1A1-G23 ローカルマップ統合によるユーザとモノの屋内位置推定

概要

論文の詳細を見る
We propose a new method that can simultaneously estimate the positions of both user and sensors attached to objects in indoor environments. The position estimation of them is achieved by hierarchically structuring local positioning and global positioning. In local positioning, local maps are automatically generated and updated continuously, in addition to estimating the user position simultaneously by utilizing the suitably selected log data for improving the accuracy of position estimation according to a reliability factor in the user model combined with SLAM. In global positioning, the global map is automatically generated by integrating local maps, focusing on the common sensors which exist on two or more local maps and the distances (which are weighted by likelihood on the local map) between sensors. Advantages of the proposed method are (1) calibration free, (2) dynamic extension of covered area. Evaluation results show that estimation errors were less than 400mm in both user and sensors positions, and small enough to develop context-aware indoor service taking into account the personal space.
一般社団法人日本機械学会の論文