Navigation of autonomous systems based on situation control with dynamic replanning
The solution to the problem of autonomous mobile systems navigation is a complex task, traditionally presented in the form of solving the sequence of subtasks: perception of information about the environment, localization and mapping, path planning, and motion control. A large number of scientific w...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ivan Kozhedub Kharkiv National Air Force University
2020-09-01
|
| Series: | Системи обробки інформації |
| Subjects: | |
| Online Access: | https://journal-hnups.com.ua/index.php/soi/article/view/413 |
| Summary: | The solution to the problem of autonomous mobile systems navigation is a complex task, traditionally presented in the form of solving the sequence of subtasks: perception of information about the environment, localization and mapping, path planning, and motion control. A large number of scientific works are devoted to the solution of the listed subtasks. However, existing research does not pay enough attention to the integration of individual elements of the navigation cycle solutions into a single homogeneous system. This leads to an additional accumulation of errors in the process of a complex solution to the navigation problem. In previous works, a model was proposed that provides homogeneous integration, using for this a multi-level structure of representing an autonomous system's knowledge in the form of sets of fuzzy rules and facts. The five-level model represents the autonomous system's knowledge of goals, paths, an environment map, strategies, and specific controls necessary to achieve the goal. To ensure adequate processing of fuzzy rules, a modified Takagi-Sugeno-Kang fuzzy inference model is proposed. In this work, the previously proposed model is expanded. The model was tested in conditions of noisy sensor data. A method is proposed for the formation of level 2 rules, which describe an autonomous system's cartographic knowledge about the environment, using the well-known methods of global path planning. Extension of the model provides dynamic paths replanning of the autonomous system, using the processing of present knowledge about existing paths. Such replanning is effective in terms of computational time and independent of the completeness of the knowledge base of complete paths. |
|---|---|
| ISSN: | 1681-7710 2518-1696 |