Research Information System
Thesis Type: Postgraduate
Institution Of The Thesis: Kahramanmaras Sutcu Imam University, Graduate School of Science and Technology, Informatics, Turkey
Approval Date: 2021
Thesis Language: Turkish
Student: HAMZA AYDEMİR
Supervisor: Mehmet Tekerek
Abstract:
Elfes emphasized the importance of the concept of autonomous mobile robots that can work in unknown environments to expand the application and spread of robots, especially in the context of research and industry. In this direction, it has become basic requirement over time to ensure that a robot can navigate using only its sensors when placed at any point in an unknown environment. This demand, which is expressed as autonomous navigation, has led to the need for the robot to plan a path. Path planning is trajectory planning for the robot to move efficiently from a starting point to a destination point. Path planning may require navigating the entire workspace according to the autonomous mobile robot's job description. This problem is referred to as Complete Coverage Planning (CCP) in the literature. CCP is a path planning task that passes through all points of an area while avoiding obstacles. The problem addressed in this thesis is to process a partially filled cell as completely full in Grid Based Complete Coverage Planning (GBCCP), which is the most widely used CCP method in the literature. In this context, the use of the clustering method, which will be created by taking into account the characteristics of the environment of the coverage problem encountered in the GBCCP method, has been determined as a research question. In this direction, it is suggested to use the K-means++ algorithm, which is a widely used clustering algorithm and partitioning technique introduced in the literature by Arthur and Vassilvitskii in 2007. The proposed K-means++ Complete Coverage Planning (Km++CCP) method has been validated by experiments on the robot in the simulation environment and the real world. At the same time, the performance of the Km++CCP method and the performance of the GBCCP method were compared. According to this; In all experiments, it has been calculated that the indoor coverage performance of the Km++CCP method is higher than the GBCCP method. When the findings obtained from the experiments are examined, it is seen that the proposed Km++CCP method covers 96.86% of the indoor in the simulation environment, 95.97% of the real-world obstacle-free indoor, and 94.32% of the real-world indoor with obstacles. It has been observed that the GBCCP method covers 89.76% of the indoor in the simulation environment, 82.47% of the real-world obstacle-free indoor, and 77.95% of the real-world obstacle-free indoor.