Answers are grounded in this paper and cite their sources.
Proposes a new safety approach to maintain control performance in data-driven systems during input deviations, suggesting continuous operation under unexpected conditions.
In data-driven control systems, machine learning-based controllers are widely adopted, but their performance and safety may degrade when inputs deviate from the training data range. While conventional fail-safe designs rely on anomaly detection followed by shutdown, modern applications demand continuous operation even under unexpected conditions. This study proposes a safety design approach that detects early signs of input deviation based on changes in model outputs and performs model correction or retraining as needed. The goal is to achieve fail-soft operation that ensures safety without interrupting control.
No takes yet. Share an insight, caveat, or question.
Saito et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: