Hacker News Reader

Comment by nateberkopec | Hacker News Reader

I volunteered for a SAR team in the US southwest for 5 years, the last half of that or so we had an active drone unit.

Drones are good at covering terrain that's difficult to traverse on foot. Canyons, cliff walls, the like. As some of the rescues on the site show, they're also good for getting another angle that's not human-being eye level, which is sometimes all you need to spot a clue or subject.

Drones are not very good at covering large areas of ground quickly. It's also extremely difficult to spot anything small than an entire human being on the drone's camera. That means you miss valuable things like bootprints, pieces of equipment, etc.

They're a very useful tool in the toolbox, but I don't see them replacing human beings until image recognition technology gets another level-up. "Recognize a human being or signs of one with the background of literally any possible terrain on Earth" is a bit beyond what's field-deployable at the moment.

View on HN

> It's also extremely difficult to spot anything small than an entire human being on the drone's camera. That

One of the articles mentioned a local fire dept had a drone with thermals which helped find a 14yr old boy lost in a remote area. That would certainly help. From the videos I’ve seen from Ukraine thermals are a huge advantage for finding people.

That may be true. My unit didn't have a thermal camera setup. However, we're not always looking for live subjects that will give off a thermal signature, and thermal cameras don't help with spotting inanimate objects. I would say we found signs of the subject's passage (equipment, footsteps) 80% of the time before finding the actual subject.

I worked with an imaging and aerial mission company in Colorado, and indeed thermal was used in the vast majority of search and rescue missions. Not useful for recovery, but for time sensitive operations it was critical.

> Drones are not very good at covering large areas of ground quickly.

I work with imaging drones in agriculture and every time we run the numbers the answer is almost always fixed wing. We have multirotors that have no problem flying at 80+km/h, but the battery life kills our ability to cover broad areas of land without having to land and swap batteries. With fixed-wing you get (depending on many factors) a ~10:1 efficiency improvement but also need significantly more skill as a pilot and potentially more infrastructure (e.g. a viable runway of some kind). VTOL fixed wings have some potential, but the takeoff and landing burns so much battery that you lose a lot of the advantages of the fixed wing.

Hmm, what about a launcher? Slingshot the drone any number of ways, then "landing" can be handled by an arrestor net.

Yeah, for sure! That's exactly the approach that Zipline has taken: https://www.youtube.com/watch?v=FeSCEalMOL8&t=60s

There are some practicality points to it. For example, their system works awesome when you have a fixed base but might be impractical when launching with gear that's towable with half-ton truck.

I am developing CV systems for agribusiness and have been grappling with this issue. The solution seems to be a combination of the two systems. Quads are great for some tasks like herd detection/sorting/counting and spotting problems in planting. For crop dusting, seed planting, geo surveys in general you go with fixed wings.

Yeah, one of the approaches we're looking at involves using fixed wings to cover 98% of our surveys and "mopping up" the areas we can't hit with a quadrotor (e.g. near power lines, trees, other keepout zones)

How long before someone just gets as close as possible in a truck and deploys 100s of autonomous drones in a swarm to scan an area in detail and report anything significant?

I mean the technology for controlling drones like that is there, given these drone airshows. But it's a ton of data that needs to be processed, it may be a capacity and bandwidth issue / challenge as well.

Give it time, I'd say. Only a matter of time before there's drones autonomously mapping out big areas and computers assembling the data.

Was your drone a fixed wing or rotary wing? The battery life is always getting better, but I have to imagine the flight time on a fixed wing has got to just crush a quadcopter, although I think they're more difficult for a novice to fly.

a dual camera drone like Mavic 2ET is pretty good at spotting humans at pretty long distances even in brush/ trees

Reply on news.ycombinator.com