mlsanityData to 7 Oct 2026

Contrastive Representation Regularization for Vision-Language-Action Models

arXiv 2510.01711 ↗First posted 2 Oct 2025cs.RO cs.LGcited by 16

On the map: Multimodal LLMs / vision-language models → Embodied and 3D vision-language models

also in 1 more
Abstract (arXiv)

Vision-Language-Action (VLA) models have shown strong capabilities in robot manipulation by leveraging rich representations from pre-trained Vision-Language Models (VLMs). However, their representations arguably remain suboptimal, lacking sensitivity to robotic signals such as control actions and proprioceptive information. To address the issue, we introduce Robot State-aware Contrastive Loss (RS-CL), a simple and effective representation regularization for VLA models, designed to bridge the gap between VLM representations and robotic signals. In particular, RS-CL aligns the representations more closely with the robot's proprioceptive states by using relative distances between the states as soft supervision. Complementing the original action prediction objective, RS-CL enhances control-relevant representation learning, while being lightweight and fully compatible with standard VLA training pipelines. Our empirical results demonstrate that RS-CL substantially improves the performance of state-of-the-art VLA models; it pushes the prior art to 69.7% achieving the state-of-the-art performance on the RoboCasa-Kitchen benchmark, and boosts success rates from 45.0% to 58.3% on challenging real-robot manipulation tasks.

Key facts. Cited by 16 papers in the map (data through 7 Oct 2026). Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 53 of its 58 references that are in the map. First posted on arXiv 2 Oct 2025.

Builds on

53 of 58 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025RoboBrain 2.0 Technical Report2 Jul
  2. 2025SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics2 Jun
  3. 2025Visual Embodied Brain: Let Multimodal Large Language Models See, Think, and Control in Spaces30 May
  4. 2025Knowledge Insulating Vision-Language-Action Models: Train Fast, Run Fast, Generalize Better29 May
  5. 2025RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete28 Feb
  6. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success27 Feb
  7. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control28 Jul
  8. 2023PaLM-E: An Embodied Multimodal Language Model6 Mar

Other references · 45

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  3. 2023Visual Instruction Tuningcited by 7,792
  4. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  5. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,482
  6. 2022Scalable Diffusion Models with Transformerscited by 5,931
  7. 2022Flow Matching for Generative Modelingcited by 5,426
  8. 2025Qwen2.5-VL Technical Reportcited by 5,418
  9. 2020Supervised Contrastive Learningcited by 2,803
  10. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  11. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  12. 2019Similarity of Neural Network Representations Revisitedcited by 1,695
33 more
  1. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,653
  2. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
  3. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,167
  4. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,127
  5. 2024DROID: A Large-Scale In-The-Wild Robot Manipulation Datasetcited by 801
  6. 2020CURL: Contrastive Unsupervised Representations for Reinforcement Learningcited by 745
  7. 2022R3M: A Universal Visual Representation for Robot Manipulationcited by 678
  8. 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 643
  9. 2017Time-Contrastive Networks: Self-Supervised Learning from Videocited by 453
  10. 2024CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulationcited by 411
  11. 2024RoboCasa: Large-Scale Simulation of Everyday Tasks for Generalist Robotscited by 391
  12. 2022VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Trainingcited by 384
  13. 2024The Platonic Representation Hypothesiscited by 362
  14. 2023MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrationscited by 308
  15. 2022Rectified Flow: A Marginal Preserving Approach to Optimal Transportcited by 277
  16. 2025DreamGen: Unlocking Generalization in Robot Learning through Video World Modelscited by 168
  17. 2025Cosmos-Reason1: From Physical Common Sense To Embodied Reasoningcited by 159
  18. 2022CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completioncited by 158
  19. 2025NORA: A Small Open-Sourced Generalist Vision Language Action Model for Embodied Taskscited by 119
  20. 2025ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action Modelcited by 112
  21. 2025FLARE: Robot Learning with Implicit World Modelingcited by 105
  22. 2024DexMimicGen: Automated Data Generation for Bimanual Dexterous Manipulation via Imitation Learningcited by 104
  23. 2025Video Generators are Robot Policiescited by 86
  24. 2023Multi-View Masked World Models for Visual Robotic Manipulationcited by 78
  25. 2020A Simple but Tough-to-Beat Data Augmentation Approach for Natural Language Understanding and Generationcited by 72
  26. 2025InstructVLA: Vision-Language-Action Instruction Tuning from Understanding to Manipulationcited by 71
  27. 2025Eagle 2.5: Boosting Long-Context Post-Training for Frontier Vision-Language Modelscited by 62
  28. 2024Diffusion-VLA: Generalizable and Interpretable Robot Foundation Model via Self-Generated Reasoningcited by 57
  29. 2021Not All Negatives are Equal: Label-Aware Contrastive Loss for Fine-grained Text Classificationcited by 43
  30. 2025HAMLET: Switch your Vision-Language-Action Model into a History-Aware Policycited by 42
  31. 2023Goal Representations for Instruction Following: A Semi-Supervised Language Interface to Controlcited by 32
  32. 2023Contrastive Language, Action, and State Pre-training for Robot Learningcited by 15
  33. 2025CLASS: Contrastive Learning via Action Sequence Supervision for Robot Manipulationcited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 16.

16 papers.

Sort:

By area:

  1. 2026Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignmentcited by 15
  2. 2026Learning to Act Robustly with View-Invariant Latent Actionscited by 6
  3. 2026Is Forward Prediction Enough? Physical State Grounding for JEPA World Modelscited by 5
  4. 2026Understanding the Impact of Geometric Foundation Models on Vision-Language-Action Modelscited by 4
  5. 2026Contrastive Conceptor Activation Steering (COAST): Unlocking Vision-Language-Action Models through Hidden Statescited by 3
  6. 2026EmbodiedMidtrain: Bridging the Gap between Vision-Language Models and Vision-Language-Action Models via Mid-trainingcited by 2
  7. 2026QuoVLA: Quotient Space for Vision-Language-Action Modelscited by 2
  8. 2026FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuningcited by 1
  9. 2026GeoProp: Grounding Robot State in Vision for Generalist Manipulationcited by 1
  10. 2026How Should Vision-Language-Action Models Use Proprioceptive State?cited by 1
  11. 2026Mitigating State Aliasing in Vision-Language-Action Models via Inverse Dynamics Learningcited by 0
  12. 2026Contrastive Action-Image Pre-training for Visuomotor Controlcited by 0
4 more
  1. 2026Semantic Anchoring for Robotic Action Representationscited by 0
  2. 2026WorldToken: Time-First Sequence Modeling for Robotic Imitation Learningcited by 0
  3. 2026Act with Intent: Distilling Behavior Intent for Vision-Language-Action Modelscited by 0
  4. 2026The Layer Mystery of VLA: An Information-Theoretical Analysis of VLA Latent Interfacecited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Enhancing Generalization in Vision-Language-Action Models by Preserving Pretrained Representations
  2. 2025VLA-0: Building State-of-the-Art VLAs with Zero Modification
  3. 2025VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Model
  4. 2026V-Link: Recovering Lost Visual Representations in Action DiT for Vision-Language-Action Models
  5. 2026Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment
  6. 2026JEPA-VLA: Video Predictive Embedding is Needed for VLA Models
  7. 2026MixVLA: Adaptive Mixing of Non-Invariant Information for Generalizable Vision-Language-Action Models
  8. 2026Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models