Saman Yazdannik - Faculty of Aerospace, K. N. Toosi University of Technology, Iran
Shamime Sanisales - Faculty of Aerospace, Malek Ashtar University of Technology, Iran
Morteza Tayefi - Faculty of Aerospace, K. N. Toosi University of Technology, Iran
Reza Esmaelzadeh - Faculty of Aerospace, Malek Ashtar University of Technology, Iran
Mostafa Khazaee - Faculty of Aerospace, Malek Ashtar University of Technology, Iran

This paper presents a comprehensive framework for enhancing the safety and reliability of quadrotor UAVs by integrating second-order sliding mode control (۲-SMC) and an advanced anomaly detection and prediction system based on machine learning and AI. The paper addresses the challenges of designing controllers for quadrotors by proposing a novel sliding manifold approach divided into two subsystems for accurate position and attitude tracking. The paper also provides a detailed analysis of the nonlinear coefficients of the sliding manifold using Hurwitz stability analysis. It demonstrates the effectiveness of the proposed method through extensive simulation results. To further assess the safety and reliability of the quadrotor, an anomaly detection and prediction system is integrated with the position and attitude tracking control. The system utilizes machine learning and AI techniques to identify and predict abnormal behaviours or faults in real time, enabling the quadrotor to quickly and effectively respond to critical situations. The proposed framework provides a promising approach for designing robust and safe controllers for quadrotor UAVs. It demonstrates the potential of advanced machine learning and AI techniques for enhancing the safety and reliability of autonomous systems.

Anomaly Detection, auto-encoder, fault detection, Machine Learning, Quadrotor UAVs, safety, second-order sliding mode control (۲-SMC)