Planning a hostile takeover of your local swimming pool? This might help: [Dr Anders Lyhne Christensen] sent us a note about his work at the BioMachines Lab of the Institute of Telecommunications in Portugal. They have been building a swarm of robot boats to experiment with autonomous swarms, with some excellent results.
In an autonomous swarm, each robot makes its own decisions and talks to its neighbors, and the combined behavior of the swarm produces an overall behavior, like ants in a nest. They’ve created swarms that can autonomously navigate, patrol an area or monitor the temperature in an area and return to base to report the results. In an excellent video, [Anders] outlines how they used computational evolution to create these behaviors, randomly mutating a neural net to find the best approach, which is then sent to the real boats.
Perhaps coolest of all: the whole project is open source, with the brains of each boat running on a Raspberry Pi, and a CNC milled foam hull with 3D printed component mounts. Each boat costs about 300 Euro (about $340), but you could reduce the cost a bit by salvaging components and once the less-expensive Pi Zero becomes obtainable. This project will no doubt be useful for many an evil genius who is sick of being splashed by the toughs at the local pool: a swarm of killer robots surrounding them would be an excellent way to keep them at bay.
Robot swarms are definitely cool, but has any come up with a useful application for them yet?
When I first saw this, I was thinking it could be used to address the issue of piracy at sea. It seems the military already has this operational and tested. (See below comment from ‘DRM’)
Detecting anything, really. Searching for things in the ocean is difficult. Even large things like submarines. Nevermind small things like a wayward leisure-craft.
Having a swarm of robots that can work together autonomously to completely and reliably search any given 3d section of ocean would be a big deal.
Small means even the tiniest wave will obscure it’s range of vision (optical as well as other types like RADAR), so nah it won’t work like you envision.
You might argue SONAR will work but you’d be better served with a swarm of mini submarines then, with floaters for antenna purposes. But even then, you get power and equipment-size issues to have that working in a useful manner.
Obviously, this is a small proof-of-concept sized device to work out algorithms and such. If you’re going to be doing things like searching for something in the ocean, obviously it would be with bigger craft. The military is experimenting extensively with launch-sized autonomous craft, so clearly the concept has at least some merit.
Have you not seen those drone swarms with LEDs where they align themselves to create different images? How’s that not useful?
They are useful for publishing papers. And since it is just a broad catchall buzzword, “the swarm”, academics love it, since it allows you to be *super* dishonest about what is actually going on, but everyone focuses on all the robots moving around simultaneously.
So @AndrewPullin, what is actually going on?
You’re much to mild mannered. Heck with splashed with water and sending these out, have them detect urine and home in on the source with some spinning blades…. :P
And kill all the fish that refuse to leave the ocean :D
I had an idea for a robot boat that has a GPS, a solar panel, and a destination. It does its best to make it there.
http://hackaday.com/2015/05/25/project-sea-rendering-autonomously-renders-sea-bottoms/
This is cool. I was thinking, since people make balloons that cross oceans. Would it be possible to cover it with solar panels, and make it go really long distances? Or the salt would corrode everything?
It is possible to seal things up really well so that the salt doesn’t get in. Its certainly possible. Collecting enough energy for propulsion while not collecting too much wind(unless it blowing in the right direction) seems like it would be the challenge.
Needless to say there is a military use:
https://www.youtube.com/watch?v=ITTvgkO2Xw4
Isn’t there always a military use for things?
That sounds like a drinking game.
yo dawg: boats
no IMU ?
Just a guess, the pitching and bobbing of floating in the water may make IMU measurements not very useful.
That, and the fact that the robots are basically moving in 2D (they do not go underwater). A simple digital compass is enough to find the heading
Why Wifi?
http://hackaday.com/2015/09/13/arduino-masters-ham-radio-digital-mode/
How much data would these types of systems need to transfer between each other?
Path of least resistance. It was the simplest to implement and very cheap. Future versions will have to rely on better communication technology. Regarding data, we are currently just broadcasting some basic position and heading data, but in the future the robots could share more complex sensory data among themselves.
Wifi is cheap, easy and reliable over short distances. It also doesn’t require a licence. The main downside is that its a bit power hungry.
Anything custom built would have to operate under an amateur radio licence and the restrictions that go with it. Mainly no commercial use and no encryption. To operate something custom commercially you would have to get the design FCC approved and then buy a licence to operate it.
Zigbee could also work but the range isn’t that much better than wifi.
What about just measuring water quality?!
Yup, we are going in that direction. Check our paper for some of the results of a temperature monitoring mission: http://arxiv.org/pdf/1511.03154.pdf
It’s nice to see Portuguese technology in hackaday for the first time. Good job.
Depth mapping! With like 10 such boats at Chinese prices anglers could quickly map the depth distribution in their favorite lakes!
But as far as robot swarms are concerned: why not just simulate? I hardly see any point in actual deployment if there is no real use. nobody wants 300 Dollar boats measuring temperature.
There are plenty of studies of studies with simulated swarms. Since our goal is to be able to use real swarms for useful missions, we needed to validate if swarm control synthesis techniques can actually be applied in real-world environments. Water temperature is too simplistic as a real application of a swarm, but environmental monitoring of more parameters is a huge application domain. This is just a first step.
Of course. Pretty cool, no question. I just need a few that will find the fish for me when released on a lake…
What is the motor / esc you’re using? Brushed? And runtime given what battery? What was the design goal of the hull? Lowest energy consumption? Stability?
Whether practical or not, this is probably one of the better introductory explanations of the concepts of swarm robotics and of computational evolution that I’ve seen.