CAPSafe: A Benchmark for Contextual Action and Plan Safety Reasoning in Embodied Agents
We introduce the Contextual Action and Plan Safety benchmark (CAPSafe), designed to evaluate safety reasoning at two levels: (1) an action-level dataset that tests whether a single action is safe given its environmental context, and (2) a plan-level dataset that evaluates whether agents can generate safe long-horizon plans for complex hazardous household tasks.
ProofPaper ↗
Key points
- Reasoning about safety is essential for embodied agents powered by large language models (LLMs).
- CAPSafe models safety conditions where hazards are situational (inferred from object properties), multiple hazards can co-occur within the same environment, and each is annotated with a relative severity level for fine-grained analysis.
- Evaluating nine closed and open-source LLMs reveals a substantial gap between hazard recognition and safe long-horizon planning.
- These results highlight that robust safety reasoning over long action sequences remains an open challenge.
Sources (1)
- [1]CAPSafe: A Benchmark for Contextual Action and Plan Safety Reasoning in Embodied AgentsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 03:33 PM
We introduce the Contextual Action and Plan Safety benchmark (CAPSafe), designed to evaluate safety reasoning at two levels: (1) an action-level dataset that tests whether a single action is safe given its environmental context, and (2) a plan-level dataset that evaluates whether agents can generate safe long-horizon plans for complex hazardous household tasks.
Reasoning about safety is essential for embodied agents powered by large language models (LLMs).
Extractive summary: sentences quoted from the sources.
Before this
- Oct 20, 2026Voxtral 3
- Oct 20, 2026RANGER: Retrieval Agent for Navigating Graphs with Semantic Evaluation Rewards
- Oct 20, 2026CovNav: Conformal Prediction-Guided Adaptive Inference for Object Goal Navigation
- Oct 19, 2026Evaluating Agents on Highly Subjective Tasks
- Oct 19, 2026Magnolia: A Persistent-Memory Agent Framework for Computational Chemistry
- Oct 19, 2026Introducing Clairic, a new System One class of extraction models