Research On Indoor Autonomous Navigation For Quadrotor UAV Based On Multiple Sensors

In disaster rescue, unmanned aerial vehicles(UAVs) gradually show more and more advantages. In recent years, UAV navigation based on GPS or SLAM has greatly progressed. However, how to navigate UAV autonomously in an indoor environment that is devoid of obvious physical cues becomes a new challenge. This thesis uses quadrotor UAV control technology and data fusion of multiple sensors to realize autonomous navigation, which has high scientific and application value.Firstly, state and motion model of UAV are established based on Bayes filter and aerodynamics theories. The delays of data processing and communication between UAV and ground station are eliminated when states are predicted. The posture from inertial measurement unit (IMU) is used to transform the perspective of the image. The displacement of image is computed by optical flow algorithm. Then the actual distance is calculated by using scale transformation and fusion of optical flow and IMU data. Experiment results show that this method has high positioning accuracy for a short time.Secondly, data association algorithm is used to look for optimal transformation between local and global map. The global position is computed by using Monte Carlo localization method. In order to resolve the problem of particle diversity reduction, several methods are presented, such as reducing weight variance, monitoring changes in weight. The experiment results show that this method has strong robustness to meet the needs of long time localization.Thirdly, the hardware and software of AR.Drone quadrotor are improved. The navigation and control system on ground station are developed based on Robot Operating System (ROS). The control system uses PID controller with limiter, and reduces the impact of the delay. Navigation system combines Bayes Filter and Monte Carlo localization, which enables the system to locate UAV for short and long time.Finally, in order to verify the navigation algorithms and techniques, a simulation system is established based on Gazebo. The navigation algorithm is tested in the simulation environment The simulation results show that the system meets the needs of indoor UAV localization in timeliness, accuracy and robustness.