Homemade AI drone software finds people when search and rescue teams can't

6 months ago (wired.com)

It's basically the FREE POINT square on the bingo card at this point. When someone builds a cool robot that they don't know what to do with, it's inevitably for SAR. I've worked on a couple of them myself.

  • Can't forget having to signal away from military. Accessible drone AI can be used to stop petty crime, man the borders, catch drug smuggling boats & ofc, kill people effectively. These uses are drenched in politics, so have to signal something more palatable.

    Truly accessible drone automation software is instant unicorn status. It would've been acquired by Anduril or Dji before it saw the light of day.

    We live in an interesting time where the 1st world's biggest problems (illegal immigration, drug smuggling, obesity, effective transportation, solving crimes, infinite energy) are already solved from a technical standpoint. But we artificially limit their use due to dystopian vibes.

    • Good! The "vibes" are real. We don't live in Star Trek, we live in Black Mirror. We dont have a world where Superman would exist, we would have Homelander.

      The ability for an authority to have perfect knowledge of who is in every area of a city, and conversely to have instant access to where every person is, is terrifying.

      At no point in human history has society been built on expecting perfect, uniform enforcement of laws and that has gone from "beyond wildest imagination" to technically doable in two generations.

    • Surely this is sarcasm. None of the things you list are solely technical problem, or even technical at all. They are all political problems.

      The politics that you claim artificially limit solutions are simply a set of politics different than your own.

      The 1st world’s biggest problems start with the climate change they are causing.

  • It even has an xkcd: https://xkcd.com/2128/

  • Free business ideas, because I want this to exist:

    Use drones with IR cameras:

    * Find deer after they're shot. Right now you need to hire a blood hound and it takes hours

    * Do wildlife surveys for conservation and management departments

    * Pest management for farmers

    • > * Find deer after they're shot. Right now you need to hire a blood hound and it takes hours

      They tried this exact thing with the kentucky freeway shooter using both helicopter based FLIR system and IR camera equipped drones and failed. Eventually the dudes body was found by a group of ... as far as i can tell, wilderness youtubers working with a police search party.

      Even the dogs didn't find him.

    • I've worked on 2 (and also for SAR). Not many people working in that field still and there's very little open data.

      Thermal is really good in the UK because the ground is rarely that warm. You can easily see a person hiding among rocks, for example. It's basically Cops. In equatorial Africa you can only use it at dawn/dusk (or overnight) because the ground is so hot.

    • If you google “thermal drone for hunting” you will find some YouTube videos about people solving the first problem.

      Pest management is a heck of a good idea. The province of Alberta is officially rat free - if Alberta doesn’t have something like this I bet they would be interested. Especially if it could do double duty for wildlife surveys.

      1 reply →

    • The problem is that night-vision-y technologies are heavily controlled and price inflated. Only few are willing to buy QCIF 30Hz cameras at $500 from AliExpress or from legitimate resellers at $2k.

  • Tardy add-on: after reading TfA this is not a case of cool robot becomes SARbot. This is a suite of specific tools meant to find people in the mountains. Come on y'all, you can't let me get away with middlebrow dismissal like that!!!

Search And Rescue (SAR) is sometimes VERY political. We build long-range RC airplanes to help find boats in distress at sea on the Mediterranean [1]. As you might expect, we could do 10x better, with less resources, if the powers that be wouldn't make it impossible to e.g. launch from an island and return to an island. So we launch from a boat, because it's not national territory and they can't make it impossible.

Shameless Plug: If you wanna join, let us know. We definitely would benefit from better on&off-board image recognition. But there are many, less buzzword-y, challenges as well: designing, building & testing airplanes, training pilots, delivering planes & batteries (challenging due to Watt-Hour restrictions), remote issue debugging, etc.

[1] https://tha.de/searchwing/

Put weapons on it (as already seen in current conflicts) and it becomes a seek-and-assassinate tool. Drones are cheap enough it could even be done en masse. It is a scary future, and it’s not far away at all.

  • S&R has always been a front for weaponized robotics, IMHO.

    The last DARPA grand challenge (Subterrainean) had automated drone networks that could find and identify humans in caves and tunnels. They were at least up front about the military challenges in these environments. (https://www.darpa.mil/program/darpa-subterranean-challenge), but the nod at civilian first-responders doesn't seem fair. Honestly, is cave-in such a big civilian problem that we need to prioritize it as a talking point at all levels?

    • > Honestly, is cave-in such a big civilian problem that we need to prioritize it as a talking point at all levels?

      Considering (1) the number of people who are employed in mining occupations, (2) the frequency of serious accidents in mines, yes. Particularly in developed countries, societies expect that great lengths will be gone to rescue or recover the victims, and mine rescue is incredibly dangerous work.

      (1) BLS says ~200K in the US in 2024, although only a minority of them work underground.

      (2) BLS says "underground mining machine operators" is the 9th deadliest job in the US, and that is with a large and well-equipped mine rescue system (MSRA says 250 teams across the country).

      3 replies →

    • It's generally hard to say what's a "front" for what, unless you mean "what can you get someone to grant you research money for when you really expect to parlay the learnings into another topic."

      Everything about the rocketry needed to get to orbit started from warfare purposes, for example. And ARPANET was a foray into how to build a disruption-resistant network for military purposes.

      Science and knowledge are a bit of a soup.

  • A read a comment here a while ago about "search and rescue" being a euphemism for military applications and that's the first thing I thought of when I saw this story.

  • I've been saying that any armistace on drones won't come until the US starts being hit by drone warfare. Especially by a foreign militia or nation state

Satellite SOS, recently introduced in iPhone and Google Pixel phones, will help a lot with lost hiker cases. However, drone-based search will still be useful in case the hiker broke the phone or is too incapacitated to use it.

  • There's another pretty common case: children who get lost somehow or decide to go on walks by their own and don't find their way back home but seek shelter wherever they can.

    I'm training for drone-based SAR at the German THW (Technical Relief Agency) at the moment, and that's probably going to be the majority of usecases.

I don‘t see any hint of AI being used here, but rather a handcrafted computer vision algorithm. Can anyone more involved in the matter elaborate if there was an actual AI model used?

  • We don't have a formal classification of which technologies can be considered "AI", but computer vision would feel like a valid entrant to me.

    • I thought AI meant "ML" + marketing.

      I joke, but not. I'm a researcher and AI has been a pretty ambiguous term for years, mostly because intelligence is still not well defined. Unfortunately I think it's becoming less well defined in the last few years (while prior to that was getting better defined) via the (Fox) Mulder Effect.

    • Maybe? I am currently going through 'artificial intelligence modern approach' by Russel&Norvig and from historical perspective alone, it seems vision would qualify.

      It is just that the language drifted a little the way it did with cyber meaning something else to post 90s kids. So now AI seems to be mostly associated with llms, but not that long ago, AI seemed to almost include just use to of an algorithm.

      I am not an expert in the field at all. I am just looking at stuff for personal growth.

    • No, even before the current AI era classical computer vision was not considered to be "AI"... because it isn't. That's just a fact.

  • Deep learning is just a subset of AI which has officially been a thing since 1956. A chess algorithm is smarter than any human yet it's just classical search.

  • Handcrafted CV algorithms and this level of autonomy is textbook AI, it’s just not Machine Learning.

  • I'm so tired of this argument. AI is a blurry term as it's used in the world. Who the fuck cares if this is "officially AI" or not? Can we just stop having this discussion?

Everyone is thinking the same thing reading that headline. In a stroke of comedic genius the link still says "bodies" instead of "people".

  • I’m amazed this problem isn’t fixed in every CMS, or at least publishing team processes, by now (the problem is that the link slug is generated from the first title and doesn’t update when the title is updated.)

I have always wanted to get into robotics. Maybe my next evolution in stuff is robotics. I'm starting to get tired of big corporate software engineering.

These silly things are neat. I'm also really interested in the snake/worm robots that dig to find people and inflate to move rubble.

  • I had a similar thought years ago. Recently I built a CNC machine and it was a great blend of hands on, practical building work and software work. The software domain being a new challenge was fun too. It felt great to excercise existing skills in a new domain, to reach a new totally novel goal for me.

    Now I barely use it, which does suggest that building it is what I really wanted to do all along and that kind of validates that robotics is probably a hobby I'd enjoy.

As somebody familiar with the area described in the article I cannot fathom how somebody could be lost, conscious (article mentions messages were exchanged) for so long.

> The weather was unusually mild for the season, and Kelly thought he might even have time to “bag” a second Munro,

I really hate when people use very uncommon terms without defining them. (or sometimes even people's names)

It's not that I couldn't make a guess based on context, but it's distracting, and I feel like my eyes must have skipped over something and I often keep going back over the text to see what I must have missed reading.

I imagine this is sometimes caused by sloppy editing, especially when they refer to a last name of a person who has yet to be introduced in the article, but I think sometimes it's a deliberate choice and I object.

  • Honest question, what was the most confusing part for you? I am guessing bag as that one might be more obscure but even then in the context I think its guessable but maybe a struggle for non-native english speakers. Munro seems difficult but since your selective quote makes it worse imo.

    "...a second Munro, as the Scottish mountains above 3,000 feet are known."

    The opening paragraph describes him climbing/hiking a mountain in Scotland. "His plan was to climb Creise, a 1,100-meter-high peak overlooking Glen Etive...". Which then leads into him trying to "bag" a second one.

    Just a counterpoint that it does not feel like sloppy editing at all. I struggle to see what would be difficult here for native speakers.

  • Peak bagging is a very common term in the outdoor sports world. This complaint is like a non-tech person reading a Wired article that mentions JSON and complaining that there's no explainer.

  • They introduce Charlie Kelly the previous paragraph, explain what a Munro is right after that and use quotes around “bag”. What else could you expect? “Bag” is extremely common in many industries and they defined both other different terms.

    You just ripped on an editor for absolutely no reason.

  • The word to "bag" may be more common in this context but it's not exclusive to it nor very uncommon, at least in North America. You might say "they bagged a record in the 4x400m relay" or "we bagged the contract" or another form like "that objective is in the bag." I think it's etymologically derived from hunting (literally putting game in a bag) but at this point it's just a word.