Multi-Rotor Aerial Robot Visual Localization, Environmental Reconstruction And Nonliner Control

In recent years,the UAV(unmanned aerial vehicle)has gradually become aerial robot which have self localization flight capabilities.The research on aerial robot can be divided into two parts: one is the research of aerial robot autonomous positioning based on machine vision;the other one is flight control design.However,with the requirement of miniaturization of aerial robots,a more compact and maneuverable configuration of UAV is needed,such as the tilt tri-rotor UAV described in this thesis.To improve the autonomous control ability for small aerial robots,this thesis has obtained the following achievements:1.Based on the ORB-SLAM,an onboard visual SLAM method is developed to calculated the pose and position of the aerial robot from the images obtained via onboard camera.Ground station can receive the images from aerial robot through the Wi-Fi protocol and convert it into three-dimensional dense map.To testify the proposed visual SLAM methodology,a flight test bed is developed which consist a quad-rotor UAV with a RGB-D camera as the main onboard visual sensor.Several indoor and outdoor experiments have been performed to verify the performance of the proposed algorithm,and an positioning accuracy of 5 cm has been achieved.The reconstruction has no obvious mismatch and tearing phenomenon.2.According to the principle of IMU pre-integration,this thesis present a novel visual and inertial fusion localization algorithm.The state variables of IMU and visual odometry are optimized by using nonlinear optimization algorithm,and the optimal estimation of the position and attitude of the aerial robot is obtained.Experiments on EuRoc datasets show that the proposed algorithm can improve the positioning ability of the aerial robot in complex and fast motion environments.3.A new nonlinear robust control strategy based on the super-twisting algorithm is proposed to control the tilt tri-rotor UAV which is subjected to unknown inertial tensor and unknown external disturbances.Lyapunov based stability is employed to prove the stability of the closed loop system,the finite-time convergence of the attitude control error is also achieved.The proposed control strategy is validated on the self-built tilt tri-rotor UAV test bed.