Automaton

Wing-flapping micro robots, unmanned helicopters, formation flight algorithms -- there were lots of cool UAV projects at the International Conference on Intelligent Robots and Systems (IROS) last week in San Diego, Calif. Too many, in fact, to describe them all here. But Automaton correspondent Marcel Bergerman, a systems scientist at Carnegie Mellon, has the highlights:

Greetings from San Diego! I am at IROS 2007 looking at the latest developments in unmanned aerial vehicles (UAVs). IROS is a huge conference, with 132 technical sessions in 11 parallel tracks over three days, plus three plenary sessions and several workshops.

UAVs alone got four sessions, which is an indication of the importance of the field among the myriad of disciplines that comprise robotics today. The weather was great, and luckily the forest fires that abated San Diego did not interfere with the conference. More than 1,000 people registered for and attended IROS, making is one of the premier robotics conferences worldwide.

Instead of covering each UAV paper presented in a succession, which would be just plain boring, I decided to report on those works that I found interesting or novel (in no order of importance). The pictures below were taken from the conference CD-ROM; of course, these carry the IEEE copyright, but since this is an IEEE blog, I figured it would be OK to publish a picture or two (or three).

Under the category "design of micro UAVs" I enjoyed the works of Robert Wood (Harvard) and Xinyan Deng (University of Delaware). Wood is designing insect-size UAVs that actually flap their tiny little wings to generate enough lift to keep the vehicle aloft, with some lift to spare for an onboard processor and sensors.

Transmission system for one of Robert Wood’s micro aerial vehicles.

He described the fabrication process for each of the components, which is rather involved. Videos shot with high-speed cameras show the wings flapping at 110 Hz very robustly. I look forward to some future conference where Wood shows these tiny UAVs flying with some degree of autonomy.

Robert Wood’s 60 mg, 3 cm wingspan micro aerial vehicle.

Deng is also working on insect-like UAVs; although she showed results with "large" wings on the order of several centimeters, she mentioned that her tests are coherent with real insects in terms of Reynolds number and advance ratio (the ratio of flapping frequency to flight speed).

Xinyan Deng’s prototype dragonfly robot.

Talking about bigger machines -- Our group at Carnegie Mellon (yes, I did find our work interesting) :-) presented a novel model-based cascaded controller for the Yamaha RMAX unmanned helicopter. The model is adapted from work published in 2006 by NASA's Mark Tischler in the Journal of the American Helicopter Society; I believe it will slowly replace Mettler's as the basis for small helicopter simulation and controller design models.

A quadrotor built at ETH Zurich also caught my attention, especially the video where the machine avoids people walking towards it.

ETH’s quadrotor in hover.

On a higher level of abstraction I would like to mention Rachid Alami's (from the LAAS/CNRS, France) formation flight algorithm, which allows UAVs to adopt the best possible convoy formation for a given mission profile. The algorithm takes into account threats such as enemy-operated jamming radars which would disrupt communication between the UAVs and between them and a ground control station. Alami’s group has started experimenting with three small fixed-wing airplanes.

Alami’s formation flight algorithm applied to an 8-UAV fleet.

Overall there were 20 UAV-related papers from four continents and eight countries, including Portugal, Germany, France, Switzerland, Japan, Australia, Mexico, and Brazil. This shows how international is the UAV community!

Thanks, Marcel!

The previous DARPA Grand Challenge competition -- a trip through the Nevada desert taken by autonomous vehicles-- took two tries to get right; the first year, not a single vehicle made it across the finish line. The second year was a much better showing -- four vehicles finished -- and winner Stanford University took away the $2 million prize.

This year's DARPA Urban Challenge took the robots out of the desert and into a (simulated) city. Teams had to build vehicles capable of "executing simulated military supply missions while merging into moving traffic, navigating traffic circles, negotiating busy intersections, and avoiding obstacles." Since this was the first year of this style of competition, many people wondered if it would have the same problems as the first year in the desert -- lots of failures and no one completing the course.

We needn't have worried. Of the 11 vehicles that were allowed to enter the final round of the competition, six finished the course -- though only three teams, Carnegie Mellon, Stanford, and Virginia Tech, finished under the 6 hour time limit.

The MIT vehicle waits at an intersection as a (human-driven) car makes a turn.
Photo Credit: JOHN VOELCKER

So what drives these vehicles (since it's not humans)? The short answer: lots of sensors and lots of computing power. Nearly all the vehicles had some sort of array of laser range scanners arranged on the front -- though while MIT used more than 10, the UPenn entry got away with just 2. A key player in that technology was Velodyne, developers of a high-def LIDAR unit based on their work in the first two DARPA Challenges -- they stayed out of this year's event in order to continue developing their LIDAR technology. Additionally, LIDAR units designed by IBEO and SICK (an old favorite of DARPA teams) were other popular additions to the sensor suite. Stereo vision complimented the laser sensors, and of course, differential GPS receivers and inertial measurement units (IMUs) were must-haves.

While hardware integration is no easy task, software was just as daunting. A layer of hardware interface ("What does the LIDAR say?") under a layer of navigation and control ("Where am I, where do I have to go, how far do I turn the steering wheel, and how fast do I have to go?) under a layer of behavior ("Hm, a stopped car. Wait behind it, or drive around it?") makes for some intense coding. Take the Carnegie Mellon vehicle, which required over 300,000 lines of code to run the 2007 vehicle. Some COTS tools made this easier for teams such as Virginia Tech, who used LabView to "provide the major functions of the vehicle including image acquisition and processing, systems communication, vehicle health monitoring, and vehicle control. A NI Compact RIO system [provided] steering, throttle, and braking control, as well as reading CAN-bus sensors," said NI representative Trisha McDonell.

With the impressive success of the vehicles on Saturday, is my human-driven car suddenly old fashioned? Not so, say the experts. Forbes had a nice article on the competition, and specifically quoted Stanford team leader Sebastian Thrun:

In the eyes of Stanford's team leader, Sebastian Thrun ... the world is still years away from driverless autos. "I'm positively enthused that this race has a winner," he said. "But we’re witnessing the painful birth of a new technology, and this is the first of many hours of labor."

Fair enough, Dr. Thrun. I'll settle for a car that can park itself for the time being.

Special thanks to John Voelcker for insight and photos from the field

If you've noticed the radio silence lately, my excuse is that I've been traveling a lot for work. One of the places I ended up was London, #1 on my very short "favorite international cities" list. This is not to say, however, that I was completely off Automaton duty... I took the opportunity to snap a few photos of some robot goodness.

I was amused by a sign at the ubiquitous Pret a Manger, promising its customers that no robots were involved in the making of your sandwich:

And there appears to be a "robot world" of some sort at County Hall on the south bank of the Thames; my eye was originally drawn there because of the apparently massive Star Wars exhibit in the same building, but my work schedule didn't give me enough time to check out either Star Wars or Robot World. The internet doesn't have a ton of information on Robot World, either -- I'd be interested in comments from those who have been.

The one bit of info I did find online made it sound a little like the MIT museum in Cambridge, Mass. Can anyone confirm?

An article in PC magazine last week called "Robot Consumers, Grow Up!" explains why US consumers just aren't ready for the kinds of robots that Asia will be producing. The conclusion is that the consumer robotics market will not take off with things like Roomba or Aibo. Instead, robotic technology will be embedded into everything we use every day and the consumers will never notice that "the robots have won." I think the article is overall a little pessimistic, but here are a few interesting comparisons:

Part of the problem is the Western world's relatively short history with robots. Most people point to Karel Capek's R.U.R. (Rossum's Universal Robots), a science-fiction play that premiered in 1921, as the first use of the term and America's introduction to robots. We should take a cue from the Japanese. In the book Loving the Machine, author Timothy N. Hornyak explains that robots (or at least automatons) have been part of the Japanese culture for hundreds of years. They're seen as friends, helpers, entertainers, and companions. They've always resembled their creators.

American consumers fixate on anthropomorphism and generally find androids and even android pets grotesque. You won't find a lifelike robot receptionist in the U.S., but there are already many at work in Japan.

If iRobot had made a 4-foot-tall Roomba with a face and a hand to hold a vacuum hose, the company wouldn't have sold more than ten units. Instead, it sold more than two million Frisbee-shaped, personality-free bots.

There's also an interesting statement later in the article that "American robot consumers have yet to comprehend the cost of the programming and mechanical complexity necessary to create effective, realistic, interactive robots." So that's something to think about, especially when, as a robotics geek, I'm always so involved with how cool the thing I'm making is that I don't stop to think whether others appreciate how it came to be. And it also leads me into another article...

Network World talks today about a robot "amusement park" opening in South Korea. From the article:

"[O]fficials said they consider robotics to be one of South Korea's key growth industries, emphasizing "service robots" that can clean homes and offer up entertainment. The robotics industry has grown about 40% a year since 2003, officials said. ... The parks will feature a number of attractions that let visitors interact with robots and test new products."

So the first question is whether or not this would be successful in South Korea -- obviously, the government and a group of investors think it will be no problem. But if we step back and consider the first article about American robot consumers, would something like this ever fly in the US (or in the Western world, for that matter)?

From Automaton correspondent Sally Adee:

Sandia researchers Chris Forsythe and a companion, fitted with EEG caps, test drive a "smart vehicle" that can read their brain activity. Photo: Sandia

If driving next to a semi truck makes you nervous, you're not alone.

Reports are all over the news about how many incentives truck drivers get to push through sleep-deprivation and other impediments. They drive long, monotonous routes with little stimulation, and the results are sometimes fatal.

And it's not always because the driver is distracted—researchers at Sandia National Laboratories, in Albuquerque, N.M., are looking into what happens to your driving when you're not distracted enough.

Sandia is designing "smart cars" that will alert you when you're approaching that dangerous "twilight zone" between consciousness and sleepiness. They call it the "underload" condition—you're bored, distracted, drowsy, or daydreaming because there's nothing to engage your mind.

The military application to this is convoy driving, but obviously the civilian counterpart is long-haul trucking. The original project was funded by DARPA about five years ago, but now it's been taken over by the Marines. The military often doesn't use brake lights, principal investigator Kevin Dixon told me.

"So if you're a 19-year-old lance corporal driving a 25-ton truck, you're not paying attention, and you don't notice that the car in front of you has slowed down, you're going to do a lot of damage with that vehicle."

But how to detect crushing boredom?

For the college-aged (and graduate-student-aged) among you, I wanted to point out that the NASA Robotics Academy is accepting applications for its 2008 summer program. Students are hired to work on projects in teams of 4-5 at either Goddard or Marshall Space Flight Center. The internship is residential (you'll live with other interns) and it does include a stipend.

I'm an alum of the 2005 program (its inaugural year) when I worked on a planetary rover design project with Goddard in conjunction with a lab at the University of Maryland; it was a good time and the trips to see other robotics labs -- we went to Johnson Space Center, MIT's CSAIL, and Carnegie Mellon's Robotics Institute -- was unbeatable.

Also, you get a discount at the NASA gift shops. Just saying.

Application deadline is 15 January, and you'll need some recommendations, so get crackin'!

Photo: U-Boat Worx

Truly a James Bondian contraption. The C-Quester 1 is a personal submersible. And like in a real sub, you don't need a diving suit to ride it. You stay dry and pilot the thing from inside an acrylic dome.

The sub's three electric motors can propel it to 2.5 to 3 nautical miles per hour, and you can descend as far as 50 meters. It has a computer that monitors carbon dioxide concentration, air pressure, and temperature. It's got air conditioner, too.

The C-Quester 1, made by U-Boat Worx of Breda, Netherlands, was part of Spectrum's annual roundup of gadgets and gizmos for the holidays.

OK, not a robot but pretty cool, uh? Not so cool is the price tag. The C-Quester 1 costs about US $130 000, plus taxes, import duties, and registration fees.

Happy belated Thanksgiving from the Automaton crew -- we hope you had a great holiday!

Thinking of holidays, I'm commissioning an artistically-inclined friend of mine to make me custom Christmas cards this year, and between the two of us we've been investigating ways to incorporate robots. There's a surprising lack of robot cards out there, so it's becoming a DIY venture. For the arts and crafts types among you -- or those hoping to indoctrinate your children with the awesomeness of robots -- you may be interested in these rubber stamps we found:

A company named "Bam Pop!" makes a number of what I can only describe as "cutesy" rubber stamps and other crafty paraphernalia, and they have a few robots on this page. In particular I thought that the "hellobot", shown to the left, could be very versatile for greeting cards.

(Image from bampop.com)

Slightly less cutesy, but still very friendly, is this one from Etsy. I think it wants a hug. I also think the technical considerations of it balancing on a single wheel warrant a discussion.

(Image from etsy.com)

And finally, the Robot Kid from A Muse could not only make a nice greeting card, it could also inspire your child's next Halloween costume.

(Image from amuseartstamps.com)

Why do they all have square heads? I can't think of any robot we've looked at on this blog to date that has a square head.

Anyway! Found any other good stuff for robot arts and crafts? Let us know!

Raytheon is a large, well-known defense contractor with divisions all over the world. Sarcos Robotics is a small company in Utah that makes, among other things, animatronic and robotic characters for movies and amusement parks. And now Raytheon has bought Sarcos.

This article in Mass High Tech describes how Sarcos will be folded into the Integrated Defense Systems division in Massachusetts. In addition to its business of animatronics, Sarcos works on some medical devices and MEMS. Undoubtedly the MEMS technology will come in handy as part of Raytheon's defense systems, but perhaps there is another reason: Sarcos makes a robotic butterfly (link goes to Quicktime video). Perhaps Americans should be suspicious of military butterfly spies?

At any rate, take a look at Sarcos's website (especially their mildly intimidating jobs page). They do an excellent job of posting videos of their animals, humanoids, and other products. I like the juggling robot, below. I can't even do that.

In response to yesterday's article about Raytheon's acquisition of Sarcos, a friend sent me this video:

That, my friends, is pretty crazy.

Spectrum ran a feature on exoskeletons two years ago with some interesting details on the Sarcos system's force sensors, power unit, and hydraulic actuators, below:

For its part, the Salt Lake City–based Sarcos team, led by roboticist and inventor Stephen C. Jacobsen, has been working on what may be one of the strongest exoskeletons ever built. Earlier this year, at the demonstration the group did in Fort Belvoir, an engineer wearing the Sarcos robotic system was able to carry 84 kg [185 lb]—about the weight of an average size washing machine—without feeling the payload at all. Jacobsen, Sarcos's CEO and a mechanical engineering professor at the University of Utah, says that the new exoskeleton supports the payload's entire weight even if the wearer stands on one leg.

Like Bleex 2 [the UC Berkeley exo], the latest Sarcos system is a second-generation model that improves substantially over its predecessor. Jacobsen says that while wearing the exoskeleton, you can walk and run, and if you stumble, the system is fast enough to readjust its powered limbs to keep the payload's weight off your body.

The exoskeleton relies on a network of force sensors that are in touch with the wearer's body at certain points, such as underneath the feet. These special sensors, developed by Sarcos, feed data to a control computer that in turn commands the robotic limbs to move in harmony with the wearer's arms and legs without ever obstructing them. Jacobsen calls this method "get out of the way" control, and he says using the robotic suit requires no training. "You can step into the exoskeleton, and you can immediately run it," he says.

According to Jacobsen, what makes an exoskeleton an extremely hard problem is that conventional, off-the-shelf components won't work. Sarcos had to design and fabricate each piece and, in parallel, integrate all of them into its system. The exoskeleton's power unit was one of these many pieces the company had to engineer painstakingly. It's a special internal-combustion engine that can use a variety of fuels and deliver enough hydraulic power to the actuators to meet the great strength and speed the robotic limbs require.

But even more challenging, Jacobsen says, was developing yet another component: the servo valves that control the flow of the hydraulic fluid into the actuators. The valves had to be small, extremely reliable, resistant to high pressures, and highly efficient to preserve precious power, not to mention that some of their parts had to be machined to micrometer tolerances. To make things even harder, so many complex physical processes occur in the valves, Jacobsen insists that simulation software couldn't help in the design. His group, therefore, had to go through several iterations of prototypes to get the valve it needed.

Sarcos is now preparing for demonstrations scheduled over the next few months. Team members are especially busy with the exoskeleton's upper-extremity system, which will add strength to the wearer's arms. A person wearing the full-body system will be able not only to carry a payload on a backpack but also lift heavy items, a capability that is particularly useful for logistics operations such as loading and unloading cargo vehicles and moving things in a warehouse.

PS: When my colleague Harry Goldstein and I spoke with Sarcos for that article, the company had just began developing their exoskeleton's upper-extremity part. Now, as the video shows, it seems they've made significant progress. One thing, however, hasn't changed. Note in the video: the exo has a tether attached to it, probably feeding power or control signals to the suit. Sure, it's a prototype. It will be interesting to see how exoskeleton researchers will cut the umbilical cords of their creations.

As I've mentioned before, I work for a company called Bluefin Robotics located in Cambridge, Massachusetts. I haven't really talked about our technology, but we just recently got a contract for the next generation of one of our coolest vehicles, and I really like talking about this one, so on to... the HAUV!

The Hovering Autonomous Underwater Vehicle started out as a joint project between Bluefin and MIT. It's a significant departure from our other vehicles, which are torpedo-shaped with propellers on the back. HAUV is more or less a box, not needing to be quite as hydrodynamic as its siblings, and as such has occasionally earned nicknames such as "Spongebob" in the Bluefin lab.

The HAUV is run by a main electronics housing (the brains) and a 1.5 kWh subsea battery that we make. It moves thanks to eight fancy hubless thrusters arranged to allow a full six degrees of freedom: X, Y, Z, roll, pitch, and yaw. To navigate it uses a Doppler Velocity Log (DVL), which provides the computer with velocity along the hull; an inertial measurement unit to measure orientation in space (or water!); a compass; and a GPS antenna to achieve a position lock when it's on the surface. Its payload is a Dual-frequency Identification Sonar (DIDSON), located on the front of the vehicle next to the DVL, which provides imagery of the ship hull that looks a lot like a blue ultrasound. The Soundmetrics website (linked above) shows what some of that imagery can look like.

HAUV communicates with an operator via a fiberoptic cable that runs between the vehicle and the ship, but the cable is there for transmission of sonar data, not for active control from the operator (though the operator can upload new sortie commands via the link). This is different from our other vehicles, which are not tethered in any way and instead communicate via acoustic link underwater or via Iridium satellite or an RF link on the surface (depending on distance from the ship).

So what is this all for? Hull inspection, basically. Plop this little guy in the water next to a ship and it can go to town taking images of the hull. If it sees something suspicious, the Navy or Coast Guard can send down a diver to check it out and dispose of it as appropriate. Port security is a big deal these days, so there could be a lot of work for the HAUV.