Engineers at the University of South Australia have taken a page from the days of wooden sailing ships and developed a way for drones to navigate by the stars at night using simple, lightweight equipment for areas where GPS signals aren't available.
When you look at images of old aircraft designed for long-distance flights, pull out the magnifying glass or use the image enlarge function and you might see a small plastic dome sticking out of the top of the fuselage. In the days before GPS and radio direction finders, this was how the plane's navigator would fix their position – by using the stars.
The principle dates back centuries to the time when ship captains would navigate by using sextants and similar instruments to take fixes on the Sun and stars and then, with the help of astronomical tables and chronometers, calculate their latitude and longitude with an accuracy of about 1.5 nautical miles (2.8 km).
Not bad when you've been sailing out of sight of land for months at a time.
Almost exactly the same techniques were used by aviators for most of the 20th century, with the only differences being to adjust for the fact that the star sights are being made from an unstable platform in the sky.
Though celestial navigation is still practiced by most mariners and some aviators, it's mainly as a backup for when GPS isn't available or has been degraded because of local conflicts causing the US to shut down the secondary GPS fine tuning. It's possible to put automated systems on a drone that can do the sights and calculations automatically, but such systems are very complicated because they have to compensate for the movements and directions of the aircraft to work. This is not only expensive, it means a lot of extra weight and power consumption.
To get around this, the South Australia team developed an algorithm that allowed them to use a series of images of the night sky to fix the position of a winged drone to within 4 km (2.5 miles) without the need for any outside signals or data links.
According to the team, the new strap-down system is lightweight, low cost, modular, and uses ArduPilot running on a Cube Orange flight controller. During a navigation fix, the drone flies in a circle through all points of the compass as it takes multiple images. By comparing the images, the algorithm removes any biases and aligns the camera with the drone's Attitude and Heading Reference System (AHRS).
The new system, when matured, is claimed to not only have military applications, but also can be used for long-endurance environmental monitoring in remote areas.
"Unlike traditional star-based navigation systems, which are often complex, heavy and costly, our system is simpler, lighter and does not need stabilization hardware, making it suitable for smaller drones," said UniSA researcher Samuel Teague. "This type of navigation is ideal for operations over oceans, or in warfare zones where GPS jamming is a risk. Apart from the defense sector, it could also be highly useful for environmental monitoring."
The research was published in Drones.
Source: University of South Australia