The Study On Localization Methods Based On Home Service Robot

Home service robot is a special intelligent robot that can serve the people heart andsoul in daily family life. Home service robot has kinds of functions, assist the old people,take care of old people, do housework and entertain the family, for example. It makes moreand more people pay attention to the home service robot, and the research of home servicerobot is becoming a hotspot of the field of robots in recent years. To solve the localizationproblem of home service robot is the basis to complete all family life tasks successfully. Thesubject of this paper is the location research of home service robot though the use of thenatural features of home environment, and the main contents are as follows:1、In order to solve the problem of system modeling of home service robot andmodeling of experimental environment, in this paper, the motion model of home servicerobot system is built based on the experimental platform of the Voyager-II autonomousmobile robot system, the control input model of home service robot system is determined,the representation of environmental characteristics is defined, and the observation model ofsensor is built. And we make an analysis of the uncertain influences and the errors of systemmodeling of home service robot system, and elaborate the model of environmental noise andthe model of sensor observation noise. The complete building of the experiment models laysthe foundation for the research of localization problem of home service robot.2、Feature extraction and representation is studied by the natural features of homeenvironment, and the feature map of home environment based on the natural features ofhome environment is built. The laser sensor of home service robot is mainly used to detectthe edges and corners of the regular shape object in home environment, the detected edgesand corners of the regular shape object is regarded as natural landmarks when home servicerobot positioning in home environment, these natural landmarks are defined as featurepoints of natural features of home environment, and then, the feature map of homeenvironment can be built. At last, the effectiveness of the proposed method is verified by alot of experiments of Voyager-II autonomous mobile robot system.3、The theoretical methods of solving the localization problem of home service robotsystem are studied. Combined with the SLAM model of home service robot and the methodof feature extraction by the natural features of home environment, we use the extendedKalman filter (EKF) algorithm and unscented Kalman filter (UKF) algorithm to predict and estimate home service robot’s robot poses and the position of natural features of homeenvironment all time and update the feature map of home environment based on the naturalfeatures of home environment. In the theoretical calculations, the positioning error of usingthe UKF algorithm to position is smaller than the positioning errors of using the EKFalgorithm to position.4、Two kinds of home environment is supposed. One is a bedroom, also a small-scalehome environment, and the other is a bedroom and a living room, also a large-scale homeenvironment. In two kinds of home environment, home service robot conducts positioningsimulation experiment based on the built model, the method of feature extraction by thenatural features of home environment, EKF algorithm and UKF algorithm. The experimentverifies the effectiveness of using the EKF algorithm and UKF algorithm to solve thelocalization problem of home service robot system and the positioning accuracy of using theUKF algorithm to position is higher than the positioning accuracy of using the EKFalgorithm to position.