Multi-UAV Active Simultaneous Localization and Mapping in Unknown Indoor Environments

如今，无人驾驶飞行器(UAVs)(又称无人机)在无数应用中变得至关重要。然而，大多数传统的商用无人机都是遥控或半自动的，这阻碍了它们在具有挑战性的条件和关键时间任务(如搜索和救援)中的有效部署。幸运的是，随着轻量级传感和复杂实时计算的不断发展，无人机现在可以被设计成在执行任务期间自我感知和做出自主决策。然而，只有当我们能够成功和有效地部署全自动无人机机队来完成更复杂的使命和任务时，下一代无人机才能不辜负炒作。事实上，可以部署多架无人机来加快复杂任务的执行速度，同时协作绘制和探索未知环境。协作不仅加快了测绘和探索过程，还最大限度地减少了决策过程中的不确定性。该研究项目旨在解决在未知室内环境中导航的多无人机系统设计高效可靠的定位、测绘和探测算法方面的若干挑战。特别是，挑战在于部署一支无人机机队，依靠视觉和惯性系统，并赋予有限的机载计算能力，尽可能快速有效地自主导航和绘制完全未知的环境。我们为这一技术挑战提出的解决方案包括在不同级别的控制系统和机器人中融合不同的方法，以协调无人机组完成这一任务。该项目概述的研究目标的实现将为在复杂环境中运行的完全自主的多无人机系统打开新的和令人兴奋的应用之门。该项目也是一个极好的机会，可以在这些有前途的领域培养高素质的专业人员，他们可以成为研究和技术的领导者，支持加拿大的经济增长。

Nowadays, unmanned aerial vehicles (UAVs) (also known as drones) are becoming crucial in countless applications. However, most of the traditional commercial drones are remotely piloted or semi-autonomous which places obstacles in the way of their efficient deployment in challenging conditions and time-critical missions (e.g., search and rescue). Fortunately, with the unceasing development of lightweight sensing and sophisticated real-time computing, drones can now be designed to self-perceive and make autonomous decisions during the execution of tasks. Yet, the next generation of UAVs will be living up to the hype only if we are able to successfully and efficiently deploy fleets of fully autonomous UAVs to accomplish even more complex missions and tasks. In fact, multiple drones can be deployed to speed up complex missions while collaboratively mapping and exploring an unknown environment. Collaboration not only accelerates the mapping and exploration process but also minimizes uncertainty in decision making. This research project aims at addressing several challenges related to the design of efficient and reliable localization, mapping, and exploration algorithms for multi-UAV systems navigating in unknown indoor environments. In particular, the challenge consists in deploying a fleet of UAVs, relaying on vision and inertial systems and endowed with limited on-board computational power, to autonomously navigate and map a completely unknown environment as fast and efficiently as possible. Our proposed solution for this technology challenge consists in the amalgamation of different approaches in control systems and robotics at different levels to coordinate the group of UAVs to accomplish this mission. The accomplishment of the research goals outlined in this project will open doors to new and exciting applications for fully autonomous multi-UAV systems operating in tangled environments. This project is also an excellent opportunity to train highly qualified professionals in these promising fields who can become leaders in research and technology, sustaining Canada's economic growth.