SLAM Technology Research Based On Home Service Robot

The SLAM technology of home service robot has become a hot issue in the field ofrobotics research. An excellent characters of autonomy is essential for robot to accomplishtasks independently, particularly when operating under unknown environments. Thelocalization problem is a key issue for robot self-navigation, and simultaneous localizationand mapping is of great significance for a home service robot since the self-localizationaccuracy depends on the accuracy of environment map. Our study focused on the Voyager IIrobot, which carries the data obtained by the laser sensor, as the research object; andfinished self-localization and map building simultaneously.Firstly, this paper introduced the Voyager II robot platform and confirmed theperformance of laser sensor through two experiments. Our research defined three kinds ofcoordinates; built mathematical models for robot motion control system and laser sensor,respectively; established the local map through area divided (scanned by laser sensor), linesegment distill and line segment fitting procedures. During the section of line segment distill,we applied Hough transform and iterative segmentation method, and demonstrated thesuperiority of the latter. Meanwhile, we performed dynamic threshold method for theiterative segmentation and achieved good results.Secondly, this paper utilized “line segments rang” method to improve the search speedof relevant lines. We also introduced EKF and UKF algorithm and tested the performance ofthe two: both algorithms could make the robot get a very accurate pose and UKF has ahigher accuracy compared with EKF algorithm.At last, the line segments of local map were divided into two classes: globalmap-related and global map-unrelated. By selecting the line segments with correspondinganalyzing conditions, we achieved the integration of local map and global map, furtherrenewed the global map. We finally established the global map of our laboratory aisle andverified the accuracy of the map.