Visual autonomous navigation in forest environment
Sormunen, Teemu (2020)
Sormunen, Teemu
2020
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202005077659
https://urn.fi/URN:NBN:fi:amk-202005077659
Tiivistelmä
The purpose of this study was to do a research on the state of autonomous navigation in a forest environment and to show how the same principles could be implemented in conjunction with a mobile robot. The work on building and programming a mobile robot was part of a course completed before this study, and the results are introduced here. One important goal of this thesis was also to provide the reader with an overview on some of the recent studies on autonomous navigation.
Data was gathered by studying (a relatively extensive number of) research papers on the subject. A discussion on the most relevant techniques and algorithms that are used in modern-day navigation was included in order to explain the central concepts to the reader. The abovementioned algorithms were used in developing the mobile robot.
Earlier studies indicate that the forest environment is relatively undiscovered in terms of visual navigation and mapping due to lack of practical usage. The field of study is however gaining popularity as the capability increases to create more robust ways of navigating in complex environments.
The findings indicate that in order to move forward in the domain of forest in autonomous navigation, work needs to be done on creating certain universal evaluation metrics in either real-life or in simulated environment. Moreover, a way for creating commercial applications must be found. Only after this, the quality of algorithms can be compared, and field will become more competitive. An example of resulting competitiveness can be seen in the growth explosion of machine learning and autonomous passenger vehicles.
This thesis opens up possibilities for the forest industry to become part of this growth by providing the necessary groundwork for visual navigation.
Data was gathered by studying (a relatively extensive number of) research papers on the subject. A discussion on the most relevant techniques and algorithms that are used in modern-day navigation was included in order to explain the central concepts to the reader. The abovementioned algorithms were used in developing the mobile robot.
Earlier studies indicate that the forest environment is relatively undiscovered in terms of visual navigation and mapping due to lack of practical usage. The field of study is however gaining popularity as the capability increases to create more robust ways of navigating in complex environments.
The findings indicate that in order to move forward in the domain of forest in autonomous navigation, work needs to be done on creating certain universal evaluation metrics in either real-life or in simulated environment. Moreover, a way for creating commercial applications must be found. Only after this, the quality of algorithms can be compared, and field will become more competitive. An example of resulting competitiveness can be seen in the growth explosion of machine learning and autonomous passenger vehicles.
This thesis opens up possibilities for the forest industry to become part of this growth by providing the necessary groundwork for visual navigation.