IROS (the International Conference on Intelligent Robots and Systems) starts today in California, and we're here all week to bring you the best it has to offer. If you're not familiar with IROS, it's kinda like the fall version of ICRA. You remember ICRA, right? San Francisco may be (slightly) less exotic than Shanghai, but that doesn't mean that the robotics presentations are going to be any less awesome. 2,459 papers were submitted to the conference this year, and there will be ten (ten) conference tracks going on all at the same time, with rapid-fire 15 minute live presentations, and we're going to do our level best to be in all ten of those places at once. And on top of that, there's workshops, demos, and an exhibit hall.
This morning we're kicking things off with a three-hour tour of the UC Berkeley robotics lab, followed by an afternoon packed full of presentations, including animal robots, robot table tennis, brain interfaces, insect robots, and something about a shape-memory alloy robotic heart. All of this, just in half of day one!
Please bear with us if our posting schedule gets a little wonky this week; we're gonna be running ourselves ragged trying to keep up with all the cool stuff that's going on. Stay tuned!
It's been a long time coming, but Intuitive Automata's Autom robotic weight loss coach is now up for pre-order on a dedicated "MyAutom" website. If you haven't been following the saga of Autom, it was first an MIT Media Lab robot with a significantly different look. Autom's developer at MIT, Cory Kidd, co-founded Intuitive Automata to help commercialize Autom based on the original MIT project, and it's starting to look like everything will be coming together within the next year. Not to get off topic or anything, but it's fantastic to see a research robot like this make the difficult jump into the consumer market. Congrats to Dr. Kidd!
Anyhow, back to the robot. We know that Autom is designed to be exceptionally interactive, crunching data on your health, diet, and exercise regimen and giving back friendly and constructive criticism. Studies have shown that people who use Autom stick with their diet and exercise routines for twice as long as people using more traditional weight loss methods. Don't ask me how, maybe it's something about those big blue eyes?
If this sounds good to you, you can be one of the very first people to have this friendly little robot helping you out every day with a deposit of $195. This is not the final price, however, it's just the pre-order deposit. The final price is the $195 deposit plus a balance of $670 when the robot ships, for a total of $865. This does seem a bit steep, although I'll admit to not being familiar with how much a typical weight loss program costs.
On the upside, Intuitive Automota seems to understand that cost is, uh, an issue, and they're planning on working with health insurance companies and employers to try and subsidize things a bit. Anyway, pre-orders are open now, and you can find out a bit more info (but not all the info you'd probably want before spending most of a grand on a weight loss robot) at the website below.
Dr. James Law, a researcher at the Department of Computer Science at Aberystwyth University, has had an absolutely fantastic idea: he's nominated the iCub robot to carry the Olympic Torch as part of the 2012 Olympic Games, which will be held in London (that's in England, folks) starting next summer.
Dr. Law is proposing that iCub be included in the torch-carrying relay in honor of the 100th anniversary of the birth of Alan Turing, one of the guys who arguably invented the computer and whose test for artificial intelligence robots are continually striving to pass. This is a great idea, but I think that iCub should be part of the torch relay on its own merits: it'll be a first for robots and great publicity for engineering education and all that. Or at least, it'll be great as long as iCub doesn't faceplant in a puddle and snuff the torch out.
The only problem with this idea is that the short-sighted and obviously outdated nomination rules specify that all nominees have to be at least 12 years of age, which would mean that iCub wouldn't technically qualify. On the upside, nowhere does it say that nominess have to be human, so maybe iCub has a shot at this after all.
There are a lot of contenders in the race to become the dominant software platform for robots. One of the factors that will determine the winner is, of course, achieving critical mass. By making your software open source, you can reduces barrier to acceptance to a minimum, helping it to spread faster. This is what is happening with ROS, the Robot Operating System created by Silicon Valley robotics firm Willow Garage.
Research labs across the world are adopting ROS at fast rate. And because these labs are packed with some of the freshest and most clever minds in robotics, the ROS community is constantly adding more capabilities to the software. ROS now offers many advanced packages for robot vision, navigation, and mobile manipulation, among others.
Some robotics companies understood the potential of riding this wave of innovation, and now several commercial products use ROS as middleware. These include Aldebaran’s NAO humanoid and Meka Robotics' systems. Starting with basic ROS functions, these companies have built their own custom systems specific to their products.
As I considered the evolution and adoption of ROS, I thought that it would be great to have not only start-ups and research labs but also industrial robot companies embracing ROS. (That would be good for ROS too, of course, because the industrial segment is still a big chunk of the robotics market.) But could that ever happen?
In industrial robotics, I feel that lots of people are constantly reinventing the wheel, with different robot manufacturers developing their own proprietary operating system and controller. The result: You can't program a robot from one manufacturer and reuse that same program on a robot from another manufacturer. Furthermore, if you want to develop an add-on component or peripheral for a robot, you need to establish a relationship with the robot's maker; in other words, you need "permission" to get access to the "black box" that is their robot controller.
Maybe things work this way because industrial robotics have traditionally been a slow-moving, conservative, and expensive business. And it's also been tied to manufacturing, where everybody is trying to keep their edge on new ways to make better products at better prices using advanced technologies. As I pointed out in an article about the stagnation of industrial robotics, the proprietary operating system is an important piece of the business model that industrial robot makers rely on. For this reason, I thought that having a major industrial robot manufacturer adopt ROS would never happen.
Well, I was wrong. It was a nice surprise to hear that the Southwest Research Institute (SwRI), a private R&D organization based in San Antonio, Texas, announced recently that they've reached an agreement with Motoman, one of the largest industrial robot makers in the world, to develop a ROS interface for the Motoman SIA20 7-axis robot [CAD image and photo above]. By making its robots compatible with ROS, it seems that Motoman is betting that all those tools developed by the ROS community will become very enticing for its customers. And if customers demand that those capabilities be available, relying on a proprietary system doesn't make help you.
This could be the beginning of something big. If more industrial robotics companies adopt ROS, this could enable a lot of technology transfer from the research world to real-world applications. And then, after we get every robot on ROS, imagine we could connect them through the Net so they could share a common knowledge base.
Now you’re ready for a real robot revolution.
Samuel Bouchard is a co-founder of Robotiq, in Quebec City.
There was something particularly fetching about the design of the original Mint robotic sweeper. It was clean. It was simple. It was white. All that has just been thrown out the window with the new Mint Plus, which has most decidedly gone over to the dark side.
Besides being blacker than Darth Vader's coal cellar (he's got one of those up on the Death Star, right?), the Mint Plus 5200 series has as a bunch of new features that mostly justify its $100 price bump. First off, the place where the microfiber cleaning pad mounts to the robot (check it out in our review if you're not familiar) now contains a liquid reservoir that keeps the pad moist while the robot cleans up to 350 square feet.
The other big change is that Mint's NorthStar cubes have been upgraded to NorthStar2, which endows each cube with some sort of unique identifier that Mint can detect to allow it to move from room to room. With a dry cloth, this gets you up to 2,000 square feet of cleaning. To make it that far, Mint's battery has been increased by 25%, and there's an optional new TURBO CHARGE CRADLE which allows Mint to be charged in two hours instead of four but sadly does not increase Mint's speed to turbo.
And finally, Mint Plus is smart enough to resume cleaning after you pause it to change its cleaning cloth, retaining its room map and moving on to all the places it hasn't hit yet after you put it back down and tell it to resume.
Besides all this additional mintiness which is now virtually certain to freshen your breath as well as your floors, the Mint Pro looks to have all the upsides of the original Mint (most notably simple, effective, silent hard-floor cleaning), along with the one obvious major downside: no carpets. Oh well, all you lucky people with your hardwood floors can just fork over the $299 for the Mint Pro and go about your happy, carpet-free lives.
Some of my roboticist colleagues at ETH Zurich are, somewhat surprisingly, architects. Fabio Gramazio and Matthias Kohler, both professors at ETH's Institute for Technology in Architecture, were among the first to use robots in architectural design. Since 2006 the duo has explored various manufacturing techniques, including both subtractive and additive fabrication, as well as a wide range of materials, to create astonishing structures entirely built by robots.
The use of robots, combined with digital design tools, means a new aesthetic becomes possible, with novel shapes and patterns that would be nearly impossible to achieve without the automated machines: industrial manipulators that are extremely precise and good at repetition.
Using robots, the two ETH architects, who run the Gramazio & Kohler design studio, have fabricated intricatebuildingparts out of wood, concrete, bricks, and foam, and have used these parts to build complex, beautiful installations in Zurich, London, Barcelona, New York, and other locations.
The idea of using robotic systems as reconfigurablespaces or "smart furniture" is not new. But the way Gramazio and Kohler are using robots -- to actually build large environments -- is very innovative indeed. Though their creations thus far are limited in size, the architects are currently exploring the idea of applying robotic fabrication to the design and construction of high-rise buildings.
As you can see in the photos below, the results are impressive. In one of their projects, the architects fitted a manipulator robot in a modified freight container -- a "mobile fabrication unit" that could travel anywhere in the world. They took it to Manhattan a few years ago, where the robot built a 22-meter-long (72 feet) brick structure [photos below].
In another project, they used the robot as a milling machine, to create parts that could shape the acoustics of a room [photo below].
Some of their most interesting creations, though, are the ones that use robots to assemble elaborate environments.
Here's how they describe a 2009 project to build a temporary spatial structure [photo, right] for a major public event in Wettswil am Albis, Switzerland:
The wooden structure consists of 16 contorted elements made from 372 slats. The entire construction is structural support, roof and skin of the building at the same time. The elements were constructed by a digitally controlled robot that cut and precisely placed the slats according to an algorithmic pattern. Each of the elements is individually rotated, producing a progression of subtly varied spaces. The logic of the openings and curvatures as well as the aesthetic details conform to the rules of wood construction. The digital processing bestows a new expression on the traditional wood material.
The architects are also collaborating with roboticists from the ECHORD project to give their robot more mobility. One idea is to use a base with tracks [photo below], and program the robot to recognize its position and surroundings. The biggest challenge is making sure the robot can handle construction tolerances and variations, adapting to changing conditions autonomously.
To learn more, take a look at Gramazio & Kohler's ETH website and their company website. Below, more photos of their creations and a video discussing their work.
It's been a while since we've gotten an update from Boston Dynamics about their BigDog quadruped. And this isn't really an update, I guess, as much as a video of BigDog's noble robotic lineage, with a whole bunch of, shall we say, "outtakes" thrown in for good measure:
We know that the University of Michigan's MABEL biped robot is fast and all, but it's also had some issues in the past with taking the occasional bad step with painful results. It now looks like MABEL has learned some fancy new footwork, with this demonstration of her ability to not completely faceplant when confronted with a surprise 20cm step:
And lastly, I hope you're not burned out on quadrotors yet, because this is pretty sweet. Daniel Mellinger, Alex Kushleyev, and Vijay Kumar at UPenn's GRASP Lab have taught a big quadrotor to act as a landing (and launching) platform for a little quadrotor. Oh, and there's a bunch of hula-hoop dodging with multiple quadrotors at the end, too:
Microsoft, to their credit, has done a good job of embracing Kinect as a game-changing robotics tool instead of just a... A... A video game controller, was it? Well, whatever it was originally designed as, it's all about cheap and effective robotic 3D vision now. Microsoft knows that Kinect is a big deal for robotics enthusiasts of all kinds, and they've just announced the availability of a new beta release of Microsoft Robotics Developer Studio that incorporates the full Kinect SDK that was released back in June. This includes skeleton tracking, speech, and the raw Kinect data stream for creating 3D maps of your house (or anything else).
Besides the full-fledged Kinect integration, the other big news about RDS 4 is that for the first time, Microsoft has their own hardware reference platform designed to make it fast and easy (sort of) for consumers to get straight to programming without having to actually build themselves a robot. Eddie, pictured above, features a round multi-level design that incorporates a Kinect sensor and off-the-shelf laptop. ::cough::. ::cough again::. So yes, Eddie is clearly in the same class as both TurtleBot and Bilibot, which offer similar designs and capabilities and run ROS. We probably shouldn't create some kind of Mac vs. PC thing here, but strictly by the numbers, Eddie is a significantly more expensive proposition at $1200 assembled without a Kinect sensor or a laptop, while both TurtleBot and Bilibot cost the same amount including a Kinect sensor and a netbook.
Whether or not you decide to use Eddie and Microsoft RDS, it's always great to see companies like Microsoft embracing robotics by helping give more access to the developer community with free software releases and customized hardware platforms. As Microsoft puts it,
"This beta release is one of our early steps towards realizing our long term strategy of accelerating the consumer robotics industry. Our motivation in releasing these tools is to extend and democratize access to robotics development, bringing value to the space through ease-of-use, accessibility, and a robust existing developer community."
If you're interested in checking out the RDS 4 beta, you might also be interested in Microsoft's Robotics @ Home Contest, where you could win a free robot and possibly $10,000 for coming up with "a cool idea." Yep, that's it.
Robotics Developer Studio 4 Beta is available for download for free, and includes a simulation environment to get you started without needing to buy any hardware at all. When you're ready to take the plunge, Eddie is also available now, directly from Parallax.
Looking at the product designs from OLogic, a consumer robotics company launching at DemoFall 2011 earlier this week, the words to the classic Wizard of Oz song popped into my head “If I only had a brain…” OLogic robots don’t have brains, instead, they rely on a users’ smart phones to provide the brainpower.
At Demo this week OLogic introduced the A.M.P. (Automated Music Personality), essentially, a boom box on wheels that will sell for $300 to $400. The roving music player doesn’t do a whole lot, but you can dance to it—or with it. The company also displayed Oddwerx, a $50 smartphone dock that makes your mobile phone really mobile—it can wander around your desk. Both devices seem destined for the gadget catalogs that live in airline seat pockets, but they are good reminders that we’re carrying an awful lot of computing power around in our pockets these days. Larson introduces his gizmos in the video above.
I know we've been posting a bunch about quadrotors recently, but it's hard not to when they keep doing cool new stuff. This demo comes from Pat Bouffard and Anil Aswani and shows (eventually) a quadrotor catching tossed ping-pong balls starting at about 1:40:
All that other malarkey at the beginning of the vid (you didn't skip over it, did you?) talks about the programming that goes into making sure that this quadrotor, with what I think we can all agree is a fairly small container, can reliably make catches. Essentially, the robot pays special attention to what's physically going on with itself, using experience to compensate for thing like increased lift due to ground effect.
This technique is called LBMPC (that's Learning Based Model Predictive Control), and you can see it in action when the quadrotor needs to move sideways to catch the ball, as it figures in the fact that it's going to drift a little bit after it cancels out its lateral movement. Clever.
So, if Berkeley's quadrotor teams up with this robot, this robot, maybe this robot, and of course these robots, you've got yourself a halfway decent chance at giving any Little League team a run for their juice boxes, and I for one would pay money to see it happen.