Geek Gestalt

MOUNTAIN VIEW, Calif.--For a few minutes Thursday, as I steered one of NASA's K10 intelligent robots across a small field of rocky, sandy terrain, I could almost imagine myself piloting the rover across the surface of Mars or the moon.

Until, that is, I realized I had pretty much no idea what I was doing, and saw that my struggles to steer the rover forward were actually sending it backward. Given that this little robot is worth at least as much as a mid-range Mercedes, I was relieved to see the eagle-eyed scientist standing a few feet away from it as it approached a group of large rocks that could send it sprawling, a switch in his hand capable of stopping it dead in its tracks.

I was spending the afternoon at NASA's Ames Research Center here, talking with Terry Fong, the director of the Intelligent Robotics Group (IRG), about the K10 rover program--an initiative designed for remote scouting operations on the moon or Mars. To be sure, the program has been around for a few years, but Fong and his team are constantly tweaking the robots, and so what I got my hands on Thursday (remotely, at least) was a great deal more sophisticated than would have been the case just a few years ago.

We had driven out to Ames' faux lunar/Martian landscape, a 40-meters by 80-meters field of rocks and dirt tucked away in a quiet corner on the western side of the giant NASA facility. I've been to Ames many times, but this was by far the most peaceful part of the grounds I've visited: the shriek of a red-tailed hawk as it soared high overhead was the only real sound besides a gentle wind.

Fong took us into a small shed just outside a small trailer, and sitting inside was K10 "Red," one of the two rovers his team has here. It's called red because it has a big patch of red on its body. Its twin, K10 "Black" was nearby, already prowling around, lost in its own lunar fantasy, different from K10 Red only in that its body is black.

The two rovers (see video below, but be prepared for substantial wind noise) are built to travel at "human walking speed," Fong explained, and can handle between 90 percent to 95 percent of the terrain here. That includes some softball-size rocks, plenty of loose sand and dirt and even a few steep inclines. Fong allowed that some of the bigger rocks, maybe soccer ball-sized, might be a problem, and the steepest part of the incline might cause the rovers to lose traction. But in general, these are sturdy little robots built to withstand some truly out-of-this-world conditions.

According to NASA, the K10 robots are crucial elements of the space program's directive to achieve more complete investigation of the moon than was possible during the Apollo program.

"Human missions to the moon will provide numerous opportunities to advance the scientific exploration of the lunar surface," a NASA brochure about the K10 robots reads. "Initially, human exploration of the moon will be for short periods of time--no more than a few weeks per year. To make use of the time between human missions, robots can be used to perform highly repetitive and long-duration tasks, such as site-mapping and science reconnaissance.

"NASA's K10 robots are designed to be remotely operated on planetary surfaces and act as scouts for human explorers. Scouting is an essential phase of fieldwork, particularly for geology, to help establish priorities and scientific objectives. Robotic scouting can improve human exploration of the moon by providing mission planners with detailed ground-level information to supplement and complement data collected by orbiting satellites."

Imagine, in the future, the Constellation program kicks in and NASA begins sending manned missions to the moon. Fong explained that the K10s would be essential to maximizing the research that could be done in between visits by astronauts. Indeed, they could be left behind after one mission and then be deployed to gather intelligence for the next manned mission, data that could complement what NASA can see with instruments in lunar orbit.

And while the K10s I saw ran off of Lithium-Ion laptop batteries with a life of about four hours, Fong said K10s that stay on the moon could run on an ongoing basis on solar, or on power cells.

Continuous navigation
To look at them, today's K10s are unchanged since their introduction eight years ago. But pop the hood, as it were, and what you find is an ever-changing Red Hat Linux-based brain. Every few months, Fong said, he and his team replace the standard PC laptops that serve as the K10s nerve centers with newer and more powerful ones. Even considering what a top-of-the-line laptop costs, a couple new computers are pretty cheap compared to the tens of thousands of dollars' worth of lidar, navigational equipment, sun trackers--which allow the robots to figure out precisely where they are, an advantage over compasses when they're being used for field tests in far northern parts of the world--3D surveying instruments and more.

Fong said that while much about K10s remains the same today as in the past, one recent innovation has been building in the ability to process data on the go, rather than what was possible in early missions on Mars, where rovers had to stop, calculate, move, stop, calculate, move and so on. That means, he said, that now, K10s can make real-time navigation decisions, progress that means they can cover ground much faster than their predecessors.

All told, explained Fong, a K10 robot is smart enough to figure out a path between point A and point B and determine which parts of the terrain it encounters it needs to skirt due to big rocks or other dangerous conditions.

In some cases, the IRG scientists will test the limits of what the K10 can do. Fong said that during trials last summer at the lunar-like Black Point Lava Flow, in Arizona, he and his team tweaked the K10s' algorithms so that the rovers would push on, despite confronting bush clusters that compute as rocks to be avoided.

"The robot says, 'Hey, it's rocks,' so we make it a lot more aggressive," said Fong.

Five football fields
One reason the K10s are so important to NASA is that they feature 3D laser scanners capable of surveying as much as 500 yards ahead and identifying and analyzing objects as small as pencil erasers. Similarly, the rovers carry downward pointing cameras that take very high-resolution pictures every few feet, images that can then be used to help the mission planners decide where and how to explore in the future.

Back in a lab at Ames, Fong reached into a large suitcase and pulled out what he called a K10 mini (see video below). This is a one-fifth size model rover that is an experiment to see what's possible on a much smaller scale.

The idea behind the mini rover, Fong said, was to try to understand what's possible with a robot that small. And while it's only the size of a small dog, he explained that its basic software was the same as its larger cousins, and was actually built up around the smallest Thinkpad laptop the IRG scientists could find.

And now, the IRG team is looking toward what's next for the K10s. And that, said Fong, is to determine how best to marry the rovers with human teams so that they can be used to support explorers instead of being autonomous. The Spirit and Opportunity Mars rovers were built to do everything on their own, the K10s can be employed for wide ranges of tasks that aren't efficient or productive for the folks in space suits. And not only that, but because the K10s can be mounted on the lunar exploration vehicles NASA expects to send its personnel around on the moon in, they can be used for research, surveying and exploration wherever the astronauts go.

But at the same time, NASA knows they have a great deal of utility as autonomous explorers, and on the moon, which is within reasonable striking distance of the Earth, there's even less risk involved than there would be on Mars with putting the K10s through rigorous paces since, even if one got damaged, the next manned lunar mission could come and fix them.

Of course, if you're part of the IRG team, or you're an astronaut trained in operating, or working with, a rover, you probably have a lot of confidence about what the robot's limits are. Why else would anyone trust you with such an expensive toy?

For me, however, clutching that joystick back at Ames, trying to get my K10 to go right, or hop over that little rock, every wrong move seems like potential catastrophe. I think it's going to be a while before anyone lets me play with one of those thing without having a guy watching every move I make with it, ready to push his big red button to keep me from sending it sprawling and costing taxpayers a Mercedes' worth of cash. And I'm OK with that.

SAN FRANCISCO--Since I was one of the first people to arrive Friday night for a preview of this weekend's cocktail robots exhibition here, I was going to get the first drink.

David Calkins, one of the organizers of the San Francisco version of Roboexotica--an event that has been taking place in Vienna, Austria, for a decade--had set up his robot, Chapek, and, determining it was ready, asked me to tell the machine what I wanted to drink.

This was after, of course, Calkins had finished getting Chapek ready to go.

"Let's see if it turns on and explodes," he said, "which it has in the past."

He flicked a switch and Chapek was ready.

"Hey," he said, pleased. "It didn't."

Chapek, which is named after Karel Capek, who coined the term "robot," is a small robot with a mischievous face, wiry metal arms, and an attached control box where you tell it what kind of cocktail you want it to mix up and serve you. The choices? Gin and orange juice, a gin martini, a vodka martini, and a screwdriver.

I'm not much of a martini man, and not wanting any gin, I switched the controller over to screwdriver, and pressed the button that would set Chapek in motion.

Sure enough, it came alive and slowly, its arm began to swing in the direction of four bottles that were suspended and awaiting martini glasses. Chapek swung the glass in front of the vodka, paused while some of the liquor poured in, and then continued on toward the O.J.

Finished, it reversed directions and headed back toward the beginning, where my hand was eagerly awaiting my beverage.

But just as it reached that point, the arm sped up, and instead of stopping and serving my drink, Chapek slammed the glass against its body, spilling my screwdriver all over itself and its interior electronics.

"I suppose I should look at the (computer) code," a chagrined Calkins said, before explaining to someone who had just wandered by that, "Chapek is being greedy again. He made the drink and then he stole it."

Indeed, while the spilling of the screwdriver had seemingly fried Chapek's microcontroller, rendering the cocktail robot useless for the moment, Calkins wasn't all that perturbed. It was clearly nothing that he or his colleagues hadn't seen before.

"Chapek, he's an alcoholic," said Simone Davalos, Calkins' co-organizer. "We can't take him anywhere."

For a decade, Roboexotica has been a growing hit in Vienna. Now, it has come to San Francisco, where it will be open to the public on Saturday from 8 p.m. to 2 a.m. at Space, 354 5th St.

On Friday night, there were only a few cocktail robots set up for the invited guests to see. So I was not able to see some of the machines that will be on display on Saturday, such as one that can test your blood alcohol content, or another that I believe is intended to blend a fruity drink while shooting out big fire.

But one that was on hand was Davalos' creation, which she called El Espanol Borracho.

This small robot is a cylinder that, if you weren't paying close attention, you might mistake for a kettle. But it's attached to a small container of high-pressure fuel and has spigots coming out of it that serve up the booze.

Davalos had the robot serving Spanish Coffee, a particularly strong mix of several alcohols, but at first she couldn't get it to do just what it was supposed to.

Clearly, fire was supposed to be involved, because she began to tell a funny story.

"Last year, I did this and set my arm on fire," she said. "And I didn't notice at first because the floor was also on fire."

She tried to get it running properly again, but still, no fire erupted. Instead, we all began to smell a very strong odor of fuel.

"David, unplug me please," Davalos urged Calkins. "Quickly."

By now, the robot had squirted out a full cup of alcohol. And for a moment, a small halo of flame shot out of the side of the robot, illuminating the cup.

Someone asked, "What does it taste like?"

"Probably fuel," Davalos joked.

She tinkered with it a little more and then tried again. And this time, it worked: Huge jets of flame shot out the side of the little robot, blasting the drink.

We all cheered. And then someone noticed that the floor had caught fire a little bit.

"Yeah, that happens," Davalos said matter of factly, before stamping out the little flames with her foot.

Over on the other side of the room, Magnus Wurzer, a founder of the Austrian Roboexotica, was setting up Chris Veigl's Mind Reading Martini Maker.

This looked a little bit more like a science experiment than a robot, but Wurzer explained that the key to this robot is software that is "customized for martini mixage."

He explained that the robot measures how many alpha waves on average are being detected by an electroencephalogram (EEG), and that the more alpha waves it finds, the drier the martini it makes.

Back in Vienna, Wurzer said, Roboexotica has grown to include dozens of cocktail robots, as well as others that people bring just because they're cool.

For example, he said, someone had recently brought a robot that was capable of flipping a cigarette into peoples' mouths.

Meanwhile, Calkins was trying to get Chapek up and running again, but he was frustrated by the effects of the robot's drinking problem.

"Orange juice and microcontrollers do not mix," Calkins said, exasperated.

A few minutes later, though, he offered, hopefully, "(the microcontroller) might be drying out."

To which Davalos deadpanned, "Every robot bartender needs to dry out once in a while."

On June 10, Geek Gestalt hits the highways for Road Trip 2008. I'll start in Orlando, Fla., and visit many of the South's most interesting destinations. Stay tuned, and be sure to keep up, both now and during the trip, with what I'm doing on Twitter.

Over the years, I've bought and built a lot of IKEA products: chests of drawers, office tables, bedside stands, media centers, glassware cases, and so forth.

Once, to make a little money, I even hired myself out to build some bookcases for a busy friend.

IKEA, as you probably know, is a furniture-retailing-industry phenomenon; millions of people buy its products because they're generally inexpensive and easy to put together. Plus, they almost universally come with everything you need to get going.

Almost every time I've put together an IKEA product, I've wondered as I sifted through the bolts, screws, and Allen wrenches: how do they design these things?

Well, one of the great things about being a journalist is that you get to ask such questions, and so I finally called IKEA and posed it. The answer, as you might figure, is rather complex, as I learned from IKEA product developer June Deboehmler and public relations rep Marty Marston.

"When we decide about a product, we always start with the price," Deboehmler said. "Then, what is the consumer need?"

For example, the product designers might begin thinking about designing a new flat-screen-television stand. Assuming that there's evidence such a product is needed--like a trend of many people buying flat-screen TVs--IKEA will set out to design it.

"When we start in the development process, we say we'd like to have a cabinet to hold a large screen TV that's 42 inches, and priced out to come in at X dollars," Marston said. "OK, now we've said we want it to retail at $500, arbitrarily. What can you make, what can you design, to make it at that price?"

From the beginning of the process, a variety of people get involved. Those include field technicians who are able to see what's needed in the creation of a new product and determine if IKEA has already designed something similar that can be mined for parts or design inspiration.

Another example is a packaging technician.

"They're always part of the team from way at the beginning, when the product is designed," Deboehmler said. "We always have to find the smartest way to do something so that it can be flat-packed and minimize waste of space when transporting."

Deboehmler and Marston used a recently designed product, the $139 Lillberg chair, to explain.

In the beginning, at the concept stage, the developer gives the lead designer what's called a "brief" on the new product.

"We give them all the parameters for everything the product should achieve," Deboehmler said, "the costs, the look, the style group, that kind of thing. Then we have a brief discussion, and then give them time to go away to create sketches...Then we sit down and do the real drawings we work from."

From there, Deboehmler, a lead designer, a packaging technician, and a field technician traveled to an IKEA factory in Lithuania and began work on the product on the factory floor itself.

With the Lillberg chair, the idea was to build a prototype at the factory--which the team did--and then to see what they had on their hands.

"After many, many days of trials, we thought we had it right," Deboehmler said. "'OK, this is the product.' Our designer was on his hands and knees. Then we got it back to (IKEA headquarters in) Sweden and started taking it apart again, and decided we can make it better because we can fit more in the package if we changed the arm direction."

By making a small tweak in the angle of the chair's arm, she elaborated, the designers and packaging technician figured out they could get more of the chairs in a single shipping container, and that, in the end, meant a lower cost to the consumer.

"The arm (change) meant huge savings," she said.

That's the sort of tweak that evolves organically from the design process, and may be impossible to discover until the team is well past the conceptual stage.

"When you see something on paper, it looks great," Marston said. "But it's not until you touch it that you say, 'Aha, if you turn it this way, we could get 10 arms out of this length of wood instead of 7."

The Lillberg chair took the design team about 10 months from concept to completion, including manufacturing time and global shipments.

That's about how long it takes for most new IKEA designs, Deboehmler said. An exception is lighting as that requires going through lighting tests for each country a new light will be sold in.

Another major consideration in the design process is minimizing waste.

"The whole idea of waste is very much embedded in our culture," Marston said, "not only in product development, but in all the various functional (areas of IKEA). We are so against wastefulness. It's very much a Scandinavian thought behavior.

"When some of our teams go to factories, we always look at areas where we throw things away," she said. "Sometimes we say, 'Wait a minute, we can do something with this.' And we turn things upside down and inside out to see if we can do it better."

Karlstad, Malm, Noresund...
IKEA doesn't sell anyone else's products: almost everything is designed in-house. So how to explain names like Lillberg, Karlstad, Malm, Noresund, Ljusdal, and Tryggve?

Deboehmler said many, including the company's chair and sofa products, are named after Swedish towns. "So when you're driving around in Sweden," she laughed, "you suddenly see this town name that's a sofa."

And what of the range of hardware that's used to put IKEA products together? There are seemingly dozens of different screws, bolts, fasteners, studs, and so forth.

In fact, Marston said that IKEA tries to minimize the amount of hardware used in product designs. In part, that's because many products are made in multiple factories serving many countries.

"A number of years ago," Marston said, "somebody had the bright idea that if we narrowed down our catalog of hardware that we use in our products, then we can be even more efficient."

The design teams also look for ways to make the products stand out.

That's why when designing the Lillberg chair, the team chose to incorporate what is called a "dovetail joint," which involves two pieces of wood that interlock using fingers of wood pushed together.

"It's quite a difficult thing to do on a production line," Deboehmler said. "We didn't know if we could pull it off, but we managed."

And the advantage of doing so?

"It's a design feature on very high-end furniture," Marston said, "and someone who has knowledge of high-end furniture would recognize that as an attribute."

The company is also looking for ways to maximize warehouse efficiency.

"We have (only) two pallet sizes," Marston said, referring to the wooden platforms on which goods are placed. "Our warehouses are dimensioned and designed to hold these two pallet sizes. It's all about efficiencies because that helps keep the price of innovation down."

At some of its warehouses in Europe, IKEA uses robots to move products around, even doing things like taking them off of very high shelves.

(Credit: IKEA)

In Europe, some IKEA warehouses utilize robots to "pick the goods," a term of art for grabbing products off very high shelves.

These factories, Marston said, are dark, since no lighting is needed for the robots, and run 24 hours a day, picking and moving goods around.

"You (can) stand on a catwalk," she said, "and you look out at this huge warehouse with 12 pallets (stacked on top of each other) and this robot's running back and forth running on electronic eyebeams."

At any given time, Deboehmler said, IKEA will likely be in the process of creating 5 to 10 new products, some of which are for the current year, and some for next or the one after that.

"It's an ongoing process," she said. "There's no real beginning and end to the year cycle. It's continual."

Over at Boing Boing this morning, I see that uber-blogger/novelist/speaker/electronic freedom fighter Cory Doctorow is planning on speaking at the Roboexotica symposium that gets under way in Vienna, Austria, tomorrow.

I hadn't heard of Roboexotica myself until I was in Austin, Texas, last month covering the Maker Faire there. At dinner one night with some of the Maker Faire folks, I Make Things video blogger Bre Petis started telling me about the event. And as often happens when smart people tell me about amazing things, my inner geek got very excited.

If you're not familiar with Roboexotica, this is how it's explained on the official Web site: "Until recently, no attempts had been made to publicly discuss the role of cocktail robotics as an index for the integration of technological innovations into the human Lebenswelt, or to document the increasing occurrence of radical hedonism in man-machine communication. Roboexotica is an attempt to fill this vacuum. It is the first and, inevitably, the leading festival concerned with cocktail robotics worldwide. A micro mechanical change of paradigm in the age of borderless capital. Alan Turing would doubtless test this out."

Now, I don't know what "Lebenswelt" means but I get the gist of it. In fact, if it weren't for the fact that people I know to be serious about things like this were actually traveling to Vienna, I might have thought Roboexotica was a prank. After all, "cocktail robotics?"

But it is real, and I wish I were going.

It turns out that the topics being discussed at the symposium don't all have to do with programming robots to serve gin and tonics--though, since I don't speak German, I'm not entirely sure what much of the program is about.

What I can see on the English version of the festival's site, however, looks pretty interesting. You've got Doctorow speaking about "why consciousness uploading, post-human existence and life after the Singularity are popular today, and why science fiction is always about the present," and Petis is doing his own talk on "the apocalyptic utopia."

Fun stuff.

Now all we need to do is figure out how to get the organizers to do a San Francisco edition of their event sometime in the future, and I can guarantee a rabid local response. Zombies, meet cocktail-serving robots.

SAN JOSE, Calif.--I'm down in the Valley on Thursday for the RoboDevelopment conference, and, having been to a few other robotics events, something seems very different.

In the past, I've been to RoboNexus, a show aimed at presenting kids and adults alike with the latest developments in robotics technology. And I've been to RoboGames, a display of pure robot-on-robot battle fury.

And there's no doubt that Robo Development is different. For one, it's smaller. For two, there's no pervasive grinding sound from the gnashing teeth of gears and claws and, well, saw blades that you might find at RoboGames. And there are no kids, like you find en masse at RoboNexus.

So, what is RoboDevelopment?

"There isn't an event designed for robot technology professionals," said Dana Kara, the president of Robotics Trends, the media company putting on this confab. "For people building robots for sale, there's nothing out there."

Until now, that is.

RoboNexus and Robo Business, both shows run by Kara's company, cover a broad range of robotic technology. But he said that after surveying the robotics conference landscape, he concluded there was nothing geared toward giving the people building robotics infrastructure technology a place to gather and showcase their work.

So, that's what RoboDevelopment is. (For more pictures from the show, see "Photos: Robots for all occasions.")

"Robotics is much more difficult than IT," said Kara. "It requires software engineering, mechanical engineering and electrical engineering, so it's much more difficult to build things for the real world."

So, for example, the keynote speeches here Thursday have such titles as: "Low Cost, High Impact Enabling Technologies for Mass Market Robotics"; "The Emerging Robotics Industry--Catalysts and Next Steps"; and "A Robotics Development Platform from Kindergarten to Rocket Science."

This is heady stuff, and the conference itself definitely feels that way.

After all, as Kara put it, this is an event where the providers of "enabling technologies like development environments, sensors, meters and actuators and power" come together to help the creators of robots aimed at the consumer market get on their way.

One other thing that Kara pointed out struck me, and that was the reasoning behind holding this event here.

In the past, he said, most serious robotics conferences have been held in Boston or Pittsburgh because of those cities' proximity to specific universities or military research in the field.

But now, by bringing RoboDevelopment to San Jose, Kara's company is hoping to bring the field of robotics development for commercial products to a place rife with venture capitalists looking for new entrepreneurial fields to invest in.

"This could be a huge market if they get their act together," said Kara.

Still, I didn't entirely understand what the show was about. But Kara had the answer.

"Any graduate student can build a robot that can vacuum your floor," he said. "But how do you do it for $200?...That's where these people are."