Fault-Tolerant Preintegration Method for GNSS/MEMS-IMU/Quadcopter Dynamics Model Using Factor Graph Optimization
Xin Sun, Jizhou Lai, Pin Lyu, Bingqing Wang, Haoxin Tian · IEEE Sensors Journal · 2025
Accurate and reliable positioning is a prerequisite for quadcopters to perform tasks. Low-cost Micro-Electro-Mechanical Systems Inertial Measurement Unit is often combined with global navigation satellite system to provide accurate navigation parameter for quadcopters. Factor graph optimization has been popularized in navigation parameter estimation of quadcopters. The preintegration theory based on factor graph optimization is widely used in quadcopter positioning. However, Micro-Electro-Mechanical Systems Inertial Measurement Unit often fails in harsh environments, resulting in failure of preintegration, thereby reducing the navigation performance of quadcopters. Therefore, this article introduces the quadcopter dynamics model including a drag model and a thrust model, and combines Micro-Electro-Mechanical Systems Inertial Measurement Unit to form a fault-tolerant preintegration method. Different from previous preintegration, our proposed method switches two quadcopter dynamics model-assisted preintegration in real time according to the fault of the accelerometer. The fault-tolerant preintegration factor is combined with the factor of the global navigation satellite system to obtain the navigation parameter estimate. The proposed method is not affected by accelerometer failure, improves the robustness of quadcopter navigation and positioning, and reduces dependence on the environment. The actual flight test data results show that the proposed method can detect and isolate accelerometer faults in real time. In the case of accelerometer failure, the introduction of the improved quadcopter dynamics model improves the attitude, velocity, position accuracy by more than 34.6%, 20.8%, 33.9% respectively.