Scholarlink

<p>Protecting autonomous systems from changing cyberthreats requires the use of Artificial Intelligence (AI) in robotics cybersecurity. Due to their widespread use in sectors including healthcare, manufacturing, and defense, robotics is becoming more and more susceptible to cyberattacks, which can include network invasions, hostile manipulations, and data breaches. The application of Artificial Intelligence (AI) in this field offers significant advantages over traditional cybersecurity methods by bringing advanced predictive analytics, automation, and adaptive learning capabilities. The rise of digital infrastructure has made robust cybersecurity crucial as cyberattacks continue to increase in complexity and frequency. This project focuses on the exploratory data analysis (EDA) and machine learning (ML) methodologies for detecting and predicting cyberattacks. IT-driven AI/ML models analyze OT data from sensors, cameras, and robot logs to improve robotic movement, problem detection, and efficiency. The aim is to understand patterns, enhance security measures, and improve threat detection capabilities. The project begins with a comprehensive EDA to uncover patterns and relationships in network traffic and attack behavior. This step includes analyzing distribution, trends, correlations, and outliers within the dataset to gain insights into the nature of various cyber threats, while OT devices must be protected from cyber threats like ransomware or unauthorized access [<CitationRef CitationID="CR1">1</CitationRef>, <CitationRef CitationID="CR2">2</CitationRef>]. Visualizations and statistical techniques are employed to highlight significant attack features and identify potential indicators of compromise. Algorithms like decision trees, random forests, and neural networks are used to create and train machine learning models after the EDA [<CitationRef CitationID="CR3">3</CitationRef>, <CitationRef CitationID="CR4">4</CitationRef>, <CitationRef CitationID="CR5">5</CitationRef>]. The goal of these algorithms is to accurately categorize and forecast various kinds of cyberattacks. Performance is evaluated using ROC-AUC, F1-score, precision, and recall metrics.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Adaptive artificial intelligence for securing cyber-physical robotic systems

  • 期刊论文

  • 发表日期：

    2026年4月14日

查看全文
  • A. Kishore Kumar,
  • M. Divya,
  • A. Murugarajan

摘要

Protecting autonomous systems from changing cyberthreats requires the use of Artificial Intelligence (AI) in robotics cybersecurity. Due to their widespread use in sectors including healthcare, manufacturing, and defense, robotics is becoming more and more susceptible to cyberattacks, which can include network invasions, hostile manipulations, and data breaches. The application of Artificial Intelligence (AI) in this field offers significant advantages over traditional cybersecurity methods by bringing advanced predictive analytics, automation, and adaptive learning capabilities. The rise of digital infrastructure has made robust cybersecurity crucial as cyberattacks continue to increase in complexity and frequency. This project focuses on the exploratory data analysis (EDA) and machine learning (ML) methodologies for detecting and predicting cyberattacks. IT-driven AI/ML models analyze OT data from sensors, cameras, and robot logs to improve robotic movement, problem detection, and efficiency. The aim is to understand patterns, enhance security measures, and improve threat detection capabilities. The project begins with a comprehensive EDA to uncover patterns and relationships in network traffic and attack behavior. This step includes analyzing distribution, trends, correlations, and outliers within the dataset to gain insights into the nature of various cyber threats, while OT devices must be protected from cyber threats like ransomware or unauthorized access [1, 2]. Visualizations and statistical techniques are employed to highlight significant attack features and identify potential indicators of compromise. Algorithms like decision trees, random forests, and neural networks are used to create and train machine learning models after the EDA [3, 4, 5]. The goal of these algorithms is to accurately categorize and forecast various kinds of cyberattacks. Performance is evaluated using ROC-AUC, F1-score, precision, and recall metrics.