ARXIV:2606.06491 · ROBOTICS · SUBMITTED 06 JUN · 03:17 UTC · FRESHNESS FRESH

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies

Dong Jing · Jingchen Nie · Tianqi Zhang · Jiaqi Liu · Huaxiu Yao · Zhiwu Lu · +1 at arXiv

TempoVLA enables controllable-speed robotic manipulation for flexible task execution.

Ship in 2-4 weeks›Score5.0Evidence unverified

Opportunity summary

Pain TempoVLA enables controllable-speed robotic manipulation for flexible task execution.

Evidence 0 refs | 3 sources | 50% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

TempoVLA enables controllable-speed robotic manipulation for flexible task execution. Yet existing Vision-Language-Action models (VLAs) only inherit a single fixed speed from training demonstrations.

METHOD

Full abstract

Robot manipulation alternates between low-risk transit phases that call for fast execution and high-risk contact stages that demand slow, precise motion. Yet existing Vision-Language-Action models (VLAs) only inherit a single fixed speed from training demonstrations. Prior efforts to accelerate VLAs through model compression, KV-cache reuse, or reinforcement learning only shift the policy from one fixed speed to another, and leave deceleration almost unexplored. We observe that the magnitude of each predicted action already governs how fast the robot moves, opening a direct route to controllable execution speed. We turn this observation into TempoVLA, a single VLA whose execution speed is controlled by an explicit condition. TempoVLA combines two coupled components. (1) A data-side Variable-Speed Trajectory Augmentation (VSTA) that re-times demonstration to any target speed by merging or splitting actions while preserving its motion semantics. (2) A model-side conditioning mechanism that feeds the speed to the policy. Statistics show that VSTA reaches the requested speed with negligible motion error. Experiments in simulation and on real-world tasks demonstrate that TempoVLA achieves flexible speed control in both directions, while VSTA additionally boosts the default $1\times$ performance via better data utilization. Furthermore, by cooperating with a large multimodal model, TempoVLA realizes dynamic speed control, accelerating through low-risk phases and decelerating for high-risk ones.

RESULT

ScienceToStartup currently rates this 5.0/10 on the public viability pass. Statistics show that VSTA reaches the requested speed with negligible motion error. Code availability is flagged in the production record; the public repository link…

WHY NOW

Robotics moved forward this cycle; last verified June 2026. Public score 5.0/10. Production flags indicate code availability.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score5.0

PainTempoVLA enables controllable-speed robotic manipulation for flexible task execution.

Evidence0 refs | 3 sources | 50% coverage

Blockerno shell-level blocker reported

Analysis summary

TempoVLA enables controllable-speed robotic manipulation for flexible task execution.

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

{ "contract_version": "paper-r2", "paper_id": "70c4616f-b5fe-4034-a633-5cd7cea90c50", "arxiv_id": "2606.06491", "canonical_route": "/paper/tempovla-learning-speed-controllable-vision-language-action-policies", "active_tab": "synced from current hash by the drawer client", "selected_artifact": "tempovla-learning-speed-controllable-vision-language-action-policies", "endpoints": { "paper_pack": "/api/v1/paper/tempovla-learning-speed-controllable-vision-language-action-policies/paper-pack", "build_passport": "/api/v1/paper/tempovla-learning-speed-controllable-vision-language-action-policies/build-passport", "mcp_resource": "sciencetostartup://surfaces/paper-workspace" } }

{ "surface": "paper", "mode": "paper", "query": "TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies", "normalized_query": "2606.06491", "route": "/paper/tempovla-learning-speed-controllable-vision-language-action-policies", "paper_ref": "tempovla-learning-speed-controllable-vision-language-action-policies", "topic_slug": null, "benchmark_ref": null, "dataset_ref": null }

{ "@context": "https://schema.org", "@graph": [ { "@type": "WebPage", "@id": "https://sciencetostartup.com/paper/tempovla-learning-speed-controllable-vision-language-action-policies#webpage", "url": "https://sciencetostartup.com/paper/tempovla-learning-speed-controllable-vision-language-action-policies", "name": "TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies", "description": "TempoVLA enables controllable-speed robotic manipulation for flexible task execution.", "isPartOf": { "@id": "https://sciencetostartup.com/#website" } }, { "@type": "ScholarlyArticle", "@id": "https://sciencetostartup.com/paper/tempovla-learning-speed-controllable-vision-language-action-policies#scholarlyArticle", "headline": "TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies", "description": "TempoVLA enables controllable-speed robotic manipulation for flexible task execution.", "url": "https://sciencetostartup.com/paper/tempovla-learning-speed-controllable-vision-language-action-policies", "sameAs": "https://arxiv.org/abs/2606.06491", "identifier": { "@type": "PropertyValue", "propertyID": "arXiv", "value": "2606.06491" }, "isAccessibleForFree": true, "isPartOf": { "@id": "https://sciencetostartup.com/#website" }, "datePublished": "2026-06-04T17:59:40.000Z", "author": [ { "@type": "Person", "name": "Dong Jing", "affiliation": { "@type": "Organization", "name": "Renmin University of China" } }, { "@type": "Person", "name": "Jingchen Nie", "affiliation": { "@type": "Organization", "name": "Fudan University" } }, { "@type": "Person", "name": "Tianqi Zhang", "affiliation": { "@type": "Organization", "name": "University of North Carolina at Chapel Hill" } }, { "@type": "Person", "name": "Jiaqi Liu", "affiliation": { "@type": "Organization", "name": "University of North Carolina at Chapel Hill" } }, { "@type": "Person", "name": "Huaxiu Yao", "affiliation": { "@type": "Organization", "name": "University of North Carolina at Chapel Hill" } }, { "@type": "Person", "name": "Zhiwu Lu", "affiliation": { "@type": "Organization", "name": "Renmin University of China" } }, { "@type": "Person", "name": "Mingyu Ding", "affiliation": { "@type": "Organization", "name": "University of North Carolina at Chapel Hill" } } ], "additionalProperty": [ { "@type": "PropertyValue", "propertyID": "viabilityScore", "value": 5 }, { "@type": "PropertyValue", "propertyID": "researchDomain", "value": "Robotics" }, { "@type": "PropertyValue", "propertyID": "commercialReadiness", "value": "code" } ] }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://sciencetostartup.com" }, { "@type": "ListItem", "position": 2, "name": "Robotics", "item": "https://sciencetostartup.com/topics" }, { "@type": "ListItem", "position": 3, "name": "TempoVLA: Learning Speed-Controllable Vision-Language-Action", "item": "https://sciencetostartup.com/paper/tempovla-learning-speed-controllable-vision-language-action-policies" } ] }, { "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "name": "What is the startup potential of \"TempoVLA: Learning Speed-Controllable Vision-Language-Action\"?", "acceptedAnswer": { "@type": "Answer", "text": "TempoVLA enables controllable-speed robotic manipulation for flexible task execution." } }, { "@type": "Question", "name": "What products could be built from this research?", "acceptedAnswer": { "@type": "Answer", "text": "Productize as a software module or API for robotic systems integrators needing flexible speed control for various manufacturing or warehousing applications." } }, { "@type": "Question", "name": "What are the practical use cases?", "acceptedAnswer": { "@type": "Answer", "text": "Develop a robotics control software for industrial automation settings, allowing machines to adjust their task speeds dynamically according to operational risks and needs." } }, { "@type": "Question", "name": "What industries could this research disrupt?", "acceptedAnswer": { "@type": "Answer", "text": "This solution could replace fixed-speed robotic systems, adding major flexibility and efficiency without altering existing hardware." } } ] } ] }

TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies

TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline