Out of Distribution Detection in Self-adaptive Robots with AI-powered Digital Twins

2025-09-16

AI Summary

Addressing the challenge of proactive out-of-distribution (OOD) anomaly detection for adaptive robots operating in complex, uncertain environments, this paper proposes ODiSAR—an AI-driven digital twin framework for OOD detection. ODiSAR integrates a Transformer-based digital twin model, Monte Carlo Dropout, and reconstruction error analysis to jointly quantify both prediction error and epistemic uncertainty. Crucially, it introduces an explainability layer that maps anomalous decisions to specific state dimensions, enabling interpretable, autonomous robot adaptation. Evaluated on industrial robotic manipulation and maritime navigation tasks, ODiSAR achieves 98% AUROC, 96% true negative rate at 95% true positive rate (TNR@TPR95), and 95% F1-score—significantly improving both detection accuracy and interpretability for previously unseen anomalies. This work establishes a novel paradigm for robust operation of situated adaptive robots (SAR) under distributional shift.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Adversarial Learning & Robustness
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Self-adaptive robots (SARs) in complex, uncertain environments must proactively detect and address abnormal behaviors, including out-of-distribution (OOD) cases. To this end, digital twins offer a valuable solution for OOD detection. Thus, we present a digital twin-based approach for OOD detection (ODiSAR) in SARs. ODiSAR uses a Transformer-based digital twin to forecast SAR states and employs reconstruction error and Monte Carlo dropout for uncertainty quantification. By combining reconstruction error with predictive variance, the digital twin effectively detects OOD behaviors, even in previously unseen conditions. The digital twin also includes an explainability layer that links potential OOD to specific SAR states, offering insights for self-adaptation. We evaluated ODiSAR by creating digital twins of two industrial robots: one navigating an office environment, and another performing maritime ship navigation. In both cases, ODiSAR forecasts SAR behaviors (i.e., robot trajectories and vessel motion) and proactively detects OOD events. Our results showed that ODiSAR achieved high detection performance -- up to 98% AUROC, 96% TNR@TPR95, and 95% F1-score -- while providing interpretable insights to support self-adaptation.

Problem

• Detecting out-of-distribution behaviors in self-adaptive robots
• Using AI-powered digital twins for uncertainty quantification
• Providing interpretable insights to support robot self-adaptation

Innovation

• Transformer-based digital twin forecasting
• Reconstruction error and Monte Carlo dropout
• Explainability layer linking OOD to states