mlsanityData to 14 Oct 2026

RoHIL: Robust Human-in-the-Loop Robotic Reinforcement Learning Against Illumination Variations

arXiv 2605.19924 ↗First posted 19 May 2026cs.ROcited by 0

On the map: Deep reinforcement learning methods → RL for robot locomotion and manipulation

Abstract (arXiv)

Human-in-the-loop reinforcement learning systems achieve near-perfect success on the workstation where they are trained, but collapse when the same robot is moved to a workstation a few meters away due to shifts in the visual input distribution caused by new lamp positions and window light. Re-collecting demonstrations and re-running HIL on every workstation is incompatible with deployment, and naively fine-tuning on shifted-light data triggers catastrophic forgetting of the source workstation. To close this cross-domain gap, we present RoHIL, an offline fine-tuning framework that uses no extra real-robot interaction. RoHIL combines (i) a world-model-based image relighter that re-synthesises the visual stream of source-workstation trajectories under multiple virtual HDRI environments, leaving actions and rewards real; (ii) Illumination-Retention Replay (IRR), a data-level anti-forgetting mechanism that interleaves relit adaptation transitions with original-light retention transitions to preserve source-workstation Bellman coverage; and (iii) an anchored Bellman-actor regulariser that constrains representation and policy drift from the original source-workstation policy. Across four real-robot manipulation tasks under significant cross-workstation illumination variations, RoHIL substantially improves shifted-light performance where standard HIL-RL collapses, while preserving source-workstation performance, eliminating the need to re-collect data and retrain for every new workstation and environment. Project page: https://anonymous4365.github.io/RoHIL/

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in RL for robot locomotion and manipulation, one of 15 directions in Deep reinforcement learning methods. Builds on 37 of its 43 references that are in the map. First posted on arXiv 19 May 2026.

Builds on

37 of 43 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024RLDG: Robotic Generalist Policy Distillation via Reinforcement Learning13 Dec
  2. 2024Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learning29 Oct

Other references · 35

  1. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,885
  2. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  4. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  5. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasetscited by 2,581
  6. 2016Learning without Forgettingcited by 2,496
  7. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  8. 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,333
  9. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  10. 2017Gradient Episodic Memory for Continual Learningcited by 1,898
  11. 2020Conservative Q-Learning for Offline Reinforcement Learningcited by 1,809
  12. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
23 more
  1. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,498
  2. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  3. 2021Offline Reinforcement Learning with Implicit Q-Learningcited by 1,293
  4. 2018Experience Replay for Continual Learningcited by 1,095
  5. 2020Dark Experience for General Continual Learning: a Strong, Simple Baselinecited by 911
  6. 2020Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixelscited by 585
  7. 2020Reinforcement Learning with Augmented Datacited by 468
  8. 2015Policy Distillationcited by 440
  9. 2021Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learningcited by 372
  10. 2023Efficient Online Reinforcement Learning with Offline Datacited by 363
  11. 2024Diffusion Policy Policy Optimizationcited by 280
  12. 2018HG-DAgger: Interactive Imitation Learning with Human Expertscited by 266
  13. 2023Cal-QL: Calibrated Offline RL Pre-Training for Efficient Online Fine-Tuningcited by 237
  14. 2025ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policycited by 154
  15. 2023VidToMe: Video Token Merging for Zero-Shot Video Editingcited by 113
  16. 2025Cosmos-Transfer1: Conditional World Generation with Adaptive Multimodal Controlcited by 106
  17. 2023Imitation Bootstrapped Reinforcement Learningcited by 64
  18. 2024Efficient Online Reinforcement Learning Fine-Tuning Need Not Retain Offline Datacited by 63
  19. 2025DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Modelscited by 57
  20. 2025UniRelight: Learning Joint Decomposition and Synthesis for Video Relightingcited by 40
  21. 2025Lumen: Consistent Video Relighting and Harmonious Background Replacement with Video Generative Modelscited by 13
  22. 2024Learning Visual Generative Priors without Textcited by 8
  23. 2025Physically-based Lighting Generation for Robotic Manipulationcited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025RaC: Robot Learning for Long-Horizon Tasks by Scaling Recovery and Correction
  2. 2026Can Vision-Language-Action Models Learn from Real-World Data Continually without Forgetting?
  3. 2026From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
  4. 2026HiL-ResRL: A Model-Agnostic Finetuning Adapter via Human-in-the-loop Residual Reinforcement Learning
  5. 2021Robust Robotic Control from Pixels using Contrastive Recurrent State-Space Models
  6. 2026EvoHIL: Self-Evolving Reward and Flow-Matched Policy Optimization for Robust Human-in-the-Loop Reinforcement Learning
  7. 2024SERL: A Software Suite for Sample-Efficient Robotic Reinforcement Learning
  8. 2026VE2VF: Vision-Enabled to Vision-Free Distillation via Real-world Reinforcement Learning for Robust Contact-Rich Manipulation