Vision-Based Fast SLAM For Small Scale Unmanned Helicopter

With the rapid development of various advanced integration technology,unmanned helicopter, as a kind of cheap aircraft which is controllable andmultipurpose as well, has been widely used in military and civilian field. Navigationsystem is a localization system to guide the airplane to fly safely, which plays animportant role in the autonomous flight of unmanned helicopter. It mainly learns theairframe’s position and posture through the outside information. For the successfulapplication of SLAM algorithm in the mobile robot navigation in recent years, agood solution is provided for the low cost, light and small scale navigation ofunmanned helicopter.The common SLAM algorithm can’t be used in large-scale outdoorenvironment for the high computation. The SPS-UFASTSLAM algorithm, based onparticle filter and Spherical simplex unscented Kalman filter, is proposed in thispaper. On the basis of UFastSLAM algorithm framework, this algorithm usesSpherical simplex unscented Kalman filter instead of Unscented Kalman filteringprocess, which reduces the computation and improves the running time of thealgorithm. By the experiment, it is proved that when the location accuracy is almostthe same, SPS-UFASTSLAM algorithm improves significantly in operation speedcompared with UFASTSLAM.