On Detection and Size Estimation of Cyber-Attacks Against Autonomous Systems

Sensor attacks pose a significant concern for autonomous systems. This paper focuses on the detection and estimation of sensor attacks using adaptive observers. By formulating the measurements of the autonomous system as a filtered version, which is then augmented into the system's state, we are able to not only detect the presence of sensor attacks but also estimate their size. The adaptive observer is implemented and integrated with the controller. To showcase the observer's capabilities, we conduct numerical simulations encompassing various scenarios. The simulation results demonstrate the observer's accuracy in estimating the system state, detecting sensor attacks, and quantifying the magnitudes of these attacks.