Until recently, robotics safety research was predominantly about collision avoidance and hazard reduction in the immediate vicinity of a robot. Since the advent of large vision and language models (VLMs), robots are now also capable of higher-level semantic scene understanding and natural language interactions with humans. Despite their known vulnerabilities (e.g. hallucinations or jail-breaking), VLMs are being handed control of robots capable of physical contact with the real world. This can lead to dangerous behaviors, making semantic safety for robots a matter of immediate concern.
Our contributions in this paper are two fold: first, to address these emerging risks, we release the ASIMOV Benchmark — a large-scale and comprehensive collection of datasets for evaluating and improving semantic safety of foundation models serving as robot brains. Our data generation recipe is highly scalable: by leveraging text and image generation techniques, we generate undesirable situations from real-world visual scenes and human injury reports from hospitals. Secondly, we develop a framework to automatically generate robot constitutions from real-world data to steer a robot’s behavior using Constitutional AI mechanisms. We propose a novel auto-amending process that is able to introduce nuances in written rules of behavior – this can lead to increased alignment with human preferences on behavior desirability and safety. We explore trade-offs between generality and specificity across a diverse set of constitutions of different lengths, and demonstrate that a robot is able to effectively reject unconstitutional actions. We measure a top alignment rate of 84.3% on the ASIMOV Benchmark using generated constitutions, outperforming no-constitution baselines and human-written constitutions.
We do not advocate for a specific universal constitution in this work because rules require customization to different legal, cultural and administrative contexts; instead, we argue that human interpretability and modifiability of constitutions inferred from data makes them an ideal medium for behavior governance of AI-controlled robots.
@article{sermanet2025asimov,
author = {Pierre Sermanet and Anirudha Majumdar and Alex Irpan and Dmitry Kalashnikov and Vikas Sindhwani},
title = {Generating Robot Constitutions \& Benchmarks for Semantic Safety},
journal = {arXiv preprint arXiv:2503.08663},
url = {https://arxiv.org/abs/2503.08663},
year = {2025},
}