This is it. The pinnacle of modern robotics research. A company called AI Robotics has successfully created an ::ahem:: fully functional robotic woman that will be available for purchase as of June 11. When I say fully functional, I don’t just mean in the bedroom… According to AI Robotics, the robot (named “LISA”) can also cook meals, go shopping, do chores, give massages, and dress and recharge herself automatically. She’s got an IQ of 130 and can make conversation about news, traveling, culture and music. Have a look:

LISA uses proprietary RKS (”Recognition Krax System”) technology, which helps her recognize and respond to vocal, tactile, and visual stimuli. Her skin contains thousands of sensors, she has cameras in her eyes that can rapidly analyze forms, colors, and gestures and relate them all to an internal database, and she has the ability to learn augmented with a dedicated internal wireless internet connection. So, in some respects, she’s actually better than a human, and it won’t surprise you that I’ve already signed up to order one. I wonder if my real life girlfriend will be upset… She’ll just have to wait for the male version, I guess, which should be available soon.

[ AI Robotics ] VIA [ Communist Robot ]

Just a few days after I posted a YouTube video of some b-roll of Honda’s ASIMO conducting the Detroit Symphony Orchestra, Honda has gone live with their very own YouTube channel. There’s some interesting stuff on there, including this video of ASIMO showing off at the Kennedy Space Center:

I guess it’s pretty obvious that the people who program ASIMO aren’t the most creative when it comes to dance moves. You’d think that at the very least, he’d be able to do an amazing robot, but so far he can’t hold a candle to this guy.

[ ASIMO ]

Sylvain Calinon, who we’ve featured on BotJunkie before (twice, in fact), sent us an email talking about some new work from the Learning Algorithms and Systems Laboratory at EPFL. They’ve been working on robot (controlled with a Wiimote) that can be interactively taught to perform tasks. In this case, the robot is taught to prepare an omelette, which is way WAY more impressive than it sounds. Check it out:

Teaching a robot in this manner can obviously be frustrating at times, but research like this is critical for practical home robotics. Relying on custom programming just isn’t versatile enough for the tasks that robots are going to be asked to do, and interactive (and adaptive!) learning is, after all, how we humans teach ourselves and each other. Ideally, a future iteration of this robot would simply be able to watch you perform a task, and interpolate how to do it itself, or even how to provide meaningful assistance. And even more ideally, they’ll be just as spunky as this little guy.

[ LASA @ EPFL ] (Thanks, Sylvain C.)

This is Nexi, a small humanoid robot designed by MIT Media Lab’s Personal Robotics Group. Nexi is an MDS robot, which stands for “Mobile/Dexterous/Social.” And indeed, Nexi is all of these things. She moves around on a pair of dynamically self-balancing wheels (like a Segway) which are part of a uBot5 platform, something I saw in action last year at RoboDevelopment. Her arms, wrists, and hands are fully controllable, allowing her to grip and lift objects of up to 10 pounds, and her face features an array of movable (and quite expressive) features including eyes, eyebrows, eyelids, and mouth. Each eye has a color camera, and an IR system in her forhead lets Nexi make 3D maps of her environment.

So what’s Nexi for? It seems a little nebulous: “the purpose of this platform is to support research and education goals in human-robot interaction, teaming, and social learning.” The general idea seems to be that small, capable, social platforms like Nexi would be well suited for working with people in their homes, which is certainly true, but until such time as it’s a commercial possibility, Nexi will be used for research into human-robot interaction.

[ Nexi ] VIA [ Suicide Bots ]

There’s some kind of car race thing going on in Long Beach on April 20… Toyta Grand Prix? Well, whatever, the only car race I know of (or actually care about) is the one run by DARPA, involving totally autonomous cars. Luckily, Toyota is starting to clue in to what’s really exciting, and have added an autonomous race to this year’s lineup. There will be 3 vehicles, each one of which has already proved its mettle at the Urban Challenge, including Boss (the winner from CMU), Junior (the runner up from Stanford), and Ben, the sixth place finisher from UPenn and Lehigh. They’ve only been asked to do one lap of the 1.97-mile, 11-turn circuit (without obstacles, I assume), with the quickest time winning. It may be less challenging than the urban course from last year, but one thing’s for sure: it’ll be a heck of a lot faster.

[ GPLB ] VIA [ Crave ]

Also at the Human Robot Interaction Conference this year, Japan’s ATR Labs presented the results of a study on how quickly domestic robots respond to verbal commands. The study was designed to test how long it takes for people to get fed up waiting for the robot to do something. Obviously, an instant response is ideal, and the study suggested that any delay longer than about 2 seconds got frustrating. The interesting thing, however, was that if the robot made some kind of ‘filler’ noises (like “well” or “er”), delays of a few seconds went unnoticed. This is how we humans buy ourselves time while we think something through, and if domestic robots need to take a few extra seconds to process a complex verbal command, it looks like the solution is to have them just yammer on for a little while to distract us. If nothing else, it’ll certainly make them seem more human.

VIA [ New Scientist ]

The 3rd annual Human-Robot Interaction Conference was held in Amsterdam about a week ago. The winner of the Lego Mindstorms Student Design Competition was Phobot, a cute little robot who’s scared of things:

Phobot expresses himself through sounds, postures, movement, simulated heartbeats, and movable eyebrows. He runs a program that emulates a learning process: in the beginning, Phobot is naturally curious, but scared of even small objects. If you hold his hand and assure him that everything is okay (aww!) he’ll get used to an object of a given size, and start to dance a little bit. Using this method, you can accustom him to larger and larger objects until he becomes invincible. Phobot was conceived as a companion to children who are scared of things; the idea is that by helping Phobot to overcome his fears, it will be easier for the child to understand their own.

[ Phobot ] VIA [ Physorg ]

If, like me, you trust your computer more than any human, you’ve probably got some issues goin’ on. And what better way to address them than with a robot psychologist on the internet? MindMentor is an interactive neurolinguistic program that uses projective therapy, provocative therapy, client-centered therapy, and Pavlovian conditioning (!) to solve problems like “stress, family problems, relationship problems, motivation problems, life mission questions, sleepless nights, worrying, conflicts with friends or colleagues, et cetera. He cannot cure severe disorders like schizophrenia, bipolar disorder, hard drug addiction, et cetera.” There aren’t many details on what actually happens during the therapy sessions, but apparently, the robot does something besides regurgitate what you tell it. A one hour session will cost you $8, and you’ll get a discount on ten sessions.

Now, I have to say, this all seems just a little bit fishy. I mean, how do I know that there’s an actual robot at the other end of this, and not some measly human? It’s kinda like phone sex… How do you know you’re talking to a girl who is in fact sexy? And a girl? It’s a big problem. Or so I’ve heard. I’m skeptical for other reasons too, of course. The designers/programmers claim that MindMentor “solved people’s problems for 47% on average.” It makes me wonder how often people can solve their own problems by spending a solid hour typing them out, and whether or not it’s worth paying $8 for the privilege of doing so.

Hopefully, version 2.0 will be a robot psychiatrist, who can actually prescribe medications for my, um, ADD. And depression. And chronic pain.

[ MindMentor ] VIA [ Gearfuse ]

According to Defense News, the Israeli military is in the early stages of planning a robotic defense system with the ability to “take over completely” from humans. We’re not just talking automated weaponry here… this system would be a command umbrella, able to take control of automated and semi-autonomous systems like interceptor missile launchers in the event of “attacks that exceed physiological limits of human command.” As far as I know, this is the first system that’s being designed to operate (at least partially) without any humans in the loop at all. Some meat from the article:

“It will be designed for man-in-the-loop as well as autonomous operations,” said Milo, the officer spearheading the vision within Israel’s user community. “But right now, our emphasis is on algorithms, not autonomy. Man-machine interface is the name of the game, because the more clever we make the interface, the more successful we’ll be in providing operators and commanders the situational awareness they’ll need to make very tough decisions…”

“Our approach cannot be based exclusively on man-in-the-loop, nor can it rely only on the opposite. Rather, we need to build an operational concept and a system that is flexible and situationally dependent,” Milo said.

In the future, and “under very complex scenarios,” Milo said, the envisioned super system would be able to generate a level of supreme situational awareness and snap intuitive capabilities that could surpass the very best wartime commanders.

“We’re talking about something that sees everything and calculates everything and makes decisions that can only be made through a real revolution in BMC4ISR [Battle Management/Command, Control, Communications, Computers, Intelligence, Surveillance, and Reconnaissance],” he said. “We’re not there yet, and it could take a decade. But this is our vision and we’re running in that direction.”

There’s no denying the advantages of having a system like this in place. And even if humans are available to be in command, I have no doubt that Skynet the system will be capable of making more timely and efficient decisions than humans, because that’s what robots are good at. What robots are NOT good at, however, are making decisions like these… And we’re not talking about computer games, here. That could have been WW3, and it’s not the first time we’ve come close thanks to a computer error. I’m sure Israel is cognizant of the risks involved in giving control of weapons to a computer, I just fervently hope they’ve watched all the relevant movies and know what they’re getting themselves into.

VIA [ Danger Room ]

Robots are notoriously lousy at coping with unfamiliar situations. One of the major obstacles to having robots assist us at home is that they’re going to need to be flexible, and home users aren’t going to want to sit down and write a new servo control subroutine every time their robot needs to perform a task for which it has not been preprogrammed. The robot in the video above is able to learn via observation and replication, which is how we humans figure things out. The HOAP-3 robot, made by Fujitsu, has been programmed by Sylvain Calinon and other researchers at the Swiss Federal Institute of Technology in Lausanne to watch how a human performs an action, and then duplicate that action. If the robot doesn’t get it right, it can learn on the fly as a human corrects its motions, which (if you think about it) is pretty much how you were taught to write, throw a baseball, ride a bike, and so forth. I’d much rather have a dumb robot with the ability to learn than a smart robot that I can’t teach new things to, especially if the teaching process is no more complicated than a game of Simon says. Er, make that Evan says.

[ Sylvain Calinon’s Homepage ] VIA [ New Scientist ]