Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
Transportation

$500 Drone Calculates Its GPS Coordinates Offline from Downloaded Google Maps and a Camera (dronenr.com.au) 59

From a report: A team of drone enthusiasts have built a sub-$500 drone that uses a camera and Google Maps to provide itself with GPS co-ordinates, removing the need for a GPS satellite signal. And all of this was done in 24 hours during the El Segundo Defense Tech Hackathon. The drone the trio opted for is a custom designed and 3D printed fixed wing featuring a large single motor towards the rear and a downward facing camera used for geo-referencing...

Doesn't Google Maps still require internet, you may ask? Google Maps allows users to download segments of maps ahead of time, usually for use when you are travelling or camping out in remote areas. In this instance, the team used this feature to their advantage, allowing the drone to continue operating...

Thanks to long-time Slashdot reader schwit1 for sharing the news.
This discussion has been archived. No new comments can be posted.

$500 Drone Calculates Its GPS Coordinates Offline from Downloaded Google Maps and a Camera

Comments Filter:
  • by Software ( 179033 ) on Sunday February 25, 2024 @09:52PM (#64268368) Journal
    They're just coordinates.
    • by ls671 ( 1122017 )

      They're just coordinates.

      Well, while at it, there are no "GPS" coordinates, they're all are simply coordinates, it doesn't matter how you get them and coordinates existed way before GPS has.

      Now, armies will need to print giant rivers, lakes, roads, mountains, buildings, etc. ans lay them on the ground to fool the GPS positioning system. /s

    • Yet another example of generification. Not so long ago, GPS specifically meant a single US satellite navigation service. Nowadays it seems to mean any such service (including Galileo and GLONASS), the Iridium satellite service (which is communication not navigation) and even the humble latitude/longitude/altitude triplet which has been used in navigation for centuries.
      • Not so long ago, GPS specifically meant a single US satellite navigation service.

        In my mind GPS is the generic term and Navstar is the name of the singular USA specific variant. I guess that thinking just makes me old and stubborn.
        The following Wiki link contains "Navstar" but gets redirected to "GPS":
        https://en.wikipedia.org/w/ind... [wikipedia.org]

        The generic term for GPS has evolved into "GNSS" or simply "satellite navigation".

    • by ceoyoyo ( 59147 ) on Sunday February 25, 2024 @10:44PM (#64268446)

      Lol. Tell that to a navigation geek.

      They're WGS84 coodinates. WGS84 was created to supercede it's predecessor, WGS72, but the US Department of Defence to support future high accuracy positioning systems. I'm not sure they ever stated *specifically* that those systems were GPS, but that is the major one.

      "GPS coordinates" isn't a bad laymanization of that.

      "Just coordinates" is just wrong.

      • by Entrope ( 68843 )

        Ah, but which edition/version of WGS84? The original 1987 edition, or the update from 1991, 1997, 2000, 2004, 2014 or 2024?

        In practice, I'm not sure that Google's map data and photos are consistent in using the same reference frame, so perhaps the right phrase would be "Google Maps coordinates".

        • by ceoyoyo ( 59147 )

          Exactly. There ain't no such thing as "just coordinates."

          Google maps uses WGS84 internally, presumably one of the later revisions, although I couldn't find out which. However, it displays a Mercator projection. No matter what, the map data won't match your camera views, so you'd have to transform. The obvious way to validate the system would be to match your computed coordinates against actual GPS coordinates, the specific version depending on the firmware of whatever receiver you used on the drone.

          • by pacinpm ( 631330 )

            Exactly. There ain't no such thing as "just coordinates."

            Google maps uses WGS84 internally, presumably one of the later revisions, although I couldn't find out which. However, it displays a Mercator projection. No matter what, the map data won't match your camera views, so you'd have to transform. The obvious way to validate the system would be to match your computed coordinates against actual GPS coordinates, the specific version depending on the firmware of whatever receiver you used on the drone.

            Nah, you don't have to transform. For areas which can be seen from such a drone there is no practical difference between different coordinates.

            • by ceoyoyo ( 59147 )

              Take a look out your window. Now take a look at Google maps. See how they're different? Yes, you have to transform between those two viewpoints.

              You might not need to correct for the map projection, depending on how high you're flying and how accurate you want to be.

              You don't need to transform between coordinate systems because Google maps and GPS use the same system.

      • by Anonymous Coward

        "GPS coordinates" isn't a bad laymanization of that.

        "Just coordinates" is just wrong.

        ”Coordinates” is less accurate but it is not “just wrong” no matter how passionate that geek is.

        WSG84 is an “Earth-centered” system, and lat/long is our oldest coordinate system.

    • LAT/LON coordinates have existed far longer than GPS has existed.
    • Thank you!

      I think they meant geodetic coordinates.

    • Aeroguardian for example https://navigationguard.com/ [navigationguard.com]
    • Guessing that's journalist speak for spoofing / translated into UBX or NMEA packets that the drone software is expecting, as it was probably programmed for a physical GPS on a UART serial port, CAN bus or some other solution.
    • Exactly. journalists.

    • Fun fact, my dad had a GPS receiver circa 1993. This was back when the military. They had an offline map and a GPS receiver - no internet. You could buy them at the sporting goods store for about $500. He use it for hunting. Glue that to an RC hobby helicopter which has been around since the 60s, and you basically have this. They re-invented the wheel, except made it less good by stripping away innovation in the space (namely internet connectivity).
      • I had one too, and it still needed to receive the GPS signals form the satellites: if I got in under brush in a narrow canyon it wouldn't know where I was etc. The system they are talking about does not receive data from satellites when out in the field.

  • by Baron_Yam ( 643147 ) on Sunday February 25, 2024 @10:09PM (#64268382)

    https://dronexl.co/2024/02/19/... [dronexl.co]

    Built for the US Department of Defence, so I wonder how long until the code is passed to Ukraine?

    Put that system in something with long thin wings and fly it at high altitudes and you might make something that could go long distances to deliver gifts to Russia without the delivery being interrupted.

    • by ceoyoyo ( 59147 ) on Sunday February 25, 2024 @10:45PM (#64268454)

      It doesn't work well at high altitudes. There are things called clouds.

      It works well at low altitudes though. The US Department of Defence is rather familiar with drones with wings that go long distances using terrain following cameras and deliver gifts. They call them "Tomahawks."

      • Tomahawks cost more than $500.

      • In high altitudes maybe it's better to look up instead of down and use celestial navigation. Not so precise, but it can be corrected better when flying lower to reach the destination.

      • by Cyberax ( 705495 )

        It doesn't work well at high altitudes. There are things called clouds.

        You can use good old inertial and magnetic navigation, periodically diving below the cloud cover to correct for any drift. It won't be foolproof, of course, but it will work in a much wider range of conditions.

      • Go back far enough in time and it wasn't even cameras. The AGM-86B "terrain following" IIRC used the radar altimeter and compared the rises and falls to the stored terrain map.

    • by AmiMoJo ( 196126 )

      Japan used it with their precision Moon landing. Because there is no satellite navigation on the Moon, they used cameras to look at features on the ground and at the horizon. The probe knew what to expect based on simulations, and was able to land well within the target area.

    • I'd be pretty surprised if they didn't do this already. I mean, this is a great hackathon effort and all, but you're basically just talking about taking an image that you are already able to scale match, and then cross correlating it with satellite map imagery. If you have a rough idea where the drone started out, even brute forcing this with a pretty naive approach would likely be good enough given modern computing power.

      There are a lot of optimisations you could do of course, such as (of the top of my hea

      • by jabuzz ( 182671 )

        Military missiles already do this, including using terrain profile and inertial guidance. It is handy to have a GNSS, especially for the start coordinates but it is not necessary to achieve good precision.

      • by hawk ( 1151 )

        sure, it's nothing the military isn't already doing--but this has a *lot* less zeros on the price per unit!

  • If you thought an enemy might do something like that, you could possibly thwart the attempt via large scale landscape changes - cut down a few acres of forest, put up large scale nettings overhead that looked like rocks where there were none before.

    • The countermeasure is obvious: Use inertial dead reckoning to pass over ambiguous terrain.

      All drones have a 3D accelerometer, so the countermeasure is just software.

    • Sounds a bit impractical, given that they can update the imagery on Google maps faster and cheaper than you can terraform the countryside :)

      • The presumption is that there's radio jamming in effect to prevent navigation by GPS and other radio transmissions, as well as any human aid by radio link. This would be for navigation of a drone on a mission with no radio contact, or possibly a manned mission where the humans in the craft need computer navigation aids to avoid errors and add precision. There will be no updates to the maps once the aircraft leaves the ground. The immediate landscape could have been modified in an attempt to prevent navig

        • Such jamming would potentailly block the connection to the internet however it is obtained.
      • they can update the imagery on Google maps faster and cheaper than you can terraform the countryside :)

        True for adjusting acres of trees, but randomly putting up tarps that look like various material from overhead could be done much faster than Google Maps can update an area...

        However not faster than a major government satellite can give essentially live feeds for areas of current interest. So it's really on a technique that works against fairly poor attackers.

    • Re: (Score:3, Interesting)

      by MacMann ( 7518492 )

      I thought the same but with the addition of "glitches" like those used to confuse self-driving cars. If there's patterns on road signs, vehicles, or whatever that resemble some camera failure then the computer will give up on processing an image, either because it detected a bad image or the image contained something it didn't know how to process. I don't know if I'm describing the ways to fool the computers well and I'm not finding examples with a quick web search.

      If there's patterns the image recognitio

      • ... radio navigation is becoming obsolete.

        Radio navigation travels much further than (QR) signs painted on hillsides and barn roofs. It's great on the coast where the ocean allows LoS and also at altitude.

        Use the AM band and a lot of people can navigate and hear the news with simple electrical technology.

        • I should have wrote that radio navigation as we know it today could become obsolete.

          Smart phones can establish location from pattern recognition and cameras, and with inertial navigation, with nothing really new in hardware added. What cell phones did for location data before GPS was the norm was use navigation from cellular towers. Phones could revert to navigation by cellular signal should something happen to satellite navigation, and it might not even require any new hardware, then use visual cues for

    • If you're having much trouble with such drones, is much simpler to just use lasers powerful enough to blind the drone's cameras.

      • If you're having much trouble with such drones, is much simpler to just use lasers powerful enough to blind the drone's cameras.

        Maybe but how do you find the drones to blind them?

  • by fahrbot-bot ( 874524 ) on Sunday February 25, 2024 @11:32PM (#64268522)

    It uses Apple Maps ... :-)

  • Overheard during an April job interview:

    "Optimizing a drone to use a local map to simulate a GPS system doesn't really match the job description. I'm afraid you just aren't a good fit for our culture. Have you considered joining a code camp? If you'll excuse me I'm Tiktoking our new kale dishes and eucalyptus towels for lunch."

  • I wonder if they found my wallet?

  • Terrestrial guidance has been around for a long time. Our high end weapons do not depend on GPS. They use a combination of inertial and terrestrial guidance systems. What is posted here is a 2D terrestrial map guidance system. Some Cruise Missiles use 3D terrestrial guidance systems which use a combination of radar and optical sensors and when "blinded" fall back on inertial guidance. I don't know how they pulled this off with a sub $500.00 drone. You would need multiple computers. First the flight

You know you've landed gear-up when it takes full power to taxi.

Working...