Presents an approach for comparing SLAM algorithms, which allows you to find the most accurate. The emphasis of the research is made on 2D-SLAM-algorithms, and 2D maps are considered, which is built after the algorithm is executed. This article presents three indicators for the evaluation of maps. The purpose of this article is to present methods for quantifying the quality of the 2D laser SLAM algorithm. The object of consideration is the resulting map, and by its visual evaluation it is possible to find out which algorithm creates the best map, i. e., to determine which map contains the least amount of noise, the most accurate walls, the lowest number of artifacts, etc. This article presents some indicators of quantification. The article also presents a map estimating framework. Maps, which are built by each algorithm considered, are compared with the use of several indicators, and for each launch an evaluation is assigned. Thus, it is possible to evaluate SLAM algorithms, even if the true map or trajectory is not provided for the data sets under consideration.

Authors: Ar. Yu. Filatov, An. Yu. Filatov, K. V. Krinkin, B. Chen, D. Molodan

Direction: Informatics and Computer Technologies

Keywords: SLAM comparison, map, trajectory, gmapping, Cartographer

View full article