Multi-UAVs network communication study for distributed visual simultaneous localization and mapping

Abstract

In this paper, the problem of multi-UAVs (Unmanned Arial Vehicles) Visual Simultaneous Localization and Mapping (SLAM) is considered by using a new framework for pose and map estimation using monocular vision and reduced communications capabilities. The problem of localization and mapping is solved by fusing monocular visual data with odometry measurements through Graph SLAM formulation. Using each robot's map data representation, a proposal is made for a good and robust communication between UAVs to perform efficient data exchange while keeping SLAM performances. A mesh network is chosen to import solutions to wireless networking. Finally, some validation experiments are performed in an Ad Hoc Network and a Wireless Mesh Network using Better Approach To Mobile Ad Hoc Network (BATMAN) protocol.

Diseño Distribuido de un Sistema de Mapeo y Localización basado en visión para un Robot Móvil

“…Actualmente, la creciente utilización de vehículos aéreos en diversas aplicaciones ha desencadenado numerosos estudios sobre esta plataforma. Al igual que todos los robots móviles, los vehículos aéreos como los drones también requieren de soluciones SLAM para su funcionamiento autónomo, tal esel caso de (Williams et al, 2015), (Mahdoui et al, 2016) que proponen el uso de drones para obtener, mediante la aplicación del paradigma de PTAM (Klein and Murray, 2007), un mapa de forma colaborativa. Por otro lado, teniendo en cuenta la limitación de peso que los drones pueden transportar en comparación con los robots terrestres, el procesamiento distribuido toma un valor fundamental ya que permite balancear tanto carga como procesamiento en dispositivos externos y de esta manera liberar al robot de muchas responsabilidades.…”

Pound: A multi-master ROS node for reducing delay and jitter in wireless multi-robot networks

“…This is quite frequent in distributed applications where the exchanged data have a periodic and perishable nature. This is the case, for example, in the collaborative or distributed construction of a map (see [4], [5], [6] or [7] where the laser and localization data that reach the base station must do so coherently and within firm time windows to allow the task to be completed successfully) or cooperative localization (see [5] where LIDAR data sent to Comparison between wired and wireless configuration. Period measured at the destination node in a system in which a flow with period of 200 ms is established over wireless and wired connections.…”

“…To investigate the performance of Pound, we compare it to the state-of-the-art approaches such as Nimbro 3 [11], RT-WMP 4 [12], and the standard ROS communication protocols. The comparison is made using a scenario with two flows of information of different importance in terms of jitter, delay and bandwidth.…”

Relative Visual Localization for Unmanned Aerial Systems

“…This creates the foundation for the implementation of multi-agent coverage control and ensures timely collaboration between the distributed drones. In particular, the relative pose information needs to be reported with minimal latency [5]. While communication-based models can cope with these problems, not all UASs are equipped with such capabilities or may be operating in conditions where they are impaired [6].…”