As a particular “special military operation” (aka war) approaches a year, please share a thought for the considerable suffering each day. I’ll quote WOPR from WarGames “The only way to win is not to play“, and very much applies in this instance to Russia.
But that quote doesn’t mean we can’t play, it meant war so let’s troll a little.
On the 29th of October 2022, somebody took offence to the Russians parking their ships in Ukraine (Sevastopol, Crimea is part of Ukraine, not Russia; sham referendums and illegal annexing doesn’t count) and they decided to do something about it.
The audacious plan went like this.
The explosive boat drones were most likely navigating using GPS, then steering radio-control style (possibly via StarLink satellites) with the operator seeing through a camera, but tbh, I have no idea.
I got the impression that the invaders weren’t very impressed, leaving the 2nd best army in Ukraine (aka Russia) looking inept yet again.
I wish to qualify in my trolling that I am not anti-Russian. Russia is a great nation and I welcome the day we can all go back to getting along peacefully. But whilst their army rape, murder, and terrorise Ukraine, they are fair game being trolled.
Please don’t use the comments to fight/argue over the legitimacy or not of the “3-day” special military operation, I will delete them.
Before we start, I cannot help but notice it seems every manufacturer feels they need “AI” to sell their product. Here’s an example of the world gone mad, a toaster with AI. Yes, you read that right.
“aims to “provide a system that can consistently toast bread products to the desired level of browning”, using a neural network trained on thousands of images of toasted bread taken directly through the toaster.”
Most of us have burnt our toast, so I am sure we all understand the problem he’s trying to solve. It might be the best thing since sliced bread. Sorry about that pun.
With my toaster, the first two slices of bread on setting “2” provide the desired level of toasting. If I then cook two more slices on the same setting straight after, it burns the toast.
Why?
The toaster cooking element heats up whilst cooking the first two slices. For the subsequent slices, it is already hot, and therefore the cooking “time” needs to be reduced.
Not rocket science and appears solvable by simple calibration (settings = degree of cooked, not the amount of time heating). If it wasn’t a mechanical timer, the way to fix is simply to have the toaster warm itself up first, then insert toast.
A manual and un-environmentally-friendly fix (and a tad inefficient given today’s electricity prices) is to put the toaster on without toast for it to warm up. For cooking, lower the amount from e.g for mine “2” to “1.5”. Consecutive toast should be caramelised similarly.
If you are curious about the science behind perfect toast, here’s a link, one of many the internet has to offer… I don’t see how AI is going to be a value-for-money improvement.
What has toast got to do with ships? Because some unwanted ships were nearly “toast” afterwards. #troll
Our “drone” boat challenge
- Drone boat uses GPS points to get to the “kill box”
- In the “kill box” the AI steers the drone boat to target the ship
For our little challenge, we’ll go for authenticity, yes this is the Russian corvette “Admiral Makarov”. I would have used the “Moskva”, but it’s currently being used for an undersea special operation.
The picture below was taken before the Makarov was left parked where it shouldn’t be.
Let’s put some limitations on our approach.
I chose to avoid using libraries, and tons of code. YOLO is absolutely the kind of AI that I want. YOLO recognises objects and gives bounding boxes. We could train it on the ship of interest (or a multitude of ships), and if it sees them we steer towards the centre of the bounding box.
So what can we achieve without it?
It’s hard to appreciate from the image, but the AI is steering the drone. The upside-down “v” is the position that the AI is instructing the boat to head. Sure enough, it homes in on the Makarov, destroying it (sadly, only virtually).
On page 2 we’ll look at this in more detail.
