What’s the best way to test out the robustness of your robot’s navigation system? Just set it loose to wander around the office for a week and see what happens. 70 kilometers later, this PR2 ended up with a funny hat, some stickers, and a new sense of confidence.

For 7 days, the PR2 was assigned to autonomously navigate the Willow Garage offices, locating outlets and recharging itself when necessary. If it got stuck, it was allowed to text a human for help, and that human could use a website to help the PR2 out. In 70 kilometers, this only happened twice (twice!), and I’m more than a little curious as to what those circumstances were. Apparently, this PR2 is still clocking in the kilometers, and the data it’s collecting on robust navigation will be made available to the ROS community.

In other PR2 news, Willow Garage has just sold their first four PR2 units to CNRS Laboratory of Analysis and Architecture of Systems (LAAS-CNRS) in Toulouse, France; George Washington University in Washington, DC; Samsung Electronics in Suwon, Korea; and University of Washington in Seattle, WA. This is good for them financially, of course, but the important thing is that institutions are taking PR2 and ROS (and what they stand for) seriously enough to shell out major cash to get a high quality piece of the action. It’s one thing to give away a bunch of robots and call them a success; it’s another thing when people believe in your product and what you’re doing enough to invest hundreds of thousands of dollars in your concept. Congrats to Willow Garage, and may they sell enough PR2s to drive the cost down to the point where I can afford one.

[ Willow Garage ]

This is just the first taste of what a hacked-up Kinect sensor is capable of… That motion capture and teleoperation system looks pretty sweet, and as Willow Garage says, they’ve basically just started messing with the capabilities of the sensor, and things are already progressing very quickly.

Kinect is $150, and the open source drivers are free. Go crazy.

[ Willow Garage ]

TUM’s PR2 (named James) and another resident robot named Rosie have teamed up to make some pancakes. This is pretty awesome. But all the stuff that’s going on before the actual pancake making even starts is just as awesome. To begin with, James as no idea how to make pancakes. So, he does what we would do, and looks up the recipe on the internet. From instructions he finds there, he’s able to generate his own pancake-making program, which includes using image recognition to determine what the right bottle of pancake batter looks like when he goes searching for it in the fridge. Both robots are also able to dynamically adapt to errors and obstacles and show up in their environment, common in a real kitchen.

As we’ve seen before, pancake flipping isn’t easy, but Rosie manages it with finesse. And I have to admit: when Rosie flipped that pancake, I clapped in real life. And I may have giggled.

So, how long until I’ve got a couple robots in my kitchen serving me pancakes whenever I want? Considering that this particular pancake likely involved about a million dollars worth of robotic hardware, it might be a while… But the autonomy and adaptability demonstrated here are laying the foundation for a James/Rosie lovechild in every home and pancakes for one and all.

VIA [ Willow Garage ]

You’ve got more nerves in your fingertips than in (almost) any other part of your body. This is because you get a lot of critical information from your fingertips: every time you touch something, or pick something up, your biological sensing system sends your brain information about weight and strength and keeps you from wantonly crushing things. PR2 may not have nerves, but it does have tactile sensors on its grippers, and PhD student and Willow Garage intern Joe Romano has taught PR2 to recognize a variety of tactile cues and act on them intelligently.

This appears to build to some extent on previous PR2 research, like that brilliant bottle squeezing technique we wrote about a year ago… But there’s still a lot of potential here for fine manipulation, as well as keeping PR2 from inadvertently tossing you through a wall.

It’s pretty cool (and borderline shocking) how to see how fast PR2 is improving in all of those little ways that differentiate between a robot that’s simply functional, and a robot that’s… Comfortable. For example, remember PR2′s beer fetching demo? At the end, in order to get the beer from PR2′s hand, you had to pull up somewhat unnaturally on the bottle, moving the robot’s entire arm to let it know that you wanted it to let go. With this new programming, it would be a much more natural motion, since PR2 can tell when you’re touching the bottle and gently release its hand.

Make sure you stick around through the very end of the video to see what the whole point of this research really was.

[ Willow Garage ]

We heard about this a couple weeks ago, but now it’s officially official: PR2 is for sale (go buy one right now!), with a list price of $400,000 (plus shipping). Your choice of color. Luckily, a substantial discount is available for people who’ve demonstrated leadership in the open source community… If you’ve got a track record of providing quality open source software, you can apply to Willow Garage for their Open Source Discount Award, which is worth a staggering $120,000 (!) towards the purchase of a PR2. If that’s not a ringing endorsement of ROS and open source culture, I don’t know what is.

If you’re wondering where that $400k goes, well, robots are expensive, especially robots as comprehensively capable as PR2. The PR2 has lots (and lots) of sensors, each of which can cost well into the tens of thousands of dollars. Mix in a pair of robust arms, all the computer hardware in the base, and assembly and testing by hand, and you end up with a robot that costs about as much as a reasonably sized house. The good news is that as the market grows (not just for the PR2, but for the components separately as well), Willow Garage expects the price to drop dramatically.

I hate to keep making this comparison, but it’s like the personal computer 40 years ago. Computers went from being very expensive, very specific tools (breaking codes) to very expensive, very general tools (doing math problems), at which point everybody with a math problem wanted one, creating a market and forcing computers to get cheaper. This is the point in robotics that (hopefully) has just been crossed with the PR2, which is arguably one of the first seriously capable generalist robots to be commercially available.

Remember, the general idea behind the PR2 is to remove the whole having to build a robot bit from robotics, allowing software developers to focus on, you know, developing software, instead of building robots over and over. Since this is the first really serious attempt at such a robot, the market is both wide open and somewhat undefined, so it’s going to be interesting (to say the least) to see how well PR2 does as a commercial product, as well as who ends up buying one.

Now, if you can’t afford a PR2, there are still some options… You can download a very solid PR2 simulator, right now, for free, and start developing code. Of course, no matter how well something works in simulation, you’ll still need to test out your code on an actual robot, but there are plans in the works (most notably from Bosch, one of the recipients of the original 11 PR2s) to develop a sort of remote testing lab where people without a PR2 of their own can borrow some time on a communal robot for a little bit to run their code on. And don’t forget that ROS runs on lots of stuff, so you can start playing around with it on robots besides the PR2.

And speaking of playing around with it, it’s worth mentioning that there is no provision to keep people from using a PR2 for evil. I asked. So remember, kids: black power outlets and round doorknobs will keep you safe.

For now.

[ PR2 ]

Yesterday, we posted an update on PR2′s for sale status, and included a video from an informal contest sponsored by Willow Garage founder Scott Hassan, inviting PR2 beta teams to submit cool / funny / useful PR2 videos to be judged by himself, his wife, and his kids. The above video (from Ping Chuan Wang, Stephen Miller, Mario Fritz, Trevor Darrell, Pieter Abbeel at UC Berkeley) took first place and a cool $5000, which is way more than any person has ever been paid for folding two socks together.

Second place went to Bosch, for their PR2 mailman, whose name appears to be “Alan:”

UPenn took third place with their one robot band:

Really, each of these videos is deserving of its own post, but I wouldn’t do that to you… You can check out the other six (there’s also PR2 StrongBot, posted yesterday) after the jump. (Read more…)

Now that the PR2 has proven to be mildly successful (to put it, uh, mildly), Willow Garage is getting ready to make them commercially available. As of next month, according to the Willow Garage blog, “you can purchase your very own PR2.” They’re not talking about you, though, unless you can dig $400,000 (ish) out from under your couch cushions. The exact pricing is going to have a few more as yet to be determined aspects, of course, but I’m told that you’ll at least get to choose the color.

While $400k may not be exactly affordable for you personally (and if it is, have you considered sponsoring a quality robotics blog?), commercial availability in general means that well endowed (in the financial sense) research institutions who weren’t lucky enough to get one of the first ten eleven should be able to get their hands on one soon. And that means, more cool PR2 videos for us to enjoy and be amazed at. Oh, and open source research and development and stuff, yeah, that too.

Speaking of cool PR2 videos:

This PR2 is from Georgia Tech, and (if you have no idea what’s going on) the video was inspired by this:

This video (the first one) was part of an informal contest sponsored by Willow Garage founder Scott Hassan, inviting PR2 beta teams to submit cool / funny / useful PR2 videos to be judged by himself, his wife, and his kids. There were about 10 submissions, and we’ll have the rest of the contest videos for you in our next post.

[ PR2 ]

We keep going on about how ROS is all free and open source and platform independent and distributable and adaptable and anything can run it and so on, but this little project from ModLab at UPenn illustrates the whole idea brilliantly. They took a bunch of ckBot modules (which, incidentally, can reassemble themselves after being kicked to pieces) and made some 6 DoF arms with grippers plus a movable base, essentially duplicating all of the basic capabilities of the Willow Garage PR2. As you can see from the vid, the mini PR2 can run programs straight off of the PR2 simulation environment, and even duplicate the actions of the full-size robot, all thanks to ROS.

A bit more on ROS, after the jump. (Read more…)

Oh, this is a safety video and not a how to? Oh.

PR2 is a lot like a little kid… Tons of potential, but best not to leave unsupervised. Except, I guess, that PR2 is the size of a small refrigerator and (according to the safety video) can quite easily squish you, stab you with knives, and set your house on fire. Also, it comes with a wireless kill switch, and I’m pretty sure that little kids don’t have one of those.

[ Willow Garage ] VIA [ I Heart Robotics ]

This is it… PR2 has finally reached its full potential: beer delivery. After many long hours teaching the robot to not ever bring anyone a Bud Light under any circumstances, Willow Garage has successfully completed their beer delivery hackathon. PR2 can now take orders for different kinds of beer, locate the fridge, open the door, scan all the different bottles and locate the tasty ones, pick them up, and then bring them back to you using facial recognition software to tell when it’s found a thirsty person. Oh, and it even remembers to close the fridge door.

And it opens bottles.

If it would just pour the beer into my mouth, I’d never have to move again.

Of course, there are lots of ways to go about getting a beer, and not everybody is going to have the same sort of fridge setup, so some tweaking will be required to get this software working robustly in your house on the PR2 that you don’t have yet. Remember, though, that all of their code is modular and open source, so it’ll be easy for others to take on each module (fridge opening, beer IDs, etc) and adapt them for other fridges and other tastes in beer. I have it on good authority that the Willow Garage team plans to put in many long hours honing their beer delivery software, testing it rigorously to make sure that the PR2 can deliver, say, a bottle of Guinness to a person over and over (and over) without shaking the bottles or spilling a drop. Yep, that’s gonna take a lot of testing, but it’s for science, so it’s worth it.

I stopped by Willow Garage late last week to see this whole process in action, and while it may take a little while (note the time compression on the video), it’s still pretty wild that they put this together in such a short amount of time. And like, the robot brings you beer, how can you not just totally geek out about that?

Don’t forget that there are 11 other PR2s on the loose now, and if this hackathon is any indication, we’re entitled to expect some pretty great stuff in the near future.

[ Willow Garage ]