Failing Gracefully: Mitigating Impact of Inevitable Robot Failures
Duc Nguyen, Saad Abdul Ghani, Andrew Marshall, Allison Andreyev, Gregory Stein, Xuesu Xiao
AI summary
Problem
Current robot safety methods focus on preventing or recovering from failures, neglecting the consequences of inevitable, unpredictable malfunctions in dynamic household environments, and lack standardized benchmarks for evaluation.
Approach
The authors introduce a mathematical formulation that quantifies the probability and severity of harmful interactions between robot components and environmental entities during failures, integrating it into a motion planning objective alongside a MuJoCo-based simulation framework with a configurable failure injector.
Key results
- Novel failure impact formulation combining interaction probability and severity metrics
- FailBench simulation framework with curated scenes, planners, and a configurable failure injector
- Demonstration that the formulation effectively guides risk-aware motion planning in simulated failure scenarios
Why it matters
Provides a critical foundation for developing and benchmarking safer, more resilient service robots capable of operating reliably in human-shared spaces despite inevitable system failures.
Abstract
Service robots operate in household environments shared with humans, pets, and everyday objects, where they are highly susceptible to failures such as software crashes, hardware degradation, or unpredictable interactions. While roboticists strive to minimize failures, some remain inevitable, making it critical to mitigate their potential consequences for safe and reliable deployment. This paper introduces a novel safety formulation that evaluates both the probability of impactful interactions between robots and surrounding entities during failures, and the severity of their outcomes. By quantifying the impact of failures on different entities, our approach enables robots to make informed planning decisions that balance safety with task efficiency. To support systematic evaluation, we also present FailBench, a MuJoCo-based simulation framework for studying robot-environment interactions under diverse fail- ure modes, including sensing issues and actuator malfunctions. Together, our safety formulation and FailBench provide a foundation for developing safer and more robust motion plans and learned policies in real-world household environments.