Robustness Study Of UAV Localization Method Based On Multi-Sensor Fusion

The UAV localization problem is the core problem for its ability to fly autonomously.With the continuous improvement of hardware arithmetic power such as GPU,it becomes possible to realize large-scale positional optimization problems,and multi-sensor information fusion technology becomes a research hotspot and an important development direction for localization algorithms.Especially in indoor flight scenarios,achieving locally accurate and globally drift-free state estimation of UAVs is an urgent problem for current localization algorithms.This study is oriented to the UAV multi-sensor fusion positioning technology,using a quadrotor UAV as a platform,and based on the original VIO(Visual-Inertial Odometry)positioning algorithm,the innovative fusion of laser ranging radar sensor information enables the UAV to obtain accurate and robust positional information in indoor environment,which improves the UAV’s autonomous flight capability in GPS-free scenarios.The main research contents of the paper are as follows.1.build the hardware platform with the application scenario of indoor flight of quadrotor UAV.The power system parameters were identified and analyzed using the power test platform;the power system model,dynamics model and kinematic model of the quadrotor UAV were established;the stabilization controller of the UAV was designed with the string-level PID algorithm as the core control strategy,and the PID parameters were rectified.Finally,the joint Simulink and Flight Gear software in-loop simulation experiments are used to verify that the stability controller performance meets the UAV indoor flight requirements.2.In order to make the visual IMU location algorithm operate stably,the internal reference of the built-in IMU of the D435 i camera,the internal reference of the stereo camera,and the external reference of the camera and IMU are calibrated in turn.3.In order to solve the problem of accumulated errors in the visual IMU location algorithm,the laser ranging radar sensor is fused.The Distance-VIO algorithm is developed based on the VINS-Fusion algorithm,which incorporates displacement constraints in the altitude direction in a nonlinear optimization system.Finally,the developed algorithm was validated on an open-source dataset,and it was verified that the algorithm showed a significant improvement in positioning accuracy and robustness.4.A real experiment scenario was built,the assembly and debugging of the experimental prototype was completed,the stable controller designed in the thesis was deployed on the experimental prototype flight control board,and the communication links between the laser range sensor,flight control system,MAVROS,and the ROS modules were established.Finally,a real aircraft indoor flight test experiment was conducted in an indoor scenario.The experimental results show that the Distance-VIO algorithm effectively avoids the phenomenon of cumulative error in altitude direction and can provide centimeter-level altitude information in real time,which provides a theoretical method and technical support for UAV positioning technology based on multi-source sensor fusion.