Feel like a break from Martian Spiders? Okay. Here ya go. Spiders will be back in a few days!

As I type this, there are people who live a thousand miles south of me, who are pointing their phones at the sky and taking spectacular pictures of "Northern" lights. Yup. The Sun fired off yet another monster flare which, on hitting the Earth's magnetic field is setting up the light show of a life time¹.

And we can't see it. Here, near the shores of Lake Ontario, the cloud cover is too thick. Sky? What sky? You mean there's something else up there besides those black fluffy things? Who knew? It was the same a few months ago when that eclipse happened. My wife and I went outside to watch it. It got dark. It was kind of creepy when the horses and cows next door ran for cover. But did we actually see anything?

Nah.

So I thought, maybe if I had a drone, I could send it down the road until it saw some nice aurorae². It could send me pictures over the Internet. If it wasn't too far away, maybe I could hop in the car and go there. Wouldn't that be nice?

I think it's also illegal. The FAA doesn't like people flying drones around just anywhere they want. Something about "safety" or some such silliness. Oh well. But what can you do with a drone? That was a ridiculously long segue into a mention of a story reported by Wired last week about a do-it-yourself rescue drone. This Homemade Drone Software Finds People When Search and Rescue Teams Can’t.

The short version of the story is that some guys got together and wrote and tested software that would use a drone mounted camera to find people, probably injured or dead people, who had gone missing in the wilderness. The UK wilderness. It's hard to believe that the UK has wilderness but apparently it does. And people go missing there frequently. Also often. The software they wrote sounds complicated. It had to compensate for the motion of the drone and the uneven terrain to piece together a good picture of the ground below. Their technique for finding people is brilliant. It doesn't look for people. It looks for spots of color that are different from the surroundings.

So if you hope for someone to use a drone to find you when you're lost, don't where camo.

Sometimes the system finds things they don't want. But who really cares? It finds people when human eyes have missed them. A couple false positives is a small price to pay for one really good hit.

The thing I love about this story is that these guys aren't trying to sell anything and they didn't look for some kind of government grant to finance their prototype. They just went ahead and did it because they wanted to. They were all volunteers in rescue operations who wanted to do better rescues.

I probably noticed the story because the devastation from Hurricane Helene was all over the news at the time³ and I immediately thought that they could use some such technology in places along that storm's track. In theory, you could add infrared (IR) imaging to make it a little easier to find people trapped inside collapsed houses or stores, or cars washed downstream, or other places we'd all rather not be. I've only done a tiny amount of work with vision software and I think it would be very hard to make that work. But it would be worth it.

Plus it's easier to fly a drone over a lake that used to be a parking lot, looking for people, than to train a dog to swim out there and find them. Then again, you don't have to write fancy software to get a dog to learn interesting and potentially useful stuff. You just have to have a good supply of dog treats. And some skill at training them, of course⁴. But with all that, you can teach dogs all kinds of interesting things. Drones, generally don't learn anything. Maybe someone will write some software for that next week.

The missing-person-search drone actually just uses the drone to take pictures. The analysis of the pictures is done offline, after downloading the pictures to a computer. That shows an interesting problem in working with this kind of tech. Analyzing images takes a lot of memory and a lot of processing power. When you're using off-the-shelf equipment, that means it takes a dedicated computer. Not one of the cheap ones, either. Off-the-shelf drones just don't have the oomph (to use a technical term) to do the processing. I suppose you could stream the pictures back to the processing machine over the Internet and maybe do some real-time analytics. But when you're out in the back woods somewhere, you can't count on having a good wi-fi connection, can you?

Is this going too deep into the weeds? Sorry if it is. I'm a tech guy. Sometimes I like the weeds. The technical ones, I mean. Not the ones in the back yard. Anyway, one of the things that makes tech more difficult than you might think, is the trade-offs you have to make every step of the way. If your drone is too heavy, it won't fly. If your computer is too cheap, it won't do the analyses you want. If the body is buried in a shallow grave, you probably want a dog with a strong doggy nose instead of a drone with a camera.

You can't have everything⁵. You can study all the astronomy you like, but when the clouds move in, you're not going to see that aurora no how. And you can build software to find missing people out of spare parts and a lot of work. But they might not be alive anymore when you find them. Maybe the next drone will be faster. Maybe it will even have a defibrillator built in. But that would probably make it fly slower and then ...

Never mind. I love that we live in an age when you can buy some parts and then apply your brain and build something amazing. You just have to figure it out.

Here's that prompt: "A cartoon-style scene of a drone flying over a lush, green forest. The drone is equipped with large, exaggerated binoculars, looking down at the ground. Below, a group of cheerful hikers are waving up at the drone, smiling. The forest is bright and colorful, with tall trees, a few scattered clouds, and a sunny sky. The scene has a playful, whimsical feel, with the drone hovering high above the waving hikers."