10 embodied-AI and robotics papers accepted at NeurIPS 2026, 3 with code.10 篇被 NeurIPS 2026 接收的具身智能与机器人论文，其中 3 篇附代码。
Feed papers are listed when their arXiv comment says they were accepted (e.g. 'Accepted to CoRL 2026'); papers that do not say so are missing, and the list grows as authors post camera-ready versions. Curated entries carry a hand-checked venue.日常 arXiv 论文仅在其 arXiv 备注写明已被接收（如 “Accepted to CoRL 2026”）时列入；未注明的论文不会出现，列表会随作者上传终稿而增加。精选条目的会议/期刊信息经过人工核对。
Instead of feeding a VLA a longer frame history, learns to condense past frames into a few tokens that keep short-lived cues such as brief contacts and occlusions.不靠简单加长历史帧窗口，而是学习把过去的帧压缩成少量 token，保留短暂接触、遮挡等转瞬即逝的线索。
@misc{hong2026condensevla,
title = {{CondenseVLA: Learnable History Condensation for Efficient Multi-Frame VLA}},
author = {Feiyang Hong and Yujie Wei and Bo Zhao and Xiu Su and Hongxun Yao and Shuo Yang},
year = {2026},
eprint = {2610.25927},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2610.25927},
}
200K+ navigation-planning samples annotated from camera trajectories up to instructions; RL post-training inside a video world model, used as a cheap stand-in environment, reportedly helps on unseen scenes.包含 20 万余条导航规划样本的数据集，场景多样，标注从相机运动轨迹到高层语义指令；并把视频世界模型当作低成本交互环境做强化学习后训练，报告称在未见场景上有明显提升。
@misc{feng2026embodiedworld,
title = {{EmbodiedWorld-200K: Toward Embodied World Agents via Embodied-Planning Dataset and Interactive World Models}},
author = {X. Feng and J. Tang and Z. Lin and L. Chen and S. Shao and Q. Lu and K. Huang},
year = {2026},
eprint = {2610.25959},
archivePrefix = {arXiv},
primaryClass = {cs.AI},
url = {https://arxiv.org/abs/2610.25959},
}
Replay-free lifelong imitation learning: hierarchical dual-routing attention retrieves experts by instruction, then refines by context, plus key regularization against drift. Reports beating replay baselines on LIBERO.无回放的终身模仿学习：分层双路由注意力先按指令检索专家、再按上下文细化，并用键级几何正则抑制表征漂移带来的路由干扰；报告在多个 LIBERO 任务序列上优于无回放和有回放基线。
Learns skills separately, then consolidates them after the fact with a nested hypernetwork that generates VLA parameters instead of merging weights.先独立学习各项技能，再用嵌套超网络“生成”VLA 参数来事后整合技能，而不是直接合并权重。
Adapts a language-guided navigation policy to a new building from one iPhone scan, compiled into a training environment with synthetic instructed episodes. Reports 85% real success vs 49% for a demo-fine-tuned baseline.只需用 iPhone 扫描一次新建筑：扫描被编译成训练环境并自动合成带语言指令的导航回合，再用无评论家的闭式 Pareto 效用适配策略；报告真机成功率 85%，用人类示教后训练的基线为 49%。
@misc{yu2026one,
title = {{One Scan is Enough: Demonstration-Free Adaptation for Language-Guided Navigation}},
author = {Zhuoyuan Yu and Yuxing Long and Du Wang and Junyang Wang and Hongwei Fan and Hao Dong},
year = {2026},
eprint = {2610.33794},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.33794},
}
Multi-objective reinforcement learning is a fundamental problem for modern AI, with applications spanning post-training, reasoning models, and robotic control. A central challenge is to learn policies that balance competing objectives, such as helpfulness versus honesty in safety and alignment, or speed versus energy in robotics. The optimal policy can vary sharply with objective-weighting combinations, and the extent of this variation depends crucially on the degree of conflict among objectives, making per-preference policy learning sample-inefficient. In this paper, we introduce …
Jie Wang, Guang Li, Zhijian Huang, Jinlong Li et al.
Abstract摘要
Autonomous driving fundamentally demands dense understanding of visual, semantic, and geometric information, yet existing Vision-Language-Action (VLA) models are typically trained with sparse supervision such as language instructions or trajectory signals. This direct mapping from dense visual observations to sparse representations forces the model to compress or even discard critical scene information essential for safe driving. We argue that this limitation does not stem from insufficient model capacity, but from the lack of explicit mechanisms that encourage the formation of dense world …
@misc{wang2026think,
title = {{Think Densely, Act Sparsely: Latent Expert Cognitive Chains for Vision-Language-Action Autonomous Driving}},
author = {Jie Wang and Guang Li and Zhijian Huang and Jinlong Li and Chenxu Dang and Hangjun Ye and Yahong Han and Long Chen},
year = {2026},
eprint = {2610.25734},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.25734},
}
Vision-Language-Action (VLA) models offer a promising paradigm for autonomous driving. However, the massive parameter scale of VLAs strains onboard computational resources during real-time inference. This directly increases energy consumption and limits vehicle battery range, making it critical to minimize computational overhead without compromising driving capability and explainability. Current research on efficient VLA-based driving focuses on sequential adaptation (\textit{e.g.}, Chain-of-Thought and token adjustment), leaving structural adaptation underexplored despite parameter …
@misc{yang2026scene,
title = {{Scene-Adaptive VLA: Efficient Autonomous Driving via Dynamic Layer Routing}},
author = {Yujia Yang and Quan Yuan and Jiawei Fan and Yang Li and Tiange Fu and Xiaoyuan Fu and Guiyang Luo and Jinglin Li},
year = {2026},
eprint = {2610.24590},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.24590},
}
Zhixu Li, Keqian Tang, Litian Gong, Jingyu Yao et al.
Abstract摘要
Personalization is essential for general-purpose robots to transition into household environments and achieve true practicality. While general robotic manipulation has been extensively studied, personalized robotic manipulation (PeRM) lacks a controlled benchmark for separating task completion from preference satisfaction. We introduce LIBERO-PeRM, a benchmark for preference learning in embodied agents. LIBERO-PeRM studies five questions: 1) how policies follow preferences in zero-shot settings; 2) how they learn single preferences from demonstrations; 3) how efficiently they acquire …
@misc{li2026libero,
title = {{LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation}},
author = {Zhixu Li and Keqian Tang and Litian Gong and Jingyu Yao and Wenqian Zhang and Tianze Xu and Zehao Wang and Yuping Wang and others},
year = {2026},
eprint = {2610.22372},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.22372},
}
Motasem Alfarra, Pietro Mazzaglia, Markus Peschl, Daniel Dijkman et al.
Abstract摘要
Reinforcement-learning policies for robot locomotion can fail at deployment when operating conditions shift (e.g., rough or slippery terrain, payload changes, sensing degradation), and retraining or manual recalibration is often infeasible under on-device latency and memory constraints. We study test-time adaptation (TTA) for robotic control: adapting a pretrained policy online and in a self-supervised manner during execution, without access to rewards. Our starting point is a lightweight baseline that updates only the actor using a frozen pretrained critic as a deployment-time performance …
@misc{alfarra2026efficient,
title = {{Efficient Test-Time Adaptation For Robot Policies}},
author = {Motasem Alfarra and Pietro Mazzaglia and Markus Peschl and Daniel Dijkman and Christos Louizos},
year = {2026},
eprint = {2610.20318},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.20318},
}