MEFA: Mapping and Enabling Future Airspace

有人和无人驾驶的空域正在经历一场革命。到2030年，空中交通预计将翻两番，载人飞机和无人驾驶飞行器(UAV)的总数将翻一番。这种爆炸性的增长将改变已经很繁忙的空域，并使之拥堵。无人机占据空域的方式类似于鸟类，它们的飞行高度和速度都是重叠的。因此，正如最近盖特威克机场的无人机入侵事件所证明的那样，迫切需要能够将无人机与使用相同空域的自然生物(例如鸟类)区分开来。关于鸟类如何利用空域的详细数据有限，特别是考虑到前所未有的城市化速度，其特点是高层建筑增加，人工照明(AL)增加，以及基础设施模式的变化。已建成的基础设施与AL之间的相互作用及其对鸟类生物学的影响，现在是研究迁徙生态学，特别是鸟类的迁徙生态学，以及明亮的城市结构(如纪念碑、建筑、通信塔)造成的死亡的研究重点。高层建筑上更多地使用玻璃和其他高反光表面，增加了鸟撞的频率，从而增加了鸟的死亡率。2004年，英国鸟类学信托基金(BTO)估计，在英国，每年有1亿只鸟撞到窗户。该项目主要使用一种专门为跟踪无人机而开发的雷达传感器。与以前的雷达研究相反，单独的鸟类和无人机可以在小群体中观察到，这使得对轨迹的测量比以前实现的更精细。然而，要对受控制的无人空域进行足够可靠的监视，根本的挑战是区分小型无人机和鸟类。鸟类有特定的飞行模式，这与无人机的飞行模式是可以区分的。这项研究将开发算法来区分无人机和鸟类，小群体中的个体鸟类(通常是2-5只)，以及潜在的更大群体中的个体鸟类。深度学习算法将被开发并测试其区分鸟类和无人机以及不同鸟类群体的能力。该项目跨越了EPSRC的“与环境变化共存(生态系统挑战)”和“全球不确定性(对基础设施的威胁)”的主题，以开发一个能够同时减轻鸟类和人类的安全风险的尖端系统。

Manned and unmanned airspace is undergoing a revolution. By 2030 air traffic is estimated to quadruple with a doubling of the total number of manned aircraft and unmanned air vehicles (UAVs). This explosive growth will change and congest already heavily used airspace. UAVs occupy airspace in a similar way to birds with both flying at overlapping altitudes and velocities. Therefore, as evidenced by the recent drone incursions at Gatwick airport, there is a pressing need to be able differentiate UAVs from natural organisms (e.g. birds) that use the same airspace. There are limited detailed data on how birds use airspace, especially in light of unprecedented rates of urbanisation, characterised by increasing high-rise building, increased artificial light (AL), and changing patterns of infrastructure. All are rapidly re-shaping habitats used by migratory and non-migratory species.The interaction between built infrastructure and AL, and its influence on bird biology, is now the focus of research addressing migration ecology, especially of birds, and mortality caused by brightly lit urban structures (e.g. monuments, buildings, communication towers). Increased use of glass and other highly reflective surfaces on high-rise buildings has increased the frequency of bird strikes and thus bird mortality. In 2004, the British Trust for Ornithology (BTO) estimated 100 million birds struck windows each year in the UK.This project primarily uses a 'staring' form of radar sensor developed specifically to track drones. Contrary to previous radar research, individual birds and drones are observable within small groups that allows finer measurement of trajectories than has been achieved previously. However, for sufficiently reliable surveillance of controlled unmanned-airspace, the fundamental challenge is to discriminate small drones from birds. Bird species have specific flight patterns that are distinguishable from those of UAVs. The research will develop algorithms to distinguish between drones and birds, individual birds in small groups (typically 2-5) and potentially individual birds in larger flocks. Deep learning algorithms will be developed and tested for their ability to distinguish between birds and drones, and between different bird groups. The project cuts across the EPSRC's themes of "Living with Environmental Change (ecosystem challenge)" and "Global Uncertainties (threats to infrastructures)", to develop a cutting-edge system with the ability to simultaneously mitigate security risks to birds and humans alike.