A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an\n LSTM-based Variational Autoencoder

2017/11/02 by Daehyung Park, Park, Daehyung, Yuuna Hoshi +3 · 20 citations
Computer Science · Engineering · #Anomaly Detection Techniques and Applications #Human Pose and Action Recognition #IoT-based Smart Home Systems

paper · pdf · doi:10.48550/arxiv.1711.00614

Abstract

The detection of anomalous executions is valuable for reducing potential\nhazards in assistive manipulation. Multimodal sensory signals can be helpful\nfor detecting a wide range of anomalies. However, the fusion of\nhigh-dimensional and heterogeneous modalities is a challenging problem. We\nintroduce a long short-term memory based variational autoencoder (LSTM-VAE)\nthat fuses signals and reconstructs their expected distribution. We also\nintroduce an LSTM-VAE-based detector using a reconstruction-based anomaly score\nand a state-based threshold. For evaluations with 1,555 robot-assisted feeding\nexecutions including 12 representative types of anomalies, our detector had a\nhigher area under the receiver operating characteristic curve (AUC) of 0.8710\nthan 5 other baseline detectors from the literature. We also show the\nmultimodal fusion through the LSTM-VAE is effective by comparing our detector\nwith 17 raw sensory signals versus 4 hand-engineered features.\n

Cited by

• Soft-Constrained Optimization of Latent Space in Variational Autoencoders
• Domain-Prior-Regularized Graph Modeling for Anomaly Detection in Cyber-Physical Systems
• Opportunities and Challenges for Data Quality in the Era of Quantum Computing
• A Problem-Oriented Taxonomy of Evaluation Metrics for Time Series Anomaly Detection
• Dynamic Reward Scaling for Multivariate Time Series Anomaly Detection: A VAE-Enhanced Reinforcement Learning Approach
• Reasoning-based Anomaly Detection Framework: A Real-time, Scalable, and Automated Approach to Anomaly Detection Across Domains
• CLEANet: Robust and Efficient Anomaly Detection in Contaminated Multivariate Time Series
• Out-of-Distribution Detection for Safety Assurance of AI and Autonomous Systems
• Causal Disentanglement Learning for Accurate Anomaly Detection in Multivariate Time Series
• Moon: A Modality Conversion-based Efficient Multivariate Time Series Anomaly Detection
• Anomaly detection for generic failure monitoring in robotic assembly, screwing and manipulation
• ScatterAD: Temporal-Topological Scattering Mechanism for Time Series Anomaly Detection
• Robust and Resilient Soft Robotic Object Insertion with Compliance-Enabled Contact Formation and Failure Recovery
• AnoF-Diff: One-Step Diffusion-Based Anomaly Detection for Forceful Tool Use
• Leveraging Intermediate Representations of Time Series Foundation Models for Anomaly Detection
• Detection of Anomalous Behavior in Robot Systems Based on Machine Learning
• DQS: A Low-Budget Query Strategy for Enhancing Unsupervised Data-driven Anomaly Detection Approaches
• Enhancing Video-Based Robot Failure Detection Using Task Knowledge
• Manifold-regularised Large-Margin ℓ_p-SVDD for Multidimensional Time Series Anomaly Detection
• Time-RA: Towards Time Series Reasoning for Anomaly with LLM Feedback

Related

• CLUE-AI: A Convolutional Three-stream Anomaly Identification Framework for Robot Manipulation
• What went wrong?: Identification of Everyday Object Manipulation\n Anomalies
• Multimodal Anomaly Detection with a Mixture-of-Experts
• Unsupervised anomaly detection for a Smart Autonomous Robotic Assistant Surgeon (SARAS)using a deep residual autoencoder
• Multi-Modal Anomaly Detection for Unstructured and Uncertain Environments