Vision-Based Relative Navigation Method For Formation Flight Of Micro Air Vehicles

Micro Air Vehicle(MAV) is a new kind of aircraft, which size is close to birds or insects. The small size brings a great hidden feature and cheap price, but also a poor load capacity and anti-interference ability. The formation flight brings MAV not only a better aerodynamic characteristics, but also a better co-operation capability. The using of vision-based relative navigation method provides a direct navigation information between the teammates. It not only be used in formation keeping and collision avoidance, but also compensates for the weak navigation ability of single MAV, by information interaction.In this paper, the methods of extracting relative position and attitude information from image are discussed.Firstly, according to the building of coordinate systems and the using of pinhole imaging model, the PnP (Perspective n-Points Problem) problem solving method is proposed to be used on MAV relative navigation. A fast algorithm is given. Then, a a high precise visual relative navigation simulation platform is designed, which is based on 3D modeling software 3ds MAX and the natural landscape modeling software Vue. The high realistic flight animations and pictures is generated. Thirdly, a new color gradient adaptive threshold image segmentation method is created, for extracting the relative position and orientation information from images. The navigation algorithms based on PnP solution and circle features are simulated and compared, an improved method is given. At last we discuss the best position of MAV formation, for the vision-based relative navigation. A criteria and selection factors are discussed and simulationed.