用机器学习分析机器人日志，自动发现异常行为。
Detection of Anomalous Behavior in Robot Systems Based on Machine Learning
- 对比多种模型，用日志数据识别机器人异常
- 自编码器在复杂场景下准确率最高
- 适合关注机器人安全与故障检测的研究者
确保机器人系统的安全可靠运行至关重要，可防止潜在灾难并保障人身安全。尽管设计严谨，系统仍可能因故障引发安全隐患。本文提出基于机器学习的系统日志异常检测方法，使用CoppeliaSim采集两种场景下的日志数据，并对比评估了逻辑回归（LR）、支持向量机（SVM）和自编码器（Autoencoder）等模型。实验在四轴飞行器（Context 1）和Pioneer机器人（Context 2）场景中进行。结果表明，在Context 1中LR表现最佳，而Autoencoder在Context 2中效果最显著，说明最优模型选择依赖具体应用场景，可能源于不同平台异常模式的复杂性差异。研究强调了比较方法的价值，并展示了自编码器在检测复杂异常中的优势。
原文摘要 · Abstract (English)
Ensuring the safe and reliable operation of robotic systems is paramount to prevent potential disasters and safeguard human well-being. Despite rigorous design and engineering practices, these systems can still experience malfunctions, leading to safety risks. In this study, we present a machine learning-based approach for detecting anomalies in system logs to enhance the safety and reliability of robotic systems. We collected logs from two distinct scenarios using CoppeliaSim and comparatively evaluated several machine learning models, including Logistic Regression (LR), Support Vector Machine (SVM), and an Autoencoder. Our system was evaluated in a quadcopter context (Context 1) and a Pioneer robot context (Context 2). Results showed that while LR demonstrated superior performance in Context 1, the Autoencoder model proved to be the most effective in Context 2. This highlights that the optimal model choice is context-dependent, likely due to the varying complexity of anomalies across different robotic platforms. This research underscores the value of a comparative approach and demonstrates the particular strengths of autoencoders for detecting complex anomalies in robotic systems.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。