Reseach On Agricutual Robot Self-Positioning Tachnology Based On Machine Vision

Agricultural robot positioning technology is an important direction of robot filed, and it is crucial to implement agriculture machinery intelligentization. In fact, how to obtain the accurate robot pose information, is the prerequisite of robot intelligent navigation taskIn this thesis, a visual odometry positioning technology method that based on machine vision is studied. Main research contents and conclusions include::1. Firstly, we do a survey about robot positioning technology, and discuss the defects or shortcomings in various positioning methods. In this context, the visual odometry technique is introduced. Its principle%advantages, the research status at home and abroad as well as the technical framework are also discussed in detail.2. A new feature point extraction algorithm---SURF is introduced. Its principle is studied in-depth and realized in the VS2010software environment. In additional, SURF is compared with the Harris corner detection algorithm and SIFT algorithm. Many feature point extraction experiment in the7different scenes are done to improve the SURF algorithm’s advantage.3. The complex farmland environment of agricultural robot that wok in is discussed in this thesis, for example, illumination variation, tough terrain. Points at these difficulties and problems that maybe effect robot visual positioning accuracy, the corresponding measures are explores.4. A new stereo visual odometry system based SURF algorithm is presented. It only rely on visual input, and extract feature points as natural landmark from continuous image sequences, then match them to get the corresponding of the feature points, finally the LMedS robust algorithm, combined with least square method,is used to compute the agricultural robot precise position information. The tests show that the estimation error of visual odometry system is less than3.4%.This paper deeply studied agricultural robot visual position technology, and provided a feasible mean to achieve precise positioning of agricultural machinery.