There’s only really one way to figure out how much is too much when it comes to the more, shall we say, direct (and maybe unintentional) forms of robot human interaction. In other words, to figure out how hard a robot can punch a human before said human a.) complains or 2.) suffers permanent injury, you kinda just have to get a robot and start punchin’ some humans. A robotics lab in Slovenia has made this happen, using a small robot arm with both blunt and sharp attachments to strike humans at varying velocities to correlate velocity with pain threshold.

“We are taking the first steps to defining the limits of the speed and acceleration of robots, and the ideal size and shape of the tools they use, so they can safely interact with humans.”

The overall idea seems to be that the information gathered from this research will be used to restrict robots working around humans to moving at what is determined to be a safe velocity.

On one hand, I understand why this research is relevant and potentially important. But at the same time, I feel like limiting robots in this way is to some extent counter-productive, and is sort of akin to limiting the speed of traffic on a road when humans are walking on the sidewalk. For the humans, the idea is that you don’t just walk out into the middle of the road. And for the cars, they rely on their ability to stop quickly if a human does, for some reason, end up in their path. And I don’t see why robots couldn’t operate the same way… I mean, why restrict robots to moving at a speed where hitting humans won’t cause permanent injury when you could just program the robot to avoid hitting people entirely? I think the data that are being collected here are important, but I’m just not sure whether they should be applied in the manner that the researchers are suggesting.

Where I think it SHOULD be applied is to adaptive safety solutions like ABB’s SafeMove system. We wrote about it last year, but here’s the video:

See? Just keep the robot from hitting people in the first place, and everybody will end up much happier. It’s worth mentioning, of course, that accidents will happen… But it’s important to remember that humans are the ones programming the robots, and ultimately, any mistakes that they make can be traced back to human error.

For a slightly different take on this research, this Register article (sent to us by Aaron) offers what I can only assume is a tabloid article writer’s perspective on robot human interaction… I’d be outraged if it wasn’t so laughable.

VIA [ New Scientist ]

Many would say that Isaac Asimov wrote the book on robot human interaction, and it’s pretty cool to see that he wasn’t just thinking about the future of science, but also about the present. In this video, he’s introducing the Man-Mate industrial manipulator, developed by General Electric in 1969. The arm in the video is obviously just a demo unit, and what you can’t see is that the arm is apparently providing some sort of force-feedback, which GE called Sens-A-Lift.

This particular model, the CAM 1600, can lift up to 750 pounds and has a 22 foot horizontal reach and a 15 foot vertical reach. That’s awesome and all, but what Mr. Asimov seems to be talking about more specifically is a full body exoskeleton, which we’re only now starting to see in the form of systems from Sarcos/Raytheon, Cyberdyne, Lockheed Martin, and of course Berkeley Bionics. This kinda means that we’re starting to enter the future imagined by the likes of Asimov and Heinlein… Just try to wrap your mind around that.

Also, Isaac’s mutton chops are totally making me hungry for sci-fi.

VIA [ Cybernetic Zoo ]

Juanma Oyarzábal and Álvaro Amor are two students of Computer Engineering at the University of Deusto in Bilbao (that’s in Spain). For their final year project, they created this small swarm of construction robots who cooperate to build a structure (mostly) autonomously. The long-term concept is that robots will be able take over many of the repeatable heavy lifting tasks on a construction site with minimal human supervision.

Thanks to this new way of looking at construction, costs and construction time can be reduced considerably, and it exempts the workers of the most arduous tasks so that they can focus on more rewarding tasks such as finishing facilities and monitoring that everything runs properly.

While this is the first swarm of constructobots I’ve seen, getting robots to build structures by themselves is something we’ve seen a few times before around here… There was that robot bricklayer that’s able to lay bricks so precisely that they create pictures out of reflected sunlight, and the robot that can “print” an entire custom house out of concrete in a single day.

Incidentally, I love how the video looks like it was shot in a dorm room, and evidently they only had access to one color sensor that had to be shared between four robots. Mad props on the suit (gold cufflinks!) and fancy shoes… Way to keep it classy guys.

Thanks Juanma and Álvaro!

This is neither the first, nor the best, robot dancing video we’ve seen, but what I like about this one is how unapologetically not made for dancing this robot is. It’s an industrial robot, in an industrial setting, and you get the sense that it might be neglecting its duty as boxes pass it by in the background while it’s groovin’ out.

So why is it dancing, you ask? Because it’s a robot, and if you have access to a robot, it’s basically impossible to keep yourself from finding cool (albeit non-productive) stuff to do with it.

VIA [ YouTube ]

Getting oil out of water isn’t that hard, on principle. What is hard is getting a huge amount of oil out of an even huger amount of water. If you think about it, this is really a perfect task for a swarm of robots, since it’s simple and repeatable and just needs to be done over and over (and over and over and over) again. With this in mind, MIT’s Senseable City Lab has created Seaswarm, a swarm of networked oil spill cleanup robots:

Seaswarm is designed to be simple, cheap, and efficient. To collect oil, the robots use a wide belt covered in a special hydrophobic nanofabric (about the consistency of a paper towel) that sucks 20 times its own weight in oil (and other pollutants) out of water. The belt moves around like a treadmill, which passes the befouled nanofabric back to be cleaned while simultaneously propelling the robot forward. The video talks about heat being used to separate oil from the nanofiber, while the description on the Seaswarm website makes it seems like the oil is squeezed out using rollers… Whatever floats your robot, I guess.

What I’m not too sure about is where all of that captured oil goes. Using their solar panels for power the bots can collect for several weeks at a time, and the more oil they collect, the heavier they’ll get, and the more energy it’ll take to keep them moving. The website does mention that the oil will be ‘digested,’ which I assume implies microbes, but they’ll either have to collect oil very slowly or have some wicked crazy hungry bugs to be able to get around the problem.

Seaswarm is designed from the water up to utilize swarm behaviors. To combat a spill the size of the recent one in the gulf, about 5000 fully autonomous Seaswarm units would cooperate for a month, using GPS and WiFi to organize themselves for most efficient coverage. And of course, you get all the usual swarm benefits of scalability, adaptability, and robustness.

While Seaswarm as a whole is currently just a concept, they have built an actual working prototype (in the pic above), which was just tested out on the Charles river in Boston, I guess because the designers figured they’d teach the robot some humility when it comes to pollution clean-up.

[ Seaswarm ]

A robot stole my job back in 2008, and things are just getting worse, according to this infographic from Focus magazine. It’s a long graphic, so check out the full thing after the break. (Read more…)

Inspecting power lines is a perfect job for a robot, if for no other reason than it’s a nearly (nearly) impossible job for a human to do, especially when the lines are live, which they need to be for me to be writing this post. One of the issues with robot line inspectors, however, is that power lines tend to be held up by towery things and festooned with all manner of objects to keep wayward pilots from getting tangled up in them. In other words, there are obstacles to what would otherwise be one unbroken and extremely robot-friendly cable.

So, the obvious solution is to design a robot that’s able to get around said obstacles in a creative manner. The above video, about Hydro-Québec’s LineScout, is one of those PR puff pieces that is unfortunately pretty lame, with 6 minutes of talking heads restating over and over how fabulous the robot is without actually discussing what (in the technological sense) makes it so cool. If you watch closely, you do get to see how LineScout can traverse obstacles by detaching part of itself from the power line, moving around the obstacle, and grabbing back on. It can also operate while the lines are energized, making it minimally invasive on existing infrastructure.

Here’s another approach to the problem by the HiBot Expliner, which uses its own body as a counterweight:

It’s great that they’ve got robots inspecting power lines now, although the old way was kinda way, way cooler. Sorry robots, I love you, but sometimes progress is takes a little bit of the magic out of life.

[ HiBot Expliner ] AND [ Hydro-Québec LineScout ] VIA [ IEEE ]

We’re already familiar with Husqvarna’s robot lawn mower, but this beastly thing looks like it’s probably a lot more fun, since in addition to totally destroying your lawn, it’ll happy totally destroy anything else you decide to point it at. The Husqvarna DXR 310 is purpose built for tearing things (like buildings) to pieces. It weighs 2 tons and its extendable arm can be equipped with claws and jackhammers and other exciting stuff. It’s remote controlled, but it seems like if there was any circumstance where giving a 2-ton robot autonomy wouldn’t be a huge mistake, it would be one where the utter destruction of everything around it was actually the intention as opposed to just an unfortunate side-effect.

[ DXR 310 ] VIA [ RFJ ]

To kick off Hannover Messe (an industrial fair in Germany) last month, ABB taught one of their industrial robots to beat on stuff. Musically.

The robot percussionist was accompanied in its performance by a “real” cellist who commented “the industrial robot is a congenial partner and dominated his instruments with some virtuosity”.

The robot wasn’t an autonomous musician, at least not entirely… It looked like the guy controlling the robot was telling it what sequence of pre-programmed movements to make. If I’m not mistaken, he’s doing some stuff with a Tenori-On also, but I have no idea what… Maybe mixing background loops on the fly?

I love it when companies who build serious industrial robots take a step back and say, “let’s just do something cool,” because robots are cool, and they’re good for much much more than just totally taking over from us humans.

[ ABB ]

Thanks Maanasi!

Robots may be getting laid-off left and right due to the lousy economy, but at least they’re easily retrainable, and as long as you’ve got some programming know-how, you can get them to do all kinds of exciting things. One company called Autofuss has taken a trio of ex-industrial six axis Fanuc s430iL robots and turned them into professional camerabots. As you’d expect, they’re very fast and extremely precise, and as long as you don’t need them to, uh, move or anything, they can produce spectacularly smooth and complex shots over and over without screwing up or complaining.

Recently, the Autofuss robots (they’re named Puck, Gilda, and Rosie) worked on an ad for Louis Vuitton. The robots aren’t actually doing much fancy moving, but the commercial has Sally Ride and Buzz Aldrin and Jim Lovell in it, so that’s pretty awesome all by itself. Check it out, after the jump. (Read more…)