Mr. Daisuke Kakuma mainly conducts this research.

Semantic information about the environment lets users operate a robot using natural language. This is a comfortable way to make a robot do tasks in daily-life environments. Because a floor map in a building includes semantic information, if the robot can relate its own map to the floor map, the robot can access this information. The robot obtains the map of the unknown place by using simultaneous localization and mapping (SLAM). In this project, we propose a method using graph matching to align a floor map with an occupancy grid map generated by SLAM. Our experimental results verified that this method can perform a gross alignment of the maps in an actual situation. To verifying a robot can use the floor map information, I made a robot navigate by setting the goal position on the floor map aligned to the occupancy grid map.

The result shows the alignment of a floor map and an occupancy grid map. As assigning much environmental information to the occupancy grid map, the floor map can be aligned.

After the alignment, we can designate the destination using the floor map.

Reference

Daisuke Kakuma, Satoki Tsuichihara, Gustavo Alfonso Garcia Ricardez, Jun Takamatsu, and Tsukasa Ogasawara, “Alignment of Occupancy Grid and Floor Maps using Graph Matching”, in Proc. of the 11th International Conference on Semantic Computing, pp. 57-60, 2017.