Multi-Robot Exploration with Strictly Limited Communication

碩士 === 臺灣大學 === 機械工程學研究所 === 98 === The main objective of this thesis is to develop a frontier-based multi-robot exploration structure with strictly limited communication. Generally, robots explore unknown areas without communication; therefore, this study aims to improve the future possibility of t...

Full description

Bibliographic Details
Main Authors: Jou-Feng Huang, 黃柔丰
Other Authors: 黃漢邦
Format: Others
Language:en_US
Published: 2010
Online Access:http://ndltd.ncl.edu.tw/handle/19757005887815691001
Description
Summary:碩士 === 臺灣大學 === 機械工程學研究所 === 98 === The main objective of this thesis is to develop a frontier-based multi-robot exploration structure with strictly limited communication. Generally, robots explore unknown areas without communication; therefore, this study aims to improve the future possibility of these robots coming into contact with one another in order to avoid overlap. First, a two-robot exploration strategy, which is the foundation of multi-robot exploration structures, is proposed. Various frontier properties, such as frontier tree tracking and frontier area, are described to assist the robot in understanding the present situation. Then, the meeting-oriented goal assignment (MOGA) is proposed to provide smooth frontier exploration orders for the two robots. This is followed by the development of the meeting-oriented frontier search (MOFS), which is aimed to make robots determine the balance between deep exploration and meeting-point approach. Rules are also summed up to complete the two-robot exploration structure. Next, the two-robot structure is extended to a multi-robot exploration structure by dividing the robots into several two-robot groups. In a given communication process, each robot has one identity, assigned from several possibilities: Full Team (FT), One Teamer (OT), Not Teamer (NT), and Not-Meeting Robot (NMR). The NMR information tracking and the OT goal assignment are proposed to ensure possible meetings of team members. MOGA for teams is developed to assign left frontiers to FT and NT, in order for each team to identify its own goals. Finally, a rejection force term is added to the multi-robot MOFS to avoid the overlap of teams. Three maps are used in the simulation environment. The strategy is compared with the “coordinated” algorithm and better results are observed. The proposed strategy performs satisfactorily, even if the environment has become larger. A real-world two-robot exploration is also implemented after solving a number of practical problems, such as map merging and wireless communication.