This paper presents a method for advanced faulttolerant control (FTC) of multirotor unmanned aerial vehicles (UAVs), which includes anomaly detection on sensor measurements, fault estimation on actuators, and a robust model predictive control (MPC). To detect anomalies on the sensor measurements, an Echo State Network is used. System states and faults are estimated using an adaptive extended Kalman filter. The system is further controlled using MPC. The method is tested in numerical simulations with a hexacopter dynamic model. Simulation results show the ability of the FTC to handle failure with different even and uneven actuator faults.

Echo State Network (ESN) is one of machine learning methods that can be used to detect anomalies in sensor readings. The method predicts output signals, from which a prediction error can be created. To enable faulttolerant control, ESN needs to be combined with a robust fault estimation method. Indeed, identifying the source of the faults, whether coming from sensors or actuators, is crucial in safety-critical Unmanned Aircraft Systems (UAS), since it will determine proper control actions when the faults occur. This paper presents a novel method to combine sensor anomaly detection using ESN with actuator fault estimation using adaptive extended Kalman filter (AEKF). Numerical results show the benefit of using the cascaded algorithm in a noisy environment. Furthermore, the presented method is validated using a hexacopter with actuator faults in indoor experiments.

In this paper, we present contact point and surface normal estimators for robotic applications with flexible tools. The estimators rely on state information of a flexible tool model; this information is obtained from an unknown input observer. The observer uses force and torque measurements at the root of the flexible tool to estimate the deflection of the tool although the force applied to the tip of the tool is unknown. The flexible tool is modeled with a finite element approximation of an Euler-Bernoulli beam model including contact forces between the flexible tool tip and the environment.The unknown input observer provides estimates of the contact point between the flexible tool and the rigid environment in addition to the contact force. This information is subsequently used to estimate a surface normal of the environment. The estimators can be deployed together with an adaptive parallel position/force controller to ensure tracking of position and force references for the tip of a flexible tool.The proposed estimation algorithm is verified in simulation and validated in real robot experiments. The method enables accurate force and position tracking in addition to adaptation to the surface geometry.