Robot Policy Generalization

Follow this topic Ask about this topic

Momentum

13 papers in the last four weeks, up 333% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 144

All topics

Many Ways to Succeed: Diversity-Driven RL Fine-Tuning for VLA Generalization
Oct 7, 2026Haoru Li, Jinmei Liu, Zhiyong Wang +5Robot Policy Generalization RL Fine-Tuning
Nanjing University · Harbin Institute of Technology (Shenzhen) · Australian National University +1
Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models
Oct 5, 2026Zaibin Zhang, Binghao Ran, Yuhan Wu +10Bimanual Robotic Manipulation Vision-Language-Action Models
Dalian University of Technology · Shanghai Jiao Tong University · The Chinese University of Hong Kong, Shenzhen +1
RoboIRS: Inference-Time Internal Representation Steering for Generalist Robot Policies
Oct 3, 2026Jiuzhou Lei, Chang Liu, Dayou Li +5Robot Policy Generalization Test-Time Adaptation of VLA Models
J. Mike Walker ’66 Department of Mechanical Engineering, Texas A&M University, College Station, TX 77843, USA · Edwardson School of Industrial Engineering, Purdue University, West Lafayette, IN, USA · Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX 77843, USA +1
Ego4WAM: What Matters When Scaling Egocentric Human Data for Robot Learning?
Sep 30, 2026Zhihao Sun, Liu Liu, Xinjiang Wang +6Data-Efficient Robot Learning Robot Imitation Learning
Institute of Trustworthy Embodied AI, Fudan University · Horizon Robotics · Huazhong University of Science & Technology +1
Scale and Selection: What Makes Automatic Harness Evolution Work for Visual-Interface Robot Agents
Sep 30, 2026Zhijie Wei, Ferris Tan, Jinghui WangAgent Harness Optimization Robot Policy Generalization
Novaxbot
Memorize, Adapt, Ignore: Diagnosing Robot Learning Mechanisms under Training Data Variation
Sep 29, 2026Ke Zhang, Danica J. Sutherland, Chao LiuDomain Randomization Robot Policy Adaptation
PRIME Robotics Lab, Department of Mechanical Engineering, The University of British Columbia · Department of Computer Science, University of British Columbia · Alberta Machine Intelligence Institute
Skill-Space Shooting for Autonomous Robot Policy Improvement
Sep 29, 2026Zihang Rui, Renhao Wang, Haoxu Huang +1Robot Policy Learning Robot Policy Adaptation
Tsinghua University · UC Berkeley · Shanghai Qi Zhi Institute
ComManip: Overfitting Manipulation Policies to Comfortable Regions
Sep 29, 2026Yidan Lai, Xinyi Chen, Shuquan Man +1Visuomotor Policy Learning Robotic Manipulation
South China University of Technology, Guangzhou, China · Nanyang Technological University, Singapore
SkillWeaver: Agentic Exploration over Neural Interaction Skills for Scalable Robot Data Generation
Sep 28, 2026He Zhu, Lusen Zhao, Kwan Man Cheng +2Robot Skill Learning Robotics Simulation
Carnegie Mellon University · University of Illinois Urbana-Champaign
Agent Priors-guided Policy Learning
Sep 28, 2026Puming Jiang, Tianrun Hu, Haozhe Du +4Robot Skill Learning Robot Imitation Learning
National University of Singapore
Recursive Harness Distillation across Agents for Robot Manipulation
Sep 27, 2026Seungyeon Kim, Junhoo Lee, Minkyu Kim +2Robotic Manipulation Robot Policy Generalization
Seoul National University · Korea Advanced Institute of Science and Technology
RACaP: Agentic Reasoning, Acting, and Coding as Policies for Evolvable Robot Learning
Sep 24, 2026Zexi Li, Yehang Zhang, Haojian Huang +10Robot Policy Learning Robot Policy Generalization
The Chinese University of Hong Kong · Knowin AI · The Hong Kong University of Science and Technology (Guangzhou)
ReShoot: Generative Visual Domain Randomization of Recorded Robot Demonstrations for Visuomotor Policy Learning
Sep 17, 2026Chiyoung Kim, Min Sung Choi, Jinho Ju +5Domain Randomization Visuomotor Policy Learning
Chung-Ang University, Seoul, Republic of Korea
RecMorph: Topology-Guided Spatial Recurrence for Generalized Morphology Control
Sep 16, 2026Quanrui Rao, Yong Liu, Xueming Xiao +4Recurrent Neural Networks Cross-Embodiment Robot Learning
Jilin University, China · State Key Laboratory of Special Vehicle Design and Manufacturing Integration Technology, China · Changchun University of Science and Technology, China
XPACE: Joint World and Action Modeling from Heterogeneous Experience
Sep 15, 2026Jiacheng Wei, Jerry Bai, Xiaoyu Yue +13Cross-Embodiment Robot Learning Synthetic Data Generation for Robotics
World Model Team, XPENG Robotics
MINERVA: How Small Can a Manipulation Policy Be and Still Solve LIBERO?
Sep 3, 2026Kohei Sendai, Tatsuya Matsushima, Yusuke IwasawaRobot Skill Learning Visuomotor Policy Learning
Matsuo-Iwasawa Lab, Graduate School of Engineering, The University of Tokyo, Japan
Enhancing Visual Domain Robustness in Behaviour Cloning via Saliency-Guided Augmentation
Aug 12, 2026Zheyu Zhuang, Ruiyu Wang, Nils Ingelhag +2Data Augmentation Domain Generalization
KTH Royal Institute of Technology · Aalto University
BooST: Bridging Semantics and Motions for Efficient Skill Transfer
Aug 11, 2026Jusuk Lee, Daesol Cho, Jonghun Shin +4Agent Skill Learning Robot Policy Generalization
Department of Aerospace Engineering, Automation and Systems Research Institute (ASRI), Seoul National University, Seoul, Korea · Georgia Institute of Technology, Atlanta, GA, USA
Curriculum Generation under Structured Parametric Environments for Robust Navigation Policies
Aug 9, 2026Prishita RayCurriculum RL Robot Policy Generalization
Cornell University
GeniWorld: A Generalizable Interactive World Model for Robotic Manipulation via Visual Actions
Aug 6, 2026Chenghao Gu, Hanyang Yu, Jingbo Zhang +7Action-Conditioned World Models Robotic Policy Evaluation
Shenzhen International Graduate School, Tsinghua University · Tencent Robotics X · The Hong Kong University of Science and Technology +1
Why Does Action Chunking Improve Behavioral Cloning Performance in Robotic Control?
Aug 3, 2026Filippo Lazzati, Kyle Stachowicz, William Chen +3Ensemble Learning Robot Imitation Learning
Politecnico di Milano · UC Berkeley
Push-Wiper: Toward General-Purpose Robotic Cleaning across Varied Stains and Surfaces with Segmented Pushing Trajectories
Aug 1, 2026Renhao Lu, Mingxin Wang, Chenyang Cao +5Robotic Manipulation Robot Policy Generalization
Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen 518055, China · Z-Lab, Zerith Robotics, Shenzhen 518055, China · University of Toronto, Toronto M5S 1A1, Canada
Diagnosing Compositional Generalization in Sequential Robot Tasks
Jul 31, 2026Yixiao Wang, Cheng-En Wu, Lingfeng Sun +5Language-Conditioned Robot Manipulation Robotic Data Collection
University of California, Berkeley · Tsinghua University
X-NavDP: Generalizing Navigation Diffusion Policy to Novel Behavior and Embodiments with Group Q-score Reweighted Matching
Jul 30, 2026Tianyu Yang, Yiming Zeng, Wenzhe Cai +5Robot Navigation Policy Optimization
1Fudan University · 3Sun Yat-sen University · 2Shanghai AI Laboratory +1
World Action Planner: Generalizable Decision-Making with Action-Conditioned World Models
Jul 30, 2026Xiangcheng Zhang, Yilun DuWorld Models for Robotics World Model-Based Planning
Harvard University
Practice Makes Policies: Bootstrapping and Consolidating Robotic Capabilities from Zero Human Demonstrations
Jul 29, 2026Jialiang Li, Yuhan Wang, Haojun Li +5Continual Robot Learning Robotic Data Collection
Shanghai Jiao Tong University · University of Sussex
Scale Up Strategically: Learning Compositional Generalization via Bias-Aware Evaluation and Data Collection for Robotic Manipulation
Jul 23, 2026Yu Qi, Zhang Ye, Xinyi Xu +6Robot Skill Learning Language-Conditioned Robot Manipulation
Northeastern University · NVIDIA
Extreme-RGMT: Continual Learning of Highly Dynamic Skills for Robust Generalist Humanoid Control
Jul 22, 2026Yubiao Ma, Han Yu, Kai Guo +5Continual Robot Learning Robot Skill Learning
School of Automation, Beijing Institute of Technology, Beijing 100081, China. · Humanoid Robotics (Shanghai) Co., Ltd., Shanghai 201203, China. · School of Mechanical Engineering, Shandong University, Jinan 250061, China.
UniETP: Unifying Environments for Generalizable Embodied Task Planning
Jul 20, 2026Peiran Xu, Jiaqi Zheng, Ziyou Wang +1Robot Task Planning Robot Policy Generalization
Peking University
Data and Learning Where it Matters for Contact-Rich Manipulation
Jul 17, 2026Oliver Hausdörfer, Linus Schwarz, Gabor Marko +7Robot Skill Learning Contact-Rich Robotic Manipulation
1TU Munich · 2Siemens AG

Oct 7, 2026, cs.ROJ/K move · Enter open · S save

Many Ways to Succeed: Diversity-Driven RL Fine-Tuning for VLA Generalization

Haoru Li, Jinmei Liu, Zhiyong Wang, Xiaoming Li+4

Nanjing University · Harbin Institute of Technology (Shenzhen) · Australian National University+1

Robot Policy Generalization RL Fine-Tuning Instruction Generalization in VLA Models

Open paper AskPDF arXiv

Reinforcement learning (RL) fine-tuning improves vision-language-action (VLA) policies through closed-loop experience, yet generalization beyond the fine-tuning distribution remains limited. Our analysis reveals a selective reshaping of exploration: RL contracts behavior globally, yet diversifies successful trajectories, elicits success with fewer rollouts, and covers more of the latent task-valid solution space than supervised fine-tuning. Broader successful-mode coverage may provide alternative strategies under distribution shifts. Inspired by this, we introduce DRIVE (Diversity-driven RL fIne-tuning for VLA gEneralization), which turns successful-behavior diversity into an explicit RL objective. DRIVE groups rollouts under matched task conditions, compares their trajectories with temporal alignment, and derives a success-conditioned intrinsic reward from relative behavioral diversity. This design encourages broader coverage of feasible solutions without rewarding diverse failures or superficial timing differences. Across LIBERO-Plus, ManiSkill3, and RoboTwin 2.0, DRIVE improves the average out-of-domain (OOD) performance over vanilla RL fine-tuning by 5.3 points on

π_0

and 2.0 points on

π_{0.5}

. On a dual-arm AgileX PiPER-X platform, DRIVE further increases average OOD success from 64.1% to 73.3% (+9.2 points), demonstrating gains that persist under physical deployment.

Momentum

Related topics

Latest papers 144