Summarize Paper logo SummarizePaper.com

AI summary · arXiv: 2610.28457v1 · cs.RO

GPT-6-Astra Understands Touch and Outperform Vision--Tactile--Language--Action Models

Ziyong Wang, Yupei Lin, Runxuan Huang, Liang Lin, Dongyu Zhang, Guangrun Wang
Original paper Submitted to arXiv on 20 Oct. 2026 https://gpt-as-tactile-policy.github.io/ License: NONEXCLUSIVE-DISTRIB 1.0

How to read this summary? Illustrated guide

The essentials

GPT-as-Tactile-Policy uses expert-provided task guidance and tactile references with GPT-6-Astra to achieve strong contact-rich manipulation without robotics-specific pretraining or per-task fine-tuning.

Based on the abstract only: check the details in the original paper.

Question

Can a general-purpose multimodal model use tactile feedback to understand contact and adjust robot actions in contact-rich manipulation? The paper examines whether expert experience and online feedback can provide these capabilities without specialized policy training.

Method

The authors summarize expert demonstrations into task guidance and tactile references, then use GPT-6-Astra to combine this context with multimodal observations and interaction history to generate control commands. They evaluate the approach on five UniVTAC tasks, reporting mean success across tasks and comparing performance with and without tactile feedback on Insert Hole.

Findings

  • Result The method achieves 93.8% mean success across five UniVTAC tasks without additional robotics-specific pretraining or per-task policy fine-tuning.
  • Result Overall performance is comparable to state-of-the-art policies, with higher success on selected tasks.
    See the passage in the paper
    Overall performance is comparable to state-of-the-art policies, with higher success on selected tasks.
  • Result On Insert Hole, tactile feedback raises success from 31% to 99%.
  • Authors’ interpretation Execution analysis indicates that the system probes and corrects actions in response to contact feedback.
    See the passage in the paper
    Execution analysis further demonstrates contact-guided probing and correction, supporting manipulation progress through explicit expert experience and online feedback.

Limitations

  • The evaluation is limited to five UniVTAC contact-rich manipulation tasks, so the abstract does not establish performance beyond this testbed.

Why it matters

The authors argue that explicit expert experience combined with online tactile feedback can enable a general-purpose multimodal model to guide contact-rich manipulation without robotics-specific pretraining or task-specific fine-tuning. This suggests an alternative to learned tactile policies that require specialized pretraining and per-task optimization.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: Multimodal models tactile feedback contact-rich manipulation robot control expert demonstrations

Read further

  • Comprehensive Summary
  • Layman's Summary
  • Blog article
  • Original abstract
The paper asks whether a general-purpose multimodal model can understand the physical significance of contact and use tactile feedback to guide robot actions, without relying on a policy specially pretrained and fine-tuned for each task. It investigates this question through contact-rich manipulation, in which a robot must interpret touch, respond to obstruction, and adjust its movements. The authors present tactile understanding as an example of the broader challenge of using feedback from physical interaction to make progress on real-world tasks. Their proposed method, GPT-as-Tactile-Policy, uses expert demonstrations to generate task guidance and tactile references. The references help the model assess contact, while the guidance supports corrective actions. GPT-6-Astra receives this information alongside multimodal observations and interaction history, then generates control commands and revises its actions in response to tactile feedback. Unlike existing learned tactile policies, the method requires neither additional robotics-specific pretraining nor per-task policy fine-tuning. Instead, it combines summarized expert experience with online feedback during execution. The evaluation covers five UniVTAC tasks. The method achieves a mean success rate of 93.8%, which the authors describe as comparable overall to state-of-the-art policies, while outperforming them on selected tasks. The clearest task-specific result is for Insert Hole: adding tactile feedback increases success from 31% to 99%. The authors also report that the system uses contact-guided probing and correction, which they interpret as evidence that expert experience and online feedback can help the model make progress. The paper concludes that a general-purpose multimodal model can act directly as a robot policy for contact-rich manipulation when supplied with task-specific guidance and tactile references. The results suggest that tactile feedback is particularly useful when actions encounter obstructions, and that expert demonstrations may help guide corrections without requiring a separate policy for every task. The authors do not claim that tactile feedback improves performance by the same amount in every case, but rather that the model can use it to make meaningful adjustments and perform strongly on the reported benchmark. The available source contains only the title, authors, and abstract, not the full paper. It therefore lacks details needed for a fuller assessment, including the experimental setup, number of trials, success definitions, comparison baselines, statistical uncertainty, demonstration-summarization method, and implementation of tactile references and control commands. Although the abstract reports results on five benchmark tasks, it does not establish how well the method generalizes to other tasks, robots, or environments. It also provides insufficient information to assess computational cost, failure modes, or whether reduced training requirements offer practical deployment advantages. The results can therefore be summarized, but the strength and scope of the conclusions cannot be independently evaluated from the abstract alone.
Created on 30 Oct. 2026

Was this synthesis useful?

Score: 0

Why do we need votes?

Votes are used to determine whether we need to re-run our summarizing tools. If the count reaches -10, our tools can be restarted.

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

Compare with the PDF

You can also chat with multiple papers at once here.

Figures and plots

Ask about a figure by number or use its question button to request visual analysis. Up to two figures and four image panels per question. Values read from plots are approximate; check the original data.
AI assistant instructions?

Tools for this paper

Explain it differently 3 free rewrites left today
Save to a collection Group papers in collections to compare them, chat with them and export them. Create a free account Log in
Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero
Citation alerts Citation alerts guide
Who works on it? The people and institutions who publish on its topic
Weekly letter Presented in "Artificial intelligence this week" of 26 October 2026. Artificial intelligence this week: What changes when AI meets the world
Journal club Premium+ Slides to present it Propose for the journal club
Versions What changed between versions
Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Papers around this one

See them on a map

Ranked by how close their title and abstract are to this paper’s. Select some to read them together.

Closest papers

  1. Very close Cited by this paper
    Systematically Exploring the Capabilities of GPT-6 Astra as Embodied Policies
    G. Chen, Xiao-Qi Cheng, Yu Deng et al. · 2026 · Not summarized yet
    Abstract

    GPT-6 Astra exhibits a remarkable ability to generate numerical robot actions, extending its role beyond high-level planning. To assess Astra's capabilities as general-purpose embodied policies, we conduct comprehensive evaluations across six domains, examining direct control, cooperation with learned policies, and feedback-driven adaptation. In gripper manipulation, Astra can correct task targets and prepare contact conditions for subsequent policy execution; hybrid control with {\pi}0.5 achieves 48% success on the evaluated RoboDojo subset. In dexterous manipulation, hybrid control achieves 50% success in ten experience-guided DexJoCo trials, while direct in-hand control struggles to coord…

  2. Very close
    FTP-1: A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation
    Chengbo Yuan, Zi-Cheng Zhang, Mingjie Zhou et al. · 2026 · arXiv.org · Not summarized yet
    Abstract

    Despite the success of vision-based generalist robotic policies, existing tactile-based policies remain tied to fixed embodiments and sensor setups. This is because tactile signals are highly heterogeneous across hardware, making cross-sensor generalization difficult. We present FTP-1,the first generalist foundation tactile policy pretrained to acquire transferable tactile manipulation abilities across diverse sensors and embodiments. FTP-1 supports varied tactile inputs, including image-, array-, and state-based signals, by using heterogeneous encoders to project them into unified morphology-aware latent tokens that are jointly modeled by a shared tactile Transformer expert. Pretrained on a…

  3. Very close
    TLA: Tactile-Language-Action Model for Contact-Rich Manipulation
    Peng Hao, Chao-Fan Zhang, Ding-Zhe Li et al. · 2025 · Robot Learning · Not summarized yet
    Abstract

    Significant progress has been made in vision-language models. However, language-conditioned robotic manipulation for contact-rich tasks remains underexplored, particularly in terms of tactile sensing. To address this gap, we introduce the Tactile-Language-Action (TLA) model, which effectively processes sequential tactile feedback via cross-modal language grounding to enable robust policy generation in contact-intensive scenarios. In addition, we construct a comprehensive dataset that contains 24k pairs of tactile action instruction data, customized for fingertip peg-in-hole assembly, providing essential resources for TLA training and evaluation. Our results show that TLA significantly outper…

  4. Very close
    Contact-Grounded Policy: Dexterous Visuotactile Policy with Generative Contact Grounding
    Zheng-Tong Xu, Yeping Wang, Ben Abbatematteo et al. · 2026 · Robotics · Not summarized yet
    Abstract

    Contact-rich dexterous manipulation with multi-finger hands remains an open challenge in robotics because task success depends on multi-point contacts that continuously evolve and are highly sensitive to object geometry, frictional transitions, and slip. Recently, tactile-informed manipulation policies have shown promise. However, most use tactile signals as additional observations rather than modeling contact state or how their action outputs interact with low-level controller dynamics. We present Contact-Grounded Policy (CGP), a visuotactile policy that grounds multi-point contacts by predicting coupled trajectories of actual robot state and tactile feedback, and using a learned contact-co…

  5. Very close
    Has GPT-6-Astra Achieved Embodied Intelligence?
    SenseNova, Joint Research Team, : et al. · 2026 · Robotics · AI summary available
    Abstract

    GPT-6-Astra demonstrates a level of competence in robotic manipulation and navigation that challenges the conventional paradigm of building specialized models for individual tasks and embodiments. In this report, we ask not only how capable GPT-6-Astra is, but what its capabilities imply for the field. First, we probe the limits of standalone GPT-6-Astra. We conduct a large-scale evaluation across 12 representative benchmarks spanning tabletop manipulation, humanoids, mobile platforms, and dexterous hands. Under a unified protocol that allows in-context learning (ICL) from training-set demonstrations and substantially increases the number of execution steps, GPT-6-Astra achieves unprecedente…

  6. Very close
    Representation-Aligned Tactile Grounding for Contact-Rich Robotic Manipulation
    Ruilin Chen, Jing-Kai Jia, Tong Yang et al. · 2026 · arXiv.org · Not summarized yet
    Abstract

    Tactile-enhanced vision-language-action (VLA) policies have been introduced for contact-rich manipulation, where critical interaction states are often hidden from vision. Future tactile prediction is a promising way to use touch because it turns tactile outcomes into supervision for action-induced contact dynamics. Yet VLA policies contain representations with different roles, from perceptual encoding to motor prediction, making it unclear where this supervision should be applied. We study this as a representation-alignment problem. Through a linear probe analysis, we find that future tactile states are most predictable from intermediate action-expert features, rather than from vision-langua…

  7. Very close Cited by this paper
    Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation
    Han Xue, Jie-Ji Ren, Wen-Di Chen et al. · 2025 · Robotics · Not summarized yet
    Abstract

    Humans can accomplish complex contact-rich tasks using vision and touch, with highly reactive capabilities such as fast response to external changes and adaptive control of contact forces; however, this remains challenging for robots. Existing visual imitation learning (IL) approaches rely on action chunking to model complex behaviors, which lacks the ability to respond instantly to real-time tactile feedback during the chunk execution. Furthermore, most teleoperation systems struggle to provide fine-grained tactile / force feedback, which limits the range of tasks that can be performed. To address these challenges, we introduce TactAR, a low-cost teleoperation system that provides real-time…

  8. Very close
    TouchGuide: Inference-Time Steering of Visuomotor Policies via Touch Guidance
    Zhemeng Zhang, Jia-Hua Ma, Xin-Cheng Yang et al. · 2026 · Robotics · Not summarized yet
    Abstract

    Fine-grained and contact-rich manipulation remain challenging for robots, largely due to the underutilization of tactile feedback. To address this, we introduce TouchGuide, a novel cross-policy visuo-tactile fusion paradigm that fuses modalities within a low-dimensional action space. Specifically, TouchGuide operates in two stages to guide a pre-trained diffusion or flow-matching visuomotor policy at inference time. First, the policy produces a coarse, visually-plausible action using only visual inputs during early sampling. Second, a task-specific Contact Physical Model (CPM) provides tactile guidance to steer and refine the action, ensuring it aligns with realistic physical contact conditi…

Prior work it builds on

  1. Close
    UniVTAC: A Unified Simulation Platform for Visuo-Tactile Manipulation Data Generation, Learning, and Benchmarking
    Bai-Jun Chen, Wei-Jie Wan, Tian-Xing Chen et al. · 2026 · arXiv.org · Not summarized yet
    Abstract

    Robotic manipulation has seen rapid progress with vision-language-action (VLA) policies. However, visuo-tactile perception is critical for contact-rich manipulation, as tasks such as insertion are difficult to complete robustly using vision alone. At the same time, acquiring large-scale and reliable tactile data in the physical world remains costly and challenging, and the lack of a unified evaluation platform further limits policy learning and systematic analysis. To address these challenges, we propose UniVTAC, a simulation-based visuo-tactile data synthesis platform that supports three commonly used visuo-tactile sensors and enables scalable and controllable generation of informative cont…

  2. Close
    VLA-Touch: Enhancing Vision-Language-Action Model With Dual-Level Tactile Feedback
    Jian-Xin Bi, K. Ma, Ce Hao et al. · 2026 · IEEE Robotics and Automation Letters · Not summarized yet
    Abstract

    Tactile feedback is generally recognized to be crucial for effective interaction with the physical world. However, state-of-the-art Vision-Language-Action (VLA) models often lack the ability to interpret and use tactile signals, limiting their effectiveness in contact-rich tasks. Incorporating tactile feedback into these systems is challenging due to the absence of large multi-modal datasets. We present VLA-Touch, an approach that enhances generalist robot policies with tactile sensing without fine-tuning the base VLA with tactile data. Our method introduces two key innovations: (1) a pipeline that leverages a pretrained tactile-language model to provide semantic tactile feedback for high-le…

  3. Close
    VITaL Pretraining: Visuo-Tactile Pretraining for Tactile and Non-Tactile Manipulation Policies
    Abraham George, Selam Gano, Pranav Katragadda et al. · 2024 · IEEE International Conference on Robotics and Automation · Not summarized yet
    Abstract

    Tactile information is a critical tool for dexterous manipulation. As humans, we rely heavily on tactile information to understand objects in our environments and how to interact with them. We use touch not only to perform manipulation tasks but also to learn how to perform these tasks. Therefore, to create robotic agents that can learn to complete manipulation tasks at a human or super-human level of performance, we need to properly incorporate tactile information into both skill execution and skill learning. In this paper, we investigate how we can incorporate tactile information into imitation learning platforms to improve performance on manipulation tasks. We show that incorporating visu…

  4. Close
    OmniVTLA: Vision-Tactile-Language-Action Models With Semantic-Aligned Tactile Sensing
    Zheng-Xue Cheng, Yi-Qian Zhang, Anni Tang et al. · 2025 · IEEE Robotics and Automation Letters · Not summarized yet
    Abstract

    Recent vision-language-action (VLA) models build upon vision-language foundation models, and have achieved promising results and exhibit the potential of task generalization in robot manipulation. However, due to the difficulty of acquiring tactile data, current VLA models significantly overlook the importance of tactile perception and fail in contact-rich tasks. To address this issue, this paper proposes OmniVTLA, a novel framework that models vision, tactile, and language for end-to-end contact-rich manipulation tasks. Specifically, we introduce ObjTac, a comprehensive tactile dataset comprising textual, visual, and tactile information for 56 objects across 10 categories. Leveraging its 13…

  5. Close
    N0-VTLA: Scaling Vision-Tactile-Language-Action Model with Latent Tactile Tokens
    NeoteAI Team, Fudan Teai Team · 2026 · arXiv.org · Not summarized yet
    Abstract

    We present $N_0$-VTLA, a vision-tactile-language-action (VTLA) foundation model capable of (1) fine-grained contact-rich manipulation with tactile perception and tactile-feedback control, and (2) offline policy improvement from stored deployment data. Building on current vision-based backbones, we propose a training recipe for tactile integration consisting of visuo-tactile pre-training, staged tactile-pathway integration, and advantage-conditioned offline policy improvement. During pre-training, the policy learns broad contact priors from NeoData, our large-scale visuo-tactile robot dataset; to our knowledge, $N_0$-VTLA is the first VTLA model pretrained on tactile data at scale. During pos…

Go further

How are these papers found?

We compare the title and abstract of this paper with those of every paper summarized on this site, and with the papers Semantic Scholar recommends, those it cites and those citing it (for a much-cited paper, its most recent citations and, from OpenAlex, the most cited ones). Among the prior work and the follow-ups, the papers that use each other in depth, a method it uses and the much-cited follow-ups come a little higher. The levels (very close, close, related) come from the similarity of the texts, not from a judgment of quality; * marks a paper without an abstract, compared by its title. The sentence under a paper is written by an AI from the two abstracts: check it in the papers. A paper marked not relevant by three readers leaves the list.

Updated on 31 October 2026.

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.