Identification of spatial objects by their monochrome images of autonomous mobile robots
Abstract
The method of image recognition of spatial objects of autonomous mobile robots is presented in this paper. This method is based on equations system in which the changes of the inertia moments depend on the change in the size and rotation of the image of a plane figure. The results of computer simulations confirming the effectiveness of the proposed method are carried out.
Keywords
Download Options
Introduction
Artificial intelligence technology has always been closely linked with robotics, whose main objective is the creation of robots that can act like a man. Besides they should have certain intelligence elements, mobility and ability of autonomous regime to perform the operations for the achievement of this goal. When they can be used in emergency situations during the rescue operations in the chemical, oil and gas and other industries in the fight against terrorism, protection of the most important objects and other extreme cases, in space researches, in the condition of strong radio interference, the radio shadow and away from the command post, then it is dangerous for a man to be near or inside the object. In addition, during high-risky situations people often make mistakes that reduce the effectiveness of the work. Robotics also allows to eliminate human error [1, 2].
Conclusion
As it is seen from the table, only the corresponding standard images identified by the smallest difference do not exceed permissible threshold. In other cases, it greatly exceeds the permissible threshold. Besides, there may be some cases of nonintersection graphs constructed by the formulas (7) and (8). This indicates that the recognizable images do not match the standard ones.
Thus, two conclusions can be drawn. First, a method for identifying monochrome images of the single and multi-piece plane figure subjected to affined transformation (which consists of a non-uniform scale and image rotation in the projection plane) considered in the paper, allows us to make the identification process on the model retaining sufficient certainty. Second, considering in the paper the method for carrying out the identification process does not require multi-processor systems, or special software.