Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : September 13, 2020 - September 16, 2020
In the past two decades, simultaneous localization and mapping (SLAM) has been actively studied and realized reliable navigation especially in unknown but static environments. Since it usually assumes a static map as a model of the environment, moving objects may disturb the accuracy of the localization and map. To overcome this issue, it is common to extract observations possibly coming from moving objects by comparing them between multiple samplings. In this paper, we propose a SLAM for dynamic environment which naturally resolves this problem. Based on our previous study, a generalized dynamic model for the map is introduced and SLAM is solved through optimization problem. It is shown that the proposed method captures the motion of the object moving along its own edge. Numerical simulations in the environment comprising static and dynamic objects as well as only moving objects validate the effectiveness of the proposed method.
