Study On Pose And Control Strategy Of Quad-rotor Micro Air Vehicls

In modern warfare, more and more countries have started to use unmanned aerial vehicles for reconnaissance and military strike. With the technological development of microelectronics, MEMS and other related fields, unmanned aerial vehicles become more and more miniaturized, and thus gave birth to the concept of micro air vehicle (MAV). Different from conventional unmanned aerial vehicles, MAVs have small size, light weight and limited load, which make them unable to use high-precision sensors for attitude measurement and difficult to estimate flying pose accurately. In addition, the main work area of MAVs is indoor, tunnel and other near-Earth environment. It is difficult for MAVs to locate themselves in those chaotic and complex environments. In this thesis, pose estimation and control strategy is studied based on quad-rotor MAVs.Most quad-rotor MAVs use gyroscopes and accelerometers for attitude estimation. The corresponding algorithm assumes that accelerometers can accurately reflect the gravity component in flight. Based on traditional dynamics model of quad-rotor MAVs, the measurements of accelerometers along the pitch axis and horizontal axis will be zero in flight, and so they can not correctly reflect the gravity component. However, the actual flight measurement data indicate that the accelerometer measurements are not zero, but close to the low-pass filtered value of the gravity component. Based on the analysis of the dynamics of rotor, the rotor resistance is introduced to the traditional dynamics model. Under the model, the theoretical measurements of the accelerometer coincide with the actual measurements. Based on this model, an improved attitude estimation algorithm is proposed. Experiments show that the pose estimation accuracy of the improved algorithm is better than traditional methods. In addition, a method for speed estimation based on accelerometer is proposed.In GPS-denied environment, a method for speed and position estimation based on optical flow is proposed. The relationship between optical flow and flight speed is derived. The classical Srinivasan and Lucas-Kannde algorithms are compared. An optical flow calculation method combining feature matching and LK algorithm is proposed. The effectiveness of the optical flow algorithm is verified by autonomous hovering experiments.In order to simultaneous localization and mapping (SLAM) in environment with obstacles, a method based line structured light is used for environment measurement. The measurement principle is introduced. The factors affecting the measurement accuracy are analyzed and the accuracy is validated by experiments. A scan matching algorithm for data association in SLAM process is proposed based on the search of the solution space. A SLAM algorithm is designed based on extended Kalman filter (EKF) using optical flow method and line structured light method. Indoor flight experiments show that the algorithm is effective for position estimation and mapping.Most quad-rotor MAVs fly smoothly at low speed. To extend the maneuvering flight ability, the attitude control methods are introduced. Several flight modes, including straight flight, Looping flight, Longitudinal circumference flight and flip, are analyzed. The trajectory forms and control methods of the flight modes are studied. Smooth steering control method in the situation of high-speed straight flight is proposed.