Skip to content
Live
Hugging Face Daily Papers·· 9 hours ago65

Mind2Act World: A Diagnostic Benchmark for Embodied Mind, Physical Action, and Mind-Action Composition in Robotic Manipulation

Mind2Act World: A Diagnostic Benchmark for Embodied Mind, Physical Action, and Mind-Action Composition in Robotic Manipulation

Brief overview

The Mind2Act World benchmark includes 180 simulation operation configurations, evaluating the ability of embodied systems to combine historical reasoning, precise execution, and environmental feedback, revealing that single-view capabilities are not suitable for combined tasks. The benchmark includes 30 tasks, divided into three views: mind (decision-making based on historical observations), action (precise execution), and mind-action combination (closed-loop tasks), used to diagnose whether the system lacks a certain capability or cannot combine existing capabilities.

Article

Published on Nov 19

#1 Paper of the day

Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

A diagnostic benchmark of 180 simulated manipulation configurations evaluates whether embodied systems can combine history-based reasoning, precise execution, and closed-loop environmental feedback, revealing that single-view competence does not transfer to composed tasks.

Embodied intelligence aims to let robots perform complex physical tasks with human-like adaptability and reliability, which requires three complementary capabilities: history-based cognitive reasoning, precise and timely physical execution, and the composition of reasoning and execution through environmental feedback. Yet existing benchmarks either evaluate reasoning and execution within the same tasks or assess them separately without their composition, so they reveal little about whether a failure comes from reasoning, execution, or their composition. We introduce Mind2Act World, a diagnostic benchmark of 30 simulated manipulation tasks from five real-world application domains. Each task has three difficulty levels and two scene settings (Clean and Random), yielding 180 configurations. The tasks form three views: Mind tests decisions based on past observations, Act tests precise and timely execution, and Mind2Act combines both in closed-loop tasks where action outcomes guide later decisions. Comparing the three views shows whether a system lacks a capability or fails to compose capabilities it already has. A diagnosis harness further attributes the failed episodes of three representative systems to Mind, Act, or their composition and summarizes each system in a profile card. Our evaluation of multimodal agents, vision-language-action models (VLAs), and world-action models (WAMs) shows that competence in a single view does not carry over to Mind2Act tasks. Even the strongest agent succeeds in 52.7% of Mind and 39.0% of Act episodes but in only 12.0% of Mind2Act episodes. The strongest agent retains task state but struggles with sustained contact, whereas end-to-end policies execute practiced motions but retain little task state. These findings point to composing reasoning and execution through environmental feedback as the main open problem.

View arXiv page View PDF Project page GitHub 21 Add to collection

Get this paper in your agent:

hf papers read 2611.26444

Don't have the latest CLI?

curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2611.26444 in a model README.md to link it from this page.

Datasets citing this paper 1

Mind2Act-World/mind2act-train

Updated about 2 hours ago

•

5.91k

•

1

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2611.26444 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

Source: Hugging Face Daily Papers · huggingface.co