Abstract
Recently, it is demanded to detect temporal changes in large environment. Though a few years ago, only cameras was available for detection, the devices for measuring 3D shapes becomes common. 3D measurement devices enable us to detect temporal changes not only in 2D appearance but also in 3D shape. We propose a method for recognizing temporal changes and classifying the changes to clarify the reason from a pair of 3D data that are measured in some interval. We first apply the ICP-SLAM algorithm to each of the data independently to obtain the environmental map. Then we detect temporal changes by estimating the differences between the pair of the maps. Finally, we classify the type of changes by comparing the changed region and the range image from actual viewpoint. We confirm the effectiveness of the proposed method by evaluating the accuracy of the detection/classification through experiments.
