Paper page - CausGround: Causal Interaction Grounding for Adaptive Embodied Planning

Source: https://huggingface.co/papers/2610.29702

Abstract

CausGround extracts reusable action preconditions and effects from interaction experience to improve embodied planning and reasoning across diverse environments without model retraining.

Multimodal large language models(MLLMs) show promising reasoning capabilities, but reliable physical interaction requires planning that accounts for when actions are executable and how they change the world. These preconditions and effects are environment-dependent and often not captured by MLLMs, making adaptation challenging. Yet existing approaches either require costly environment-specific training or rely on unstructured interaction experience that is difficult to reuse. Therefore, we introduce CausGround, a non-parametriccausal interaction groundingframework that distills interaction experience into reusable action-centered causal knowledge. CausGround identifiesaction preconditionsand effects and organizes them in afactorized world-state representationcapturing object attributes and inter-object relations. At planning time, task-relevant causal dependencies are grounded in current observations to supportbackward reasoningover preconditions andforward reasoningover effects for action selection and multi-step planning. We evaluate CausGround on simulatedhigh-level planning, interaction reasoning, and real-world robot manipulation. On adapted VirtualHome and RoboCasa benchmarks, CausGround improves vision-basedhigh-level planningacross MLLM backbones, increasing average task success from 24.9% to 53.3% on VirtualHome and closed-loop planning success from 43.0% to 57.0% on RoboCasa. On SWITCH, average interaction reasoning accuracy increases from 25.8% to 31.9%, indicating improved action selection, recovery, and verification. Moreover, real-world experiments show that CausGround reasons backward from MLLM-interpreted goals to derive executable high-level action sequences for sequential manipulation. These results demonstrate that reusableaction preconditionsand effects support long-horizonembodied planningacross diverse environments without parameter updates.

View arXiv pageView PDFProject pageGitHub1Add to collection

Get this paper in your agent:

hf papers read 2610\.29702

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2610.29702 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.29702 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2610.29702 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.