announce Type: cross Abstract: Action-conditioned world models are increasingly used as learned simulators for policy evaluation and improvement, yet their effectiveness rests on a...

By Sixiang Chen, Jiaming Liu, Jixian Wu, Yichen Guo, Tinghao Wang, Siyuan Qian, Hao Chen, Jiajun Cao, Jian Tang, Shanghang Zhang