Vision aided strapdown inertial navigation system for autonomous landing of unmanned aerial vehicles

Abstract

An autonomous Unmanned Aerial Vehicle (UAV) must operate without human intervention, yet must meet the rigorous requirements associated with any airborne platform. To eliminate the need for human supervision, a UAV must be capable of carrying out at least three essential tasks autonomously: take-off, navigation and landing. Hence the proposed work is to design a vision aided Strapdown Inertial Navigation System for the autonomous landing of UAVs. The main objectives of this thesis are to describe the underlying theory behind the SDINS and vision based navigation in order to emphasize the advantages and limitations of both the systems and understand the need and significance of integrating them, to simulate the Strapdown Inertial Navigation System using quaternion and validate the same with real time data, and to validate the algorithm using the Flight gear simulation software. The major conclusions of the thesis are the body angular rates and accelerations for an aircraft from the six degrees of freedom trajectory is simulated using a kinematic approach and these angular rates and accelerometers are used in the error modeling of Micro Electro-Mechanical System inertial sensors. The advantage of this method is that there is no need for a camera calibration which means that the focal length of the camera lens and the mounting angles relative to the aircraft are not required, the most significant conclusion for this work is the demonstration of the viability of fusing imaging and inertial sensors for navigation, the technique is incorporated into an automated vision aided SDINS using Unscented Particle filter, the SDINS/Vision integration algorithm was implemented in MATLAB/Simulink and performance verified by interfacing the algorithm with XPLANE flight simulator, the position accuracy in terms of Root Mean Square (RMS) is improved by the integration of inertial and vision data, this development holds a great deal of promise to produce a lowcost, navigation-grade optical-inertial sensor for passive environments