SCENE ANALYSIS IN MOBILE INFORMATION SYSTEMS ROBOTIC COMPLEXES

Abstract

Modern robots are capable of performing increasingly complex tasks that usually require a high degree of interaction with the environment in which they have to work. As a result, robotic systems must have deep and specific knowledge about their workspaces, which go far beyond the simple representation of indicators that a robotic system can create using visual data processing techniques, for example, in the task of simultaneous localization and mapping (SLAM). Scene analysis is the link between object recognition and knowledge about the world around us and is present in one form or another in the process of extracting information from visual data necessary to solve a specific task. The article presents a systematic approach to providing on-board STZ analysis of the scene. The technologies of scene analysis are considered as an integral part of increasing the degree of autonomy of mobile RTCs. A number of technologies have yet to be mastered and implemented, but the overall structure allows you to gradually deepen the analysis of the scene on board the RTK, thereby increasing the degree of autonomy without radically redesigning the on-board information management system and STZ, as a key part of information support. The information extracted from the visual data is integrated into a multi-layered map, providing a high-level representation of the environment, which embodies the knowledge necessary for a robotic complex to actually perform complex tasks. A multi-layered map is a form of storing knowledge about the environment and the objects in it. This map combines a spatial hierarchy of objects and places with a semantic hierarchy of concepts and relationships. The structures for representing data in various layers of this map and the mechanisms for their use are described. In particular, to describe the routes of the RTK, the principles of interpretive navigation are used to provide information about the operating conditions and objects of interest of the signature structure. The software implementation of the proposed mechanisms is based on a unified approach based on the real-time STZ software framework. Examples of the use of the described technologies in solving the problems of information support for targeted movements of ground RTCs are given

References

Скачивания

Published:

2024-04-16

Issue:

Section:

SECTION IV. TECHNICAL VISION

Keywords:

Degree of autonomy, vision system, scene analysis, multilevel cognitive maps, configuration space, real time vision system framework, robotics ontologies, signature