Learning Safe-Stoppability Monitors for Humanoid Robots

TL;DR

This paper introduces PRISM, a data-driven neural framework for safety monitoring in humanoid robots, enabling proactive emergency stopping by learning safe-stoppability boundaries through importance sampling and sim-to-real transfer.

Contribution

It formalizes policy-dependent safe-stoppability for humanoids and presents PRISM, a novel simulation-based method that efficiently learns safety boundaries with targeted exploration.

Findings

PRISM improves data efficiency in safety boundary learning.

The pretrained monitor successfully transfers from simulation to real robots.

Modeling safety as policy-dependent enhances proactive safety monitoring.

Abstract

Emergency stop (E-stop) mechanisms are the de facto standard for robot safety. However, for humanoid robots, abruptly cutting power can itself cause catastrophic failures; instead, an emergency stop must execute a predefined fallback controller that preserves balance and drives the robot toward a minimum-risk condition. This raises a critical question: from which states can a humanoid robot safely execute such a stop? In this work, we formalize emergency stopping for humanoids as a policy-dependent safe-stoppability problem and use data-driven approaches to characterize the safe-stoppable envelope. We introduce PRISM (Proactive Refinement of Importance-sampled Stoppability Monitor), a simulation-driven framework that learns a neural predictor for state-level stoppability. PRISM iteratively refines the decision boundary using importance sampling, enabling targeted exploration of rare but safety-critical states. This targeted exploration significantly improves data efficiency while reducing false-safe predictions under a fixed simulation budget. We further demonstrate sim-to-real transfer by deploying the pretrained monitor on a real humanoid platform. Results show that modeling safety as policy-dependent stoppability enables proactive safety monitoring and supports scalable certification of fail-safe behaviors for humanoid robots.