Environment Mapping Algorithm Using Semantic Description and Constrained Delaunay Triangulation

  • Daniel Figurowski
  • Pawel Dworak
Keywords: Mobile robot, Environment mapping, Semantic segmentation, Computational geometry

Abstract

The article discusses the issue of creating an environment’s map with semantic division of areas. This simplifies some aspects of mobile robot control and permits the robot to carry out tasks issued in a human-understandable form. The article presents an innovative algorithm for mapping the environment by a mobile robot. Its novelty stems from the use of a semantic description of the world and Delaunay triangulation method with constraints. This description is used to segment the map of the environment and to reduce redundant information. The developed algorithm is based on the idea of extending the borders of the already discovered areas; they are expanded as the new data are collected from the environment. These data contain information about the location of semantic types in the explored space and is used to update the areas during analysis of the environment’s map. The beforementioned triangulation method is used in that process. The performance of the proposed algorithm is tested in simulation studies. The obtained results show a good computational efficiency of the method, which is crucial in the problem of environment exploration by mobile robots with limited computational resources.

Published
2019-12-06
Section
AUTOMATION, ROBOTICS