METHOD OF CONSTRUCTING A FLAT MAP OF THE SURROUNDING SPACE FOR A MOBILE ROBOTIC SYSTEM

Abstract

The paper proposes a method of building a local map of a mobile robot, which allows determining the position of objects in space with the help of a laser range finder. In addition, the accuracy parameters of the specified method of building an environment map were analyzed, which allows for the formation of requirements for hardware for controlling a mobile robot.

It is assumed that the initial map is known and during the work (the movement of the robot) it is only clarified and supplemented. The main sensors used are odometers (wheel speed sensors) and a laser scanning range finder (lidar), which measures the distance to obstacles in a plane parallel to the plane of the robot's movement. Scanning rangefinders in machine vision systems form in this case a two-dimensional picture of the surrounding space. The final result of laser scanning is the determined spatial coordinates of the points of obstacles. The collection of laser beam reflection points forms an irregular grid with a large number of such points. According to these data, a digital model of the surface space in the form of a regular grid can be obtained by a mathematical method, after camera refinement.

Insurmountable obstacles mean only projecting obstacles that fall into the field of view of the rangefinder (that is, have a height not less than the height of the rangefinder installation). It is proposed to sort the measured current points on a known map. Changes may have occurred since the creation of the v-home map (the map may be out of date). It is necessary not only to make these changes to the map, but also not to miss the old known obstacles, which are remeasured in the process. The proposed method of constructing the map works well only in a normal environment, that is, in the case when the angular dimensions of obstacles are several times greater than the angular resolution of the rangefinder. In the presence of a large number of small obstacles, the speed of the method drops sharply and errors during map construction increase. The great advantage of the method is that the number of saved points is always known - it is the number of curve segments multiplied by three.