An IMU-Only Attitude Solution and SLAM Loosely Coupled System for UAV Localization in Camera-Failure Scenarios during Nuclear Rescues

In search and rescue missions, UAV localization largely relies on visual-inertial SLAM methods, whose accuracy depends on image quality. However, in nuclear accident scenarios, high-energy particles can damage the camera's photosensitive elements, leading to the loss of camera images in the most severe case. This paper proposes an IMU-only attitude solution and SLAM loosely coupled system for auxiliary UAV localization when the camera fails. The system employs a two-step complementary filter for the IMU-only attitude solution, and optimizes the reinitialization process of the open-source SLAM system Vins-Fusion to ensure rapid switching between these two threads. To ensure the accuracy of the IMU-only attitude solution, a low-pass filter is used for gyroscope bias compensation, a state machine is employed for magnetic interference resistance, and the fourth-order Runge-Kutta method is utilized for position estimation. Experimental results show that the algorithm can accurately and stably localize the UAV when the camera fails.