Por favor, use este identificador para citar o enlazar este ítem:
http://cimat.repositorioinstitucional.mx/jspui/handle/1008/1101| Multi-robot Exploration and Semantic Map Building: Distributed Algorithms and Heterogeneous Agents | |
| GABRIEL AGUILAR MENDOZA | |
| Acceso Abierto | |
| Atribución-NoComercial | |
| CIENCIAS DE LA COMPUTACIÓN | |
| In this thesis, we deal with multi-robots systems, such systems are of interest since the performance for several robotic tasks are in general better using multiple robots than the performance obtained with a single robot. In the first part of this document, we address problems related to minimalistic robotics. The main contributions are the following: 1) The proposed algorithms guarantee exploring the whole environment or the largest possible portion of it, in finite time, even though the robots are no capable of building an exact map of the environment, they cannot estimate their positions, and each robot does not have full information about the part of the environment explored by other robots. 2) The method only requires limited communication between the robots. 3) We have proposed an exploration strategy that guarantees covering the largest possible portion of the environment with the visibility regions of omnidirectional sensors. 4) The approach scales well to hundreds of robots and obstacles. In the second part of this thesis, the main goal is to provide a motion strategy for a team of robots to explore and build a map of an unknown environment. The team of robots consists of heterogeneous agents, namely, two terrestrial robots with different sensing and motion capabilities and an aerial drone. The strategy has to consider robots’ capabilities to determine sub-goals to explore the environment with the terrestrial robots. We focus in two main aspects: 1) the heterogeneity of the robots in the team, and 2) the uncertainty in sensing and motions of the robots. The main contributions are the following: 1) An observation model that consists of several neural network classifiers that have as an input laser data and video images providing as output a semantic map. 2) A motion model, which relativizes the accuracy of the motion of the robots with respect to the robot equipped with the most precise motors. 3) An efficient motion policy based on stochastic dynamic programing. 4) The approach is validated with simulations and experiments in real terrestrial robots and a drone. | |
| 08-09-2021 | |
| Trabajo de grado, maestría | |
| OTRAS | |
| Versión aceptada | |
| acceptedVersion - Versión aceptada | |
| Aparece en las colecciones: | Tesis del CIMAT |
Cargar archivos:
| Fichero | Descripción | Tamaño | Formato | |
|---|---|---|---|---|
| TE 810.pdf | 6.6 MB | Adobe PDF | Visualizar/Abrir |