mlsanityData to 14 Oct 2026

Focusing on What Matters: Object-Agent-centric Tokenization for Vision Language Action models

arXiv 2509.23655 ↗First posted 28 Sep 2025cs.RO cs.AI cs.CVcited by 6

On the map: Vision-language-action models → Generalist VLA models and benchmarks

also in 1 more
Abstract (arXiv)

Vision-Language-Action (VLA) models offer a pivotal approach to learning robotic manipulation at scale by repurposing large pre-trained Vision-Language-Models (VLM) to output robotic actions. However, adapting VLMs for robotic domains comes with an unnecessarily high computational cost, which we attribute to the tokenization scheme of visual inputs. In this work, we aim to enable efficient VLA training by proposing Oat-VLA, an Object-Agent-centric Tokenization for VLAs. Building on the insights of object-centric representation learning, our method introduces an inductive bias towards scene objects and the agent's own visual information. As a result, we find that Oat-VLA can drastically reduce the number of visual tokens to just a few tokens without sacrificing performance. We reveal that Oat-VLA converges at least twice as fast as OpenVLA on the LIBERO suite, as well as outperform OpenVLA in diverse real-world pick and place tasks.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Placed in Generalist VLA models and benchmarks, one of 13 directions in Vision-language-action models. Builds on 50 of its 57 references that are in the map. First posted on arXiv 28 Sep 2025.

Builds on

50 of 57 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success27 Feb
  2. 2024π0: A Vision-Language-Action Flow Model for General Robot Control31 Oct
  3. 2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulation19 Sep
  4. 2024LLaRA: Supercharging Robot Learning Data for Vision-Language Policy28 Jun
  5. 2024OpenVLA: An Open-Source Vision-Language-Action Model13 Jun
  6. 2024RoboMamba: Efficient Vision-Language-Action Model for Robotic Reasoning and Manipulation6 Jun
  7. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control28 Jul

Other references · 43

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020Language Models are Few-Shot Learnerscited by 32,419
  4. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  5. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  6. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  7. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  8. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  9. 2015Neural Machine Translation of Rare Words with Subword Unitscited by 2,918
  10. 2024SAM 2: Segment Anything in Images and Videoscited by 2,863
  11. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  12. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
31 more
  1. 2018SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processingcited by 1,989
  2. 2020Rethinking Attention with Performerscited by 1,601
  3. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  4. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  5. 2020Object-Centric Learning with Slot Attentioncited by 876
  6. 2022A Generalist Agentcited by 808
  7. 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 665
  8. 2023BridgeData V2: A Dataset for Robot Learning at Scalecited by 657
  9. 2025SpatialVLA: Exploring Spatial Representations for Visual-Language-Action Modelcited by 519
  10. 2023Vision-Language Foundation Models as Effective Robot Imitatorscited by 349
  11. 2024Robotic Control via Embodied Chain-of-Thought Reasoningcited by 346
  12. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 339
  13. 20243D-VLA: A 3D Vision-Language-Action Generative World Modelcited by 328
  14. 2024DeeR-VLA: Dynamic Inference of Multimodal Large Language Models for Efficient Robot Executioncited by 117
  15. 2023Motif: Intrinsic Motivation from Artificial Intelligence Feedbackcited by 102
  16. 2025DexGraspVLA: A Vision-Language-Action Framework Towards General Dexterous Graspingcited by 89
  17. 2025VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Cachingcited by 86
  18. 2025SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulationcited by 61
  19. 2023An Investigation into Pre-Training Object-Centric Representations for Reinforcement Learningcited by 49
  20. 2024Efficient Diffusion Transformer Policies with Mixture of Expert Denoisers for Multitask Learningcited by 48
  21. 2024A3VLM: Actionable Articulation-Aware Vision Language Modelcited by 33
  22. 2022Curious Exploration via Structured World Models Yields Zero-Shot Object Manipulationcited by 31
  23. 2024Entity-Centric Reinforcement Learning for Object Manipulation from Pixelscited by 27
  24. 2024Jack of All Trades, Master of Some, a Multi-Purpose Transformer Agentcited by 21
  25. 2023SARA-RT: Scaling up Robotics Transformers with Self-Adaptive Robust Attentioncited by 20
  26. 2025ObjectVLA: End-to-End Open-World Object Manipulation Without Demonstrationcited by 20
  27. 2024Zero-Shot Object-Centric Representation Learningcited by 19
  28. 2023FOCUS: Object-Centric World Models for Robotics Manipulationcited by 17
  29. 2024Composing Pre-Trained Object-Centric Representations for Robotics From "What" and "Where" Foundation Modelscited by 16
  30. 2024A Large Recurrent Action Model: xLSTM enables Fast Inference for Robotics Taskscited by 14
  31. 2024ClevrSkills: Compositional Language and Visual Reasoning in Roboticscited by 8

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2025A Survey on Efficient Vision-Language-Action Modelscited by 34
  2. 2026OA-WAM: Object-Addressable World Action Model for Robust Robot Manipulationcited by 28
  3. 20262D or 3D: Who Governs Salience in VLA Models? -- Tri-Stage Token Pruning Framework with Modality Salience Awarenesscited by 0
  4. 2026Expanding Spatial and Temporal Context for Robotic Imitation Learning With Scene Graphscited by 0
  5. 2026Object-Centric Conditioning for Visuomotor Flow Matchingcited by 0
  6. 2026Correcting WHERE, Preserving HOW: Compositional Generalization for Vision-Language-Action Models via Referential Guidancecited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025VLA^2: Empowering Vision-Language-Action Models with an Agentic Framework for Unseen Concept Manipulation
  2. 2025VLA-0: Building State-of-the-Art VLAs with Zero Modification
  3. 2026VLA-ACL: Action-Consistent Visual Token Pruning for Efficient Vision-Language-Action Models
  4. 2026ObjDream: Guiding Vision-Language-Action Models with Object-Level Understanding and Future State Prediction
  5. 2025IA-VLA: Input Augmentation for Vision-Language-Action models in settings with semantically complex tasks
  6. 2025VITA-VLA: Efficiently Teaching Vision-Language Models to Act via Action Expert Distillation
  7. 2025Goal-VLA: Image-Generative VLMs as Object-Centric World Models Empowering Zero-shot Robot Manipulation
  8. 2024CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulation