RoboGuard: Enhancing Robotic System Security with Ensemble Learning

Abstract

Robots are becoming increasingly common in critical healthcare, transportation, and manufacturing applications. However, these systems are vulnerable to malware attacks, compromising reliability and security. Previous research has investigated the use of Machine Learning (ML) to detect malware in robots. However, existing approaches have faced several challenges, including class imbalance, high dimensionality, data heterogeneity, and balancing detection accuracy with false positives. This study introduces a novel approach to malware detection in robots that uses ensemble learning combined with the Synthetic Minority Over-sampling Technique (SMOTE). The proposed approach stacks three (ML models Random Forest (RF), Artificial Neural Networks (ANN), and Support Vector Machines (SVM) to improve accuracy and system robustness. SMOTE addresses the class imbalance in the dataset. Evaluation of the proposed approach on a publicly available dataset of robotic systems yielded promising results. The approach outperformed individual models and existing approaches regarding detection accuracy and false positive rates. This study represents a significant advancement in malware detection for robots. It could enhance the reliability and security of these systems in various critical applications