Lateral Inhibition Map Updating Method to Prevent Map Deformation for Simultaneous Localization and Mapping in Natural Environment

In this paper, a novel real time map updating algorithm using a concept of lateral inhibition was proposed and it was evaluated by simple 2D mapping experiment. For long time and difficult map creation process of simultaneous localization and mapping (SLAM) algorithm, stable map updating is one of the important topics. By using the lateral inhibition process to the map updating process, it is able to reduce the complexity of the map updating process of the SLAM. It could reduce the noise around the wall and emphasize the border (such a wall). The experimental result of map updating process of 6.0 x 3.5 m 2D room with many obstacles shows the x-y positional and the rotational estimation precision would be measured as x,y= 1.8 ×10-6 m (less than the LIDAR measurement error about 0.04 m) and the = - 1.9×10-2 deg (1 hour measurement), and the map was updated real time when the LIDAR sensor was moving from a position to another room position with low speed. After the 38 min outdoor experiment, the occupancy map was stably maintained, and there is also slight rotation and position estimation errors (x,y = 0.0 m, = 8.3 ×10-2 deg). The proposed method could suppress the wall existence probability distribution creeping induced by the natural environment vibration such as tree or leaves.