返回Google圖書搜尋

Connectivity Optimization in Robotic Networks

出版	École polytechnique de Montréal, 2018
ISBN	10856804289781085680424
URL	http://books.google.com.hk/books?id=IdmUzgEACAAJ&hl=&source=gbs_api

註釋Because of their promising applications, the interest for machine-to-machine interaction has soared (e.g. between smartphones, laptops, robots, sensors, or routers). Autonomous homes and smart cities are just two examples among the many. Without a good communication system, devices are unable to share relevant information and take effective decisions. Thus, inter-device communication is key for successful cooperations. To guarantee suitable communication, devices need to rely on a robust network. One might think of classical centralized network architecture since it is so common - antennae relay our smartphone communications, and routers provide us with an Internet connection at home. However, this architecture is not adequate for every application. When a central node (e.g. an antenna or a router) fails, it can cripple all the network. Moreover, fixed infrastructure is not always available, which is detrimental for applications like search and rescue operations. Hence, other network designs can be more suitable, like ad hoc networks, where there is no central node and every device can route information. This work aims at establishing mobile ad hoc networks between multiple devices for search and rescue operations. This thesis is framed by a partnership between Humanitas Solutions and Ecole Polytechnique de Montreal, whose goal is to relay information between victims and first responders by the use of smartphones and flying robots (i.e. drones). For this purpose, we have to enable infrastructureless communications between devices and maintain drones connected over the disaster area. Although off-the-shelf smartphones and drones can use numerous communication protocols, it is often impractical to create a mobile ad hoc network. The current state of the art only provides solutions dedicated to Android, and most of these require rooted devices. Concerning the drones, there are many control algorithms to maintain connectivity in literature. However, these algorithms involve parameters that are set empirically, despite the fact that they greatly influence the overall performance of the solution. In this thesis, we first validates HEAVEN, a communication middleware that builds ad hoc networks between devices, including Apple products. Then we propose a methodology to automate the parameter selection for mutli-robot control algorithms. Our tests reveal that HEAVEN provides an effective throughput up to 18kB/s. Despite the fact that android-based solutions outperform HEAVEN, the latter works in user space and without rooted devices. A new version, currently under development, will provide consider.