There are several well-known techniques for SLAM: beginning with “classic” SLAM based on the Extended Kalman Filter and ending with modern graph-based algorithms. Our laboratory managed to implement its own SLAM algorithm that uses 3D laser scans as an input. The scientific novelty of this algorithm is explained by applying the Dempster-Shafer theory instead of Bayesian theory for representing the model of the cells composing the map. The classic approach involves each cell having a certain probability of being occupied, while Dempster-Shafer theory allows several masses to be stored in one cell and suggests formulae for updating these masses. The main idea of the project is to use the masses in the cells of the map to find the regularities in the map. These regularities might be represented as distinguishing borders of objects or formed areas that correspond to the whole object in the environment. The goal of this research is to find if these regularities exist and, if they do, to describe them.