Research On Localization And Mapping Of Service Robot Based On Intelligent Space

Simultaneous Localization and Mapping (SLAM) technology of service robot in family environment is the advanced and key issue of robot navigation technologies. It is also one of the basic research issues of the home service robot. Family environment is usually messed up by all kinds of objects. It is difficult for the service robot to realize simultaneous localization and mapping quickly and accurately. The combined technology of intelligent-space based SLAM can be used to solve the problem efficiently. The corresponding study works are summarized as follows:(1) The research status and related basic technologies of intelligent space and SLAM was analyzed. An intelligent-space based SLAM scheme and its basic steps were proposed according to the general model and present problems of SLAM, and also the technical difficulties and key issues encountered were introduced in detail.(2) A three-dimensional mixed feature map of environment is built in order to describe the environment in more detail and to speed up the SLAM algorithm. First, environment features can be gotten by extracting Harris corners of the objects in the environment through binocular stereovision sensor. Then, three-dimensional geometric information of these corners can be gotten by stereo matching algorithm, and the corresponding information of image features can be gotten as well. Finally, the mixed information is bound up and deposited into information database of intelligent space as real-time updating information, and then the mixed features map was constructed.(3) The technical practice of intelligent-space based SLAM. The kinematic model and observation model of robots were created, and they were reconstructed in order to realize linearity. Robot can interact with intelligent space real-timely, and it can realize fast data association based on the image features retrieved from mixed features map. Then by processing the uncertainty information through Kalman Filter Algorithm, the robot pose can be estimated, and also the information was saved to update the map.(4) Recognition-based localization method of service robot in the family environment was studied. Two different methods of objects recognizing were adopted: the dissimilar objects can be recognized by the SIFT feature matching algorithm; while the similar objects can be recognized by using the auxiliary artificial landmark. Based on the recognition of objects, the service robot can retrieve their properties from information database of intelligent space, and estimate its own pose by the prediction algorithm.