Model-Based Fail-Safe Module for Autonomous Multirotor UAVs with Parachute Systems

This paper presents development of model-based fail-safe modules for autonomous multirotor Unmanned Aerial Vehicles (UAVs) with safety parachute systems. The module is based on the adaptive eXogenous Kalman filter for actuator fault diagnosis. We assume all states can be measured, such that the primary goal of the filter is not the state estimation from the measurements, but the accurate reconstruction of the multirotor dynamics in real-time. Numerical simulations show the proposed diagnostic filter can be used to estimate the magnitude of the actuator faults accurately. Furthermore, based on simulated and real data recorded during a hexacopter UAV flight when its actuators experiencing complete failure, the experiment results demonstrate the effectiveness of the approach.