Researchers at the University of Colorado at Boulder are developing small, swarming robots -- dubbed by the team as "droplets" -- that will be able to accomplish a variety of tasks. Possible uses include building a space station or a satellite, self-assembling into a piece of hardware after being launched into space, or cleaning up an oil spill on Earth.
Swarm robotics is a fast-changing, quickly growing area of robotics research and development. We've reported on a swarm of "hedgehog" robots being developed by Stanford University to explore space, and swarms that can play Beethoven, or repair coral reefs. We've also reported on the robotic self-assembling pebbles developed in Daniela Rus' Distributed Robotics Laboratory at MIT, where the leader of the University of Colorado team, Assistant Professor of Computer Science Nikolaus Correll, did post-doc work.
Swarming robots, or droplets, developed by the University of Colorado, Boulder, may someday assemble telescopes and satellites in space. (Source: University of Colorado, Boulder)
The University of Colorado team has built a swarm 20 strong. The droplets form a "liquid that thinks" when they swarm together, said Correll in a press release. He plans to use the swarm of robots to demonstrate pattern recognition, sensor-based motion, and adaptive shape change, as examples of swarm-intelligent and self-assembly behaviors. These behaviors could then be transferred to much larger swarms that could carry out more complex tasks in water- or air-based environments.
The computer science research team also includes research associate Dustin Reishus and professional research assistant Nick Farrow. Together, the team has designed a basic robotic building block.
The platform will eventually be reproduced in large quantities for developing increasingly complex systems. Correll hopes to create a design methodology that will allow the swarm of robots to work as an aggregate in more complex behaviors. These might include assembling parts of an aircraft or a large space telescope.
In a video that describes the team's research (watch it below), Reishus says 10 of the droplets are now working and some of the software is written, but the robots aren't solving any useful tasks yet. "We are still just testing each individual robot, getting the very low-level communication between two robots working."
Reishus says that, after the droplets are completed, the team will have a platform that can be used for conducting various experiments with swarm robotics, whatever those might be.
Those experiments will probably be thought up by students working in a lab Correll has set up. There, students can use basic, inexpensive tools to explore and develop new applications for the robots. He expects that this will help accelerate the pace of development. The lab's research focuses on intelligent distributed systems, including sensing, actuation, computation and communication.
Aside from robotic swarms, researchers are working on large-scale, outdoor robot teams and smart materials.
William, you've posted this basic comment/idea so often that I'm starting to think you know something about AI that the rest of us don't. The last time I looked, they were nowhere near achieving the kind of thing you're suggesting. Can you tell us any specifics of who's doing self-awareness research on the cutting edge right now?
BUT, in the public domain there have been references to some university people working towards artificial intelligence, and they have included self awareness as one means of moving toward human type judgement. My advice would remain, to "Think very carefully about the ultimate effects of your creations", because sometimes the machine does not stop just because you push the stop button.
Since people are rather less predictable than computers and robots, conside the problems that we could have if those in-animate things became a lot less predictable. What if your washing machine developed an "attitude problem", rather than just a component failure?
Thanks, William. I knew people working on AI back in the early 70s and again in the 90s and 00s. Trying to make computers work like we do in terms of logical processes is still a far cry from also giving them sentience and self-awareness. But no, I don't relish the thought of a toaster or a fridge with a 'tude. My computer already seems to have that problem :)
Hi Ann. Any short list of people working on machine consciousness would have to include Ray Kurzweil. I suspect that's exactly what he is up to at Google. He already has a well thought out published theory of the source of consciousness in humans.
Hi William. Well, Google's official long term goal IS the creation of concious, self aware, minds like ours. We all hope the servers don't get any more feisty than they already are. We need to have a set of rules like Asimov's already established when it "wakes up". But, how does one do that before we know exactly how it works? And, if the machine claims self awareness, how do we know whether to believe it? My favorite line of all time from a writer in this regard was written for Serge Graystone, a 100% CG character on the short-lived (sigh) TV show Caprica. Serge was a household robot butler/security guard, and like the other characters of the show, in real life "Serge" had a Twitter account. It's probably still there. On Twitter, "Serge" answered fan's questions from his unique point of view. "Serge" claimed repeatedly that he was not self aware. Of course, the fans just stepped up the pressure to get Serge to admit that he was self aware. The writers ended the battle with this wonderful Godel sentence: "If I were self aware, don't you think I would know it?". In essence, how do we detect a mind that has become aware, but is hiding from us? Wandering back toward the topic ... If Kurzweil is correct that conciousness emerges from large amounts of simultaneous triggering with many levels of feedback, (my gross oversimplification) then perhaps the first conscious machine will be a large building made entirely of a huge swarm of tiny communicating meta-material robots whose collective purpose was to become and maintain the building. Hmmm... I remember a sci fi movie where a "grown" organic house eats its residents. Bon Appetit!
I seem to recall that Dan Simmons' proposition for how robots become separate, self-aware entities in the Endymion series is that--they don't. Instead, a sort of collective AI emerges from the internet of the future, i.e., many, many interconnected machines, and it/they spans multiple hardware. So I think a swarm makes a lot of sense as a model for that interconnected hardware, and for the AI "mind". Theoretically speaking, of course. I still think the whole thing is, and will remain, in the realm of sci-fi.
Ann, I read a quite scary science fiction book about that very thing a few years back, and the "swarm" was a collection of very small robotic things, smaller than mosquitoes, but when they started working togather they were very deadly. Plus they learned how to reproduce without human assistance.
So perhaps the solution is to avoid giving robots long-lasting power sources, to assure that any uprisings would be short lived.
William, I think you are describing Michael Crichton's "Prey." Sure sounds like it. And I think this time your tongue is firmly planted in your cheek, although self-assembling nano-robots is definitely an area of research and has been for a decade or so.
Having worked with, interviewed, socialized with, and even dated programmers, I agree that they're "not normal." But that can be a good thing: it often means way-above-average intelligence and a wicked sense of humor.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.