Skip to content
AI News HubLIVE
Original source2 min read

MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics

Summary

Researchers introduce MESA, a framework for systematically evaluating the generalization of language-conditioned robotic manipulation policies. It offers five test suites that isolate in-distribution performance and four axes of generalization: spatial configuration, object instance, object category, and subtask composition. The team also develops MESA-Gen, a pipeline for scalable task and demonstration generation, and releases both a single-arm MESA suite and BiMESA, the first large-scale benchmark for bimanual generalization. Across seven baselines, the best model achieves only 52% on MESA and 49% on BiMESA, indicating ample room for improvement. MESA rankings broadly align with RoboArena scores.

SourcearXiv RoboticsAuthor: Albert Wilcox, Frank Chang, Nhi Nguyen, Aishani Chakraborty, Jeremy A. Collins, Vaibhav Saxena, Benjamin Joffe, Siddharth Karamcheti, Animesh Garg
MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 3 Nov 2026]

Title:MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics

View a PDF of the paper titled MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics, by Albert Wilcox and 8 other authors

View PDF HTML (experimental)

Abstract:Despite growing interest in foundation models for robotic manipulation, systematic evaluation of their capabilities remains elusive. Existing benchmarks are either narrow and saturated, offering little signal on SOTA methods, broadly scoped, providing diversity at the cost of controlled experimentation, or require training on large, messy open-source datasets. We introduce MESA, an evaluation framework designed for precise, reproducible measurement of language-conditioned policy generalization. MESA provides five evaluation suites that isolate in-distribution performance and four distinct axes of generalization: spatial configuration, object instance, object category, and subtask composition, enabling researchers to diagnose where and why policies fail, not just whether they fail. To populate these suites at scale, we develop MESA-Gen, a pipeline for task and demonstration generation built by improving MimicGen. We instantiate the framework on two platforms and release both: the single-arm MESA suite, and BiMESA, the first large-scale benchmark for bimanual generalization. We evaluate seven baseline methods spanning diffusion policies, VLM-based approaches, and pretrained VLAs, finding that the strongest model achieves only 52% on MESA and 49% on BiMESA, providing meaningful signal while leaving ample room for progress. MESA rankings are broadly consistent with RoboArena scores, suggesting that MESA captures method-level differences relevant beyond simulation. Project website: this https URL.

Comments: 51 pages, 12 figures, including appendices. Conference on Robot Learning (CoRL), 2026. Project website: this https URL

Subjects:

Robotics (cs.RO); Artificial Intelligence (cs.AI)

Cite as: arXiv:2611.05567 [cs.RO]

(or arXiv:2611.05567v1 [cs.RO] for this version)

https://doi.org/10.48550/arXiv.2611.05567

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Albert Wilcox Iii [view email] [v1] Tue, 3 Nov 2026 20:01:03 UTC (7,748 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics, by Albert Wilcox and 8 other authors

View PDF

HTML (experimental)

TeX Source

view license

Additional Features

Audio Summary

Current browse context:

cs.RO

new | recent | 2026-11

Change to browse by:

cs cs.AI

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

InvestorsAdvanced

Key points

  • MESA provides five evaluation suites that separate in-distribution performance from four generalization axes: spatial, instance, category, and composition, enabling diagnosis of failure causes.
  • MESA-Gen, built by improving MimicGen, generates tasks and demonstrations at scale to populate the suites.
  • The framework is instantiated on two platforms: the single-arm MESA suite and BiMESA, the first large-scale bimanual generalization benchmark.
  • Seven baselines spanning diffusion policies, VLM-based methods, and pretrained VLAs are evaluated; the strongest achieves only 52% on MESA and 49% on BiMESA, and rankings are broadly consistent with RoboArena.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Research

MESA: An Evaluation Framework for Compositional, Semantic, and Spatial Generalization in Robotics | AI News Hub