Robust Estimation and Sensor Fault Management Using Probabilistic Voting Algorithm in UAVs

This article presents a fault-tolerant estimator using a probabilistic voting algorithm (PVA) for the safe maneuvering of multirotor unmanned aerial vehicles (UAVs). UAVs are widely utilized in numerous applications, but any malfunction can lead to secondary accidents. The safety and robustness of the UAV component should be guaranteed to minimize fatal accidents during flight. A flight control computer (FCC) with various sensors is one of the most important components, the robustness of which should be guaranteed. In this article, a hybrid FCC including both hardware and software is developed to improve the robustness and safety of the FCC by both hardware and analytical redundancy. Triple modular FCCs for hardware redundancy are utilized to deal with various faults. The PVA is designed to estimate the reference state of the UAV and make the consensus to select the fault-free FCC by the fault probabilities of each state measurement from the FCC estimators. Moreover, multiplexers (MUXs) switch the FCC channel based on the consensus result to compensate for faults. Then, the fault identification algorithm identifies the source of the estimator faults by information on the residual signals between the estimated states and the sensor measurements. The PVA is validated through numerical simulations and experiments. This method achieves approximately a 93% correct detection rate and a fault detection time of less than 1 s, which is sufficient to maintain the dynamic responses of the UAV. These results show that the PVA improves and ensures the safe maneuvering of the UAV in various fault situations.