Failures of autonomous robots are inevitable and the way to address them is not solely by design, but by way of methods in the robot’s control architecture that are aware of the types of perturbations, their causes, conditions, and effects on the operation of the platform. In this paper, we assemble and analyze reports from approximately 450 deployments of the Sentry autonomous underwater vehicle (AUV) to categorize the different failure cases in an effort to develop a new failure taxonomy. The goal for the taxonomy is to provide robot architecture designers with a comprehensive framework that can serve as the basis for developing methods for monitoring and appropriately responding to the different types of faults that occur in actual deployments.
@inproceedings{thieraufscheutz26ieeeeras,
title={What Matters for Designing Reliable Systems: A New Taxonomy of Failure Cause-Condition-Effect Relationships Based on AUV Missions},
author={Christopher Thierauf and Matthias Scheutz},
year={2026},
booktitle={Proceedings of the IEEE RAS International Conference on Engineering Reliable Autonomous Systems},
url={https://hrilab.tufts.edu/publications/thieraufscheutz26ieeeeras.pdf}
}