据美国《大众科学》月刊网站12月3日报道，自主无人机好不好通常由全球定位系统(GPS)决定。如果没有它，无人驾驶飞行器实际在空中毫无用处，除非人类能够远程接管飞行职责。无人机GPS干扰攻击增多，这尤其成问题。一个研究团队没有试图开发日益先进和昂贵的导航技术，而是转向人类最古老的制图方法之一：绘制恒星图。

南澳大利亚大学的遥感工程师建立了新的低成本原型系统，将天体三角测量与夜间飞行的无人驾驶飞行器的视觉算法计算结合。但与现有GPS不同，这种新颖的设计不会发出任何信号，使其不受当前干扰方式的影响。

该团队在发表于《无人机》杂志的论文中解释说：“天体导航在现代无人驾驶飞行器中很少见。稳定成像系统的尺寸和重量以及精度的缺乏往往与飞机的操作要求不一致。然而，在全球导航卫星系统被拒的环境下，天体导航是为数不多能够在夜间进行全球海洋导航的非发射方式之一。”

为使其发挥作用，工程师们仅使用树莓派5微型计算机和配备广角镜头的单色传感器，设计和制造了捷联装备。然后，他们将这一工具连接到固定翼无人机的机载自动驾驶系统上，该系统捕获并通过算法分析从夜间看到的恒星获取的视觉数据。

研究助理、报告的合著者塞缪尔·蒂格解释说：“如果我们能够识别这些恒星并将它们与数据库比较，鉴于我们知道摄像头面对的方向和拍摄图像的时间点，我们实际可以从这些数据推断飞机的位置。”

蒂格和高级研究员贾万·查尔用无人机对他们的系统进行测试，表明升级他们的无人机，使无人机能够在固定高度和飞行速度轨道上始终在4公里精度范围内，估算自己的位置。虽然目前不像现代GPS那样精确，但该工具可能很快就能在发生干扰或故障时提供强大的备份。它还需要晴朗的天空评估周围的环境，但研究小组认为，进一步的研究也可以解决这个问题。

除了作战环境和长期监视任务，蒂格和查尔希望，他们廉价的、以恒星为基础的无人驾驶飞行器附加设备也能在飞越海洋或监测远程环境条件时提供帮助。

An ancient mapping technique may offer a reliable backup tool for UAVs.

By Andrew Paul

An autonomous drone is usually only as good as its GPS navigation system. Without it, the uncrewed aerial vehicles (UAVs) are essentially dead in the air unless a human can remotely take over piloting duties. This is especially problematic given the rise in drone GPS jamming attacks. Instead of trying to develop increasingly advanced and costly navigation technologies, one research team is turning to one of humanity’s oldest mapping methods: charting the stars.

Remote sensing engineers at the University of South Australia have built a new, low cost prototype system that merges celestial triangulation with vision-based algorithmic computing for UAVs flying at night. But unlike existing GPS, the novel design doesn’t emit any signals, making it impervious to current jamming methods.

“Celestial navigation is rarely seen in modern Uncrewed Aerial Vehicles (UAVs). The size and weight of a stabilized imaging system, and the lack of precision, tend to be at odds with the operational requirements of the aircraft,” the team explained in their paper published in the journal, Drones. “Nonetheless, celestial navigation is one of the few non-emissive modalities that enables global navigation over the ocean at night in [a] Global Navigation Satellite System (GNSS) denied environment.”

To make it work, engineers designed and constructed a strapdown payload using only a Raspberry Pi 5 miniature computer and a monochrome sensor fitted with a wide angle lens. They then connected the tool to a fixed-wing drone’s onboard autonomous piloting system, where it captured and algorithmically analyzed visual data taken from stars seen at night.

“If we’re able to identify those stars and compare them against a database, given that we know the orientation the camera was facing and the point in time at which that image was taken, we can actually infer the location of the aircraft from that data,” explained Samuel Teague, a research assistant and study co-author, in an accompanying university video.

Teague and senior researcher, Javaan Chahl, tested their system with a UAV, and showed that their drone upgrade allowed it to consistently estimate its location to within an accuracy of 4 km (roughly 2.48 mi) while performing fixed altitude and airspeed orbits. While not currently as precise as modern GPS, the tool may still soon provide a powerful backup in the event of jamming or malfunction. It also still requires a clear sky to assess its surroundings, although the team believes additional research could address this issue, as well.

Apart from combat environments and longterm surveillance missions, Teague and Chahl hope their cheap, star-based UAV attachment can also help when flying over oceans, or for monitoring remote environmental conditions.