@techreport{oai:ipsj.ixsq.nii.ac.jp:00053418,
 author = {植田, 健治 and 石黒, 浩 and 辻, 三郎 and Kenji, Ueda and Hiroshi, Ishiguro and Saburo, Tsuji},
 issue = {7(1991-CVIM-076)},
 month = {Jan},
 note = {In this paper  we describe about a method to determine the robot position for fusing local maps into a global map for an unknown envirnment. We do not use internal sensor information such as acceleration  rotation velocity of wheels  and so on. We employ only visual information contained in a panoramic representation. The robot position is calculated in the coordinate of the global map obtained by fusing the range in formation observed from previous robot positions. The previous robot positions are already determined in the global map  and we choose one of them as a base point. The new robot position is obtained by calculating the relative motion parameters (rotation  derection and destance of translation) from the base point. We calculate the rotation and moving direction of the motion parameters from FOE. Then the moving distance is remained. Using those determined motion parameters as a constraint  we can restrict the candidates of the robot position along a line. We estimate each candidate by matching the new local map and the global map. The moving distance is determined by selecting the best candidate according to estimation value. Using neewly determined robot position  we fuse the local map into global map. We use this improved global map to next., In this paper, we describe about a method to determine the robot position for fusing local maps into a global map for an unknown envirnment. We do not use internal sensor information such as acceleration, rotation velocity of wheels, and so on. We employ only visual information contained in a panoramic representation. The robot position is calculated in the coordinate of the global map obtained by fusing the range in formation observed from previous robot positions. The previous robot positions are already determined in the global map, and we choose one of them as a base point. The new robot position is obtained by calculating the relative motion parameters (rotation, derection and destance of translation) from the base point. We calculate the rotation and moving direction of the motion parameters from FOE. Then the moving distance is remained. Using those determined motion parameters as a constraint, we can restrict the candidates of the robot position along a line. We estimate each candidate by matching the new local map and the global map. The moving distance is determined by selecting the best candidate according to estimation value. Using neewly determined robot position, we fuse the local map into global map. We use this improved global map to next.},
 title = {パノラマ表現を用いた環境地図の作成},
 year = {1992}
}