NormAct: A Benchmark for Hidden Social Norm Compliance in Embodied Planning

2026-06-26

AI Summary

This work addresses a critical limitation in existing embodied planning systems, which often fulfill explicit task objectives while neglecting implicit social norms, resulting in socially inappropriate behaviors. To bridge this gap, the study introduces NormAct—a novel benchmark that evaluates multimodal large language models along three dimensions: task completion, norm compliance, and overall success. Central to this framework is NormPerceptor, a context-driven, norm-aware module that leverages vision-language understanding to automatically generate norm-relevant prompts, thereby activating and integrating social norms without requiring explicit instructions. Experimental results demonstrate that baseline models adhere to social norms in only 26.4% of cases; in contrast, incorporating NormPerceptor significantly boosts overall task success from 24.2% to 46.7%, underscoring the effectiveness and novelty of the proposed paradigm.

Abstract

Multimodal large language models (MLLMs) are increasingly deployed as embodied planners in egocentric environments, where task success requires not only achieving instructed goals but also acting in socially appropriate ways. While explicit goals may render certain actions optimal, implicit social norms often impose hidden constraints. Existing evaluations typically focus on explicit goal achievement or direct norm knowledge, seldom assessing whether planners can infer and apply these hidden constraints within action sequences. We introduce NormAct, a benchmark for embodied social-norm interactions that evaluates plans on Goal Achievement, Norm Compliance, and overall Task Success. NormAct uniquely embeds hidden norms within ordinary tasks, testing whether models can realize them without explicit instruction. Experiments with state-of-the-art MLLMs (GPT-5.4, Claude Opus 4.7, Gemini 3 Pro) reveal a significant gap: models achieve explicit goals in 67.3\% of cases, but comply with hidden norms in only 26.4\%. Cue-condition experiments indicate that this gap stems not from a lack of general social knowledge, but from challenges in activating and grounding relevant norms in context. To address this, we propose NormPerceptor, a context-conditioned cue generator that infers scene-relevant norms prior to planning, increasing Task Success from 24.2\% to 46.7\%. Our results underscore the importance of enabling embodied agents to proactively detect hidden norms, ground them in visual evidence, and integrate them as action-planning constraints. Our benchmark is publicly available at https://huggingface.co/datasets/Caleb196x/NormAct.

Problem

• social norms
• embodied planning
• hidden constraints
• norm compliance
• multimodal large language models

Innovation

• NormAct
• embodied planning
• hidden social norms
• NormPerceptor
• multimodal large language models