Anomaly detection for generic failure monitoring in robotic assembly, screwing and manipulation

2025-09-30

AI Summary

Industrial robots frequently fail during assembly, fastening, and polishing tasks due to anomalies such as component misalignment or unexpected obstacles. Method: This paper proposes a general-purpose time-series anomaly detection framework applicable across diverse robotic manipulation tasks and control paradigms (position, impedance, and diffusion control). Leveraging multimodal force/torque signals, the method employs a lightweight autoencoder architecture optimized for both data efficiency and real-time inference. Contribution/Results: Experiments achieve AUROC > 0.93 on wiring and fastening tasks, enabling robust identification of typical failures; it also reliably detects severe anomalies in polishing. A systematic ablation study reveals how data efficiency, detection latency, and task dynamics jointly influence performance. To our knowledge, this is the first work demonstrating cross-task and cross-control generalization of a single anomaly detection model—establishing a transferable foundation for safe, robust autonomous assembly.

Technology Category

• Intelligent Robots: Manipulation
• Natural Language Processing: Safety and Robustness
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Out-of-distribution states in robot manipulation often lead to unpredictable robot behavior or task failure, limiting success rates and increasing risk of damage. Anomaly detection (AD) can identify deviations from expected patterns in data, which can be used to trigger failsafe behaviors and recovery strategies. Prior work has applied data-driven AD to time series data in specific robotic tasks, but its transferability across control strategies and task types has not been shown. Leveraging time series data, such as force/torque signals, allows to directly capture robot-environment interactions, crucial for manipulation and online failure detection. Their broad availability, high sampling rates, and low dimensionality enable high temporal resolution and efficient processing. As robotic tasks can have widely signal characteristics and requirements, AD methods which can be applied in the same way to a wide range of tasks is needed, ideally with good data efficiency. We examine three industrial robotic tasks, each presenting several anomalies. Test scenarios in robotic cabling, screwing, and sanding are built, and multimodal time series data is gathered. Several autoencoder-based methods are compared, evaluating generalization across tasks and control methods (diffusion policy, position, and impedance control). This allows us to validate the integration of AD in complex tasks involving tighter tolerances and variation from both the robot and its environment. Additionally, we evaluate data efficiency, detection latency, and task characteristics which support robust detection. The results indicate reliable detection with AUROC exceeding 0.93 in failures in the cabling and screwing task, such as incorrect or misaligned parts and obstructed targets. In the polishing task, only severe failures were reliably detected, while more subtle failure types remained undetected.

Problem

• Detecting anomalies in robotic assembly to prevent failures and damage
• Developing transferable anomaly detection across robotic tasks and control strategies
• Evaluating detection performance in cabling, screwing, and sanding tasks

Innovation

• Autoencoder-based anomaly detection for robotic tasks
• Multimodal time series analysis for failure monitoring
• Cross-task generalization with force/torque signal processing