Mobile robots autonomous exploration through deep reinforcement learning

• Main Author: Yin, Hanqiu
• Other Authors: Chau Yuen
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Engineering; Autonomous exploration; DQN; Frontier detection; RL
• Online Access: https://hdl.handle.net/10356/176603
• Institution: Nanyang Technological University

The mobile robot autonomous exploration problem has been a hot research topic among scholars. The traditional robot autonomous exploration problem usually contains steps such as frontier detection, frontier clustering, navigation point assignment and so on. For robots, the quality and speed of frontier detection and the speed of decision-making algorithms will have a large impact on the exploration efficiency of robots. Typically, traditional algorithms need to be designed in conjunction with expert experience, and when the exploration environment changes, the algorithms usually need to be redesigned. In order to improve the transferability and generalizability of robot autonomous exploration algorithms, the use of deep learning methods as decision-making modules is quite promising. Existing deep learning-based methods suffer from problems such as low learning efficiency and might fall into local minima. In this thesis, we propose a decision-making algorithm based on deep reinforcement learning. The algorithm adopts a deep learning method to effectively extract the environment features and automatically updates the strategy by interacting with the environment. The algorithm outputs the next target position of the mobile robot according to the current position of the mobile robot and the environment information. Through the existing proven navigation algorithms, the robot is navigated to the target point outputted by the decision module to realize the exploration of the unknown environment. Experiments show that the algorithm has good learning efficiency and adaptability to different environments.