Research Of Obstacle Detection And Path Planning For Unmanned Aerial Vehicle

The small scale unmanned helicopter is a very useful platform for military,civil and scientific application.It plays a vital role in reconnaissance,aerial photography, sampling and monitoring nuclear and chemical area and has a board application prospects.As stated above,research was carried out on detecting obstacles of the unmanned helicopter on its flight in order to take missions in lots of environments which have different obstacles to avoid.This paper focuses on the two images taken by two ordinary camera placed on the helicopter and was achieved via five discrete steps:image pre-processing,feature detection,feature match,rebuild 3-D information and obstacle detection.Gaussian filter was used for removing noise in the pre-processing section.Harris seeds was used to detect edge and point as the feature,relaxation method based on regional related match for matching features in order to get the 3-D information of the features,then layered space algorithm is used to detect and avoid obstacles.Finally,the system will present the list of obstacles and their locations with output image,and give the flight command to avoid obstacles.According to the experience,the obstacles'3-D information is same with actual ones and errors in an acceptable range,which shows that the obstacle detection based on binocular vision for small scaled unmanned helicopter has a good reliability and validity.