Vision Based Navigation For Unmanned Helicopter Landing

Unmanned helicopters (UH) are widely used in both military and civilian applications, and thenavigation technology for autonomous landing is a criticial technology for UH. Compared withtraditional navigation technologies, computer vision based navigation for automous landing an UHcan save manpower and material resources and also has the advantages of high precison andimmunity to electronic interferences.In this thesis, based on the navigation requirements for automatous landing an UH, a design forcomputer vision based lading navigation system is first proposed. In the design of the system, wedescribe a new artificial landmark used for the navigation, the hardwares of the vision basednavigation system, as well as the algorithmic flow of the navigation system, etc.Secondly，based on the concentric circles contained in the artificial landmark, landmark detectionand tracking method is proposed. This method first pre-processes the collected image to obtain abinary image. Based on the binary image, through edge detection and boundary tracking, all theboundaries in the image are extracted, and ellipse fitting are then conducted for each boundary to findall the coutours haiving ellipse shapes. All the ellipse shape contours are then further examined to findout the artificial landmark. Based on the Kalman filter, using the ellipse parameters as the states,tracking method for the artificial landmark is designed.Thirdly, based on the circle composed by three primary colours, the method for estimating therelative position and pose between UH and landmark is proposed. This method first builds theimaging geometry model for circular object, based on which, the position, roll angle, and pitch angleare estimated. Then, based on the distribution information of the different colours in the circle, themethod estimates the yaw angle. Finally, the ambiguity is eliminated, and the final results for positionand pose estimation are obtained.Lastly, simulation experiments and analysis are conducted for the proposed vision basednavigation method. The factors that affecting the precision of position and pose estimation are firstlyanalyzed through simulation experiments, and the conclusions can be used for the references ofdesigning vision based navigation system. Take autonomous landing an UH on a ship as an example,we use the method proposed in this thesis for navigating the UH landing, the related parameters of thevision based navigation system are determined based on the task, and then the simulation experiements of landing UH on a ship are conducted. The experimental results show that the visionbased navigation method for landing an UH proposed in this thesis can be effectively used in themission of autonomous carrier landing of UH.