Design And Simulation Of A Visual Navigation System For Small-scale Unmanned Helicopter Autonomous Landing

The small-scale unmanned helicopters have very important military, civil and scientific research value. Autonomous landing is one of the main flying mission phases, and it's also a phase full of accidents. Compared with other phases, it requires the navigation system not only to provide high precision parameters, real-time performance and reliability, but also off-land distance as much as possible, in order to prevent the helicopter from hitting the ground too strong which leads to an accident. As a result, design a navigation system suitable for autonomous landing not only has important research value, but also of great practical significance.This dissertation takes SJTU "Sky-walker" small-scale unmanned helicopter as the main application target, and designs a visual navigation system based on computer vision which meets the special requirements. Since lack of the ability to carry out actual flight experiments, and actual flight experiments have high risks, simulation is used to design and implement the visual navigation system in MATLAB-Simulink platform after the position and attitude measurements are carried out. And it will also assess the performance of the navigation system, such as feasibility, accuracy, robustness and etc. The main contents include the following two aspects: 1) Position and attitude measurements based on the principle of computer vision: Design landing platform graph and its image is taken by a digital camera. Extract feature points hidden in the image through image analysis and processing, and finally calculate the camera's position and attitude parameters relevant to the landing platform.2) Simulation of a visual navigation system for small-scale unmanned helicopter autonomous landing: Design camera model and its cradle head model, and make these models much closer to the reality by introducing random error. Autonomous landing simulation platform based on visual navigation system is designed, which is also combined with autonomous flight control algorithm. The navigation system's performance is also assessed. In addition, VRML is used to set up a simulation scenario, which makes the results more intuitive and lively, and also enhances the system's expressive force.The visual navigation system used for autonomous landing is designed and verified in the form of simulation. Since there are some differences between simulation and real environment, further research and exploration are needed in order to push this system to be a practical application.