mlsanityData to 14 Oct 2026

Vision-Predictor-Action: Towards Efficient Embodied AI via Neuro-Symbolic JEPA Predictors

arXiv 2610.15556 ↗First posted 11 Oct 2026cs.LO cs.LG cs.ROcited by 0

On the map: Vision-language-action models → Efficient VLA inference and adaptation also fits: Reasoning and chain-of-thought in VLA

Abstract (arXiv)

Vision-Language-Action (VLA) models commonly generate robot actions token by token through Large Language Model (LLM) decoders, causing inference latency to grow with action sequence length and limiting closed-loop control frequency. We present Goal-Conditioned Vision-Predictor-Action (VPA), which replaces autoregressive decoders with a lightweight latent control loop. A shared vision encoder and a frozen text encoder embed observations, goal images, and instructions; a neuro-symbolic selector chooses a discrete skill; a Joint-Embedding Predictive Architecture (JEPA) predictor forecasts the next latent state; and a flow-matching solver generates an entire action chunk in a few integration steps, so that sequential network evaluations per decision step do not depend on the action horizon. We analyze inference complexity, regularization against representation collapse, and conditions under which latent sub-goal thresholding limits error accumulation across task phases. As an initial proof of concept, we evaluated VPA on the LIBERO-10 suite. The trainable model has under 20M parameters (checkpoint under 75 MB, excluding a frozen 63.4M-parameter CLIP text encoder); a decision step of a single-camera reference configuration takes 10-14 ms on an Apple M4; and trained only on LIBERO-10 demonstrations in under three hours on a single GPU, a two-camera policy succeeded in 40.0% of 200 evaluation episodes. These preliminary results from a single training run suggest the approach is workable. Code and weights are available at https://github.com/ylee-sling/Vision-Predictor-Action and https://huggingface.co/ylee-sling/vpa-libero10, and archived at https://doi.org/10.5281/zenodo.22960920.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models. Builds on 32 of its 38 references that are in the map. First posted on arXiv 11 Oct 2026.

Builds on

32 of 38 references.

References · 32

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  4. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  6. 2022Flow Matching for Generative Modelingcited by 5,702
  7. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,926
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,700
  9. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  10. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  11. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  12. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
20 more
  1. 2018Spectral Normalization for Generative Adversarial Networkscited by 1,636
  2. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  3. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  4. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  5. 2021VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learningcited by 1,057
  6. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
  7. 2022Building Normalizing Flows with Stochastic Interpolantscited by 995
  8. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 809
  9. 2026LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixelscited by 273
  10. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
  11. 2024ManiSkill3: GPU Parallelized Robotics Simulation and Rendering for Generalizable Embodied AIcited by 229
  12. 2020The Lipschitz Constant of Self-Attentioncited by 174
  13. 2026VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Modelcited by 118
  14. 2023Instruct2Act: Mapping Multi-modality Instructions to Robotic Actions with Large Language Modelcited by 116
  15. 2026Hierarchical Planning with Latent World Modelscited by 43
  16. 2026JEPA-VLA: Video Predictive Embedding is Needed for VLA Modelscited by 24
  17. 2025ACT-JEPA: Novel Joint-Embedding Predictive Architecture for Efficient Policy Representation Learningcited by 8
  18. 2026NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Modelscited by 2
  19. 2026Knowledge, Rules and Their Embeddings: Two Paths towards Neuro-Symbolic JEPAcited by 2
  20. 2026SJEPA: Learning Elegant Latent Dynamics with Hybrid Symbolic-Neural Predictorscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026G0.5: One Autoregressive Stream for Robot Reasoning and Action
  2. 2026Juno: Taming Predictive Latents for Vision-Language-Action Models
  3. 2025FASTer: Toward Efficient Autoregressive Vision Language Action Modeling via Neural Action Tokenization
  4. 2026ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models
  5. 2025Think Twice, Act Once: Token-Aware Compression and Action Reuse for Efficient Inference in Vision-Language-Action Models
  6. 2025Efficient Vision-Language-Action Models for Embodied Manipulation: A Systematic Survey
  7. 2026Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments
  8. 2026JEPA-WAM: Learning Vision-Language-Action Policies with Joint-Embedding World Modeling