Author: Rui Miguel Pires Carvalho

Publisher: University of Coimbra

Published: 2016-09-01

Total Pages: 82

ISBN-13:

Over the years the growing influence of robotics in the human domain has been noticeable from industrial to space and medical applications as well as a tool in adverse environments and even in everyday tasks. Many of these applications require the use of a team of several cooperating mobile robots - Cooperative Multi-Robot System (CMRS) - to make the execution of certain tasks possible and to improve the performance achieved by only one robot.Although the cooperation capacity is innate to humans, the robotic domain features a number of new challenges: communication, timing of the information obtained and the merger of that information.When cooperation among multiple robots is applied in an exploration context challenges increase. It is essential to take the costs and utility of that exploration into account.This dissertation aims to present a solution to the aforementioned problem. Therefore, a method has been developed, capable of assigning to different robots a cooperative behavior in order to explore an environment, following a philosophy of "divide and conquer".The CoopExp, package with the operation algorithm, was developed according to a distributed approach in order to increase resistance to individual failures of the exploration agents. Accordingly, a method that is able to calculate the costs involved in a faster and more efficient way, was created. Furthermore, an approach to exploration utility was also established, based on a compendium of techniques described in the literature.The development of such programs would have practically been impossible without performing tests on its functioning. In the absence of a simulator for this type of operation, the ARENA (cooperAtive multi Robot frontiEr exploratioN simulAtor) was developed. It consists of a set of new packages specifically designed for frontier identification (aap_frontiers) and to optimize the simulation, through simplifications in the process ofachieving the maps (aap_mapping) and their subsequent combination, yielding the global map (aap_map_merger).Such solutions were validated through simulation tests, using mobile units equipped with a LRF (Laser Range Finder). These tests showed that exploration time decreases when the number of robots is increased, presenting a proper performance in terms of scalability and efficiency in exploration. Last but not least, a exploration with a real team of robots was successfully carried out that was able to communicate through a wireless network in order to validate the practical functioning of this project