Beer me! It's for science!
July 11, 2010 8:06 AM Subscribe
After reaching several milestones in their Personal Robot 2 project over the last 2 years, Willow Garage, a robotics research startup located in Menlo Park, CA., gave eleven of their PR2 units to research institutions around the world for 24 months of beta projects and further development.
Some of the current abilities of the PR2 include bringing beer to thirsty engineers, navigating an office to find power outlets to recharge, playing pool, partying like it's 1983 and cleaning up leftovers.
The Georgia Institute of Technology want to focus its research on human-robot interaction and enable the PR2 to interact with drawers, lamps and light switches to assist elderly people, the TU Munich plans to teach the PR2 kitchen tasks like preparing meals or emptying the dishwasher and engineers at UC Berkelely will attempt to upgrade the robots laundry skills.
Want to build your own robot? Here's the hardware description, the specs and the open source Robot Operating System.
Some of the current abilities of the PR2 include bringing beer to thirsty engineers, navigating an office to find power outlets to recharge, playing pool, partying like it's 1983 and cleaning up leftovers.
The Georgia Institute of Technology want to focus its research on human-robot interaction and enable the PR2 to interact with drawers, lamps and light switches to assist elderly people, the TU Munich plans to teach the PR2 kitchen tasks like preparing meals or emptying the dishwasher and engineers at UC Berkelely will attempt to upgrade the robots laundry skills.
Want to build your own robot? Here's the hardware description, the specs and the open source Robot Operating System.
Wow, I welcome our new robot butler overlords if they're going to bring me beer. However, family dogs everywhere are strongly opposed.
posted by ghharr at 8:24 AM on July 11, 2010
posted by ghharr at 8:24 AM on July 11, 2010
Now if we can only get these guys together with the Real Doll people from the other thread, then...we could get even more creeped out.
"It really doesn't seem like it wants to let go of that [insert body part here.]"
posted by umberto at 8:30 AM on July 11, 2010
"It really doesn't seem like it wants to let go of that [insert body part here.]"
posted by umberto at 8:30 AM on July 11, 2010
Why does it go so slow? Would it make more sense to have a bunch of commodity servers running somewhere and the robot just acts as sort of dumb terminal?
posted by geoff. at 8:31 AM on July 11, 2010 [1 favorite]
posted by geoff. at 8:31 AM on July 11, 2010 [1 favorite]
This very much reminds me of Heinlein's "Door into Summer".
posted by rainy at 8:36 AM on July 11, 2010
posted by rainy at 8:36 AM on July 11, 2010
This reminded me of the Colorado hippie programmers at Gamma 2 Robotics, who also built machines to deliver beer (and tea). These robots aren't as slick or autonomous as Willow's, but at least they can bring you a drink without it fizzing everywhere after the trip. And they have creepy menacing voices, which is basically a requirement for any semi-sentient domestic machine where I come from.
posted by The Winsome Parker Lewis at 8:43 AM on July 11, 2010
posted by The Winsome Parker Lewis at 8:43 AM on July 11, 2010
The PR2 robot has two eight-core i7 Xeon system servers on-board, each with 24 GB of RAM, a 500 GB internal hard drive, and a 1.5 TB external removable log drive. The computers and most of the sensors communicate over a 16-port gigabit Ethernet hub with a 32-gigabit backplane. The robot also has an on-board, dual-radio router that can be bridged into a WLAN, as well as a secondary, stand-alone access point for laptop or smart phone access.Egads! My only guess is that sensory inputs exceed the bandwidth capabilities of WLAN. But get that 8-core behemoth out of the picture and you might extend your 2-hour runtime capacity.
posted by geoff. at 8:45 AM on July 11, 2010
I hope they don't think that beer video sells robots. Even with a special fridge and tailor-made input (i.e. a web interface rather than voice command) it took like an hour to get a single beer.
posted by DU at 8:50 AM on July 11, 2010
posted by DU at 8:50 AM on July 11, 2010
It runs so slow because of the massive amount of processing -- navigation, self-localization and positioning, computer vision, and manipulation -- that's happening. Yes, there are instances of "cloud computing" in robotics, where a beefy, plugged-into-the-wall server does a lot of processing and can redirect one or more robots wirelessly, but in applications like this, the control loops are very tight and very high speed and really localized on the hardware. Especially for things like manipulation, where you're combining hardware sensors with "visual servoing" -- visually adjusting position and trajectory of a hand or arm -- the processing all has to be onboard.
It's worth noting that you will never be able to by the PR2, as it is right now, as your own personal Rosie. PR2 is meant to be overkill and provide a ton of hardware capability to software developers. Robotics as a is always in kind of this constant state of "if only" -- "If only we had a standard software platform, I could skip that development and focus on creating new/faster/cheaper/lower power mechanisms!" "If only we had a standard hardware platform, I could skip that development and focus on creating new/faster/more advanced software behaviors!" By providing the PR2 and the ROS to researchers, whether together or separately, the idea is that they're permitting researchers -- researchers, not commercial product developers (yet) -- to skip unnecessary development and focus on what they need to design for new applications.
So no. The videos are not meant to sell robots.
posted by olinerd at 9:27 AM on July 11, 2010 [5 favorites]
It's worth noting that you will never be able to by the PR2, as it is right now, as your own personal Rosie. PR2 is meant to be overkill and provide a ton of hardware capability to software developers. Robotics as a is always in kind of this constant state of "if only" -- "If only we had a standard software platform, I could skip that development and focus on creating new/faster/cheaper/lower power mechanisms!" "If only we had a standard hardware platform, I could skip that development and focus on creating new/faster/more advanced software behaviors!" By providing the PR2 and the ROS to researchers, whether together or separately, the idea is that they're permitting researchers -- researchers, not commercial product developers (yet) -- to skip unnecessary development and focus on what they need to design for new applications.
So no. The videos are not meant to sell robots.
posted by olinerd at 9:27 AM on July 11, 2010 [5 favorites]
If getting a beer to the right person is the most amazing thing they can get a $400,000 robot to do, maybe they should skip the press releases for a couple of years.
posted by digsrus at 9:49 AM on July 11, 2010
posted by digsrus at 9:49 AM on July 11, 2010
So they are "personal" Robots in that they are for research purposes?
posted by DU at 10:08 AM on July 11, 2010
posted by DU at 10:08 AM on July 11, 2010
We've had robots that can fetch cans of beer since the 80's, but now the beer can be in a slightly randomized position.
We'll know we've achieved something when the location of the beer can be anywhere in the space/time continuum.
posted by RobotVoodooPower at 10:13 AM on July 11, 2010
We'll know we've achieved something when the location of the beer can be anywhere in the space/time continuum.
posted by RobotVoodooPower at 10:13 AM on July 11, 2010
If they combine these things with the stabbing robots, we'll really be getting somewhere.
posted by e.e. coli at 10:41 AM on July 11, 2010
posted by e.e. coli at 10:41 AM on July 11, 2010
So they are "personal" Robots in that they are for research purposes?
Well, so that's an interesting question. For a little bit of history, both PR2 and to some degree Willow Garage grew out of a Stanford AI lab project called PR1 (guess what it stands for) which was sort of similarly approximately humanoid. In the industry there seems to be a general assumption that a personal robot should be humanoid. There are plenty of things to argue about that, but that's another discussion. Anyway, this research project was meant to be an early concept/prototype for a personal robot that would be commercialized. Then it spun out into the quasi-commercial entity Willow Garage, and became the PR2.
I've followed WG for a few years now, and their "business plan", such as it is, has really changed a lot over those years. They started with several projects in place -- an autonomous car based on Stanley, an autonomous surface craft (sort of jetski-ish), and the PR2. I think the intention was to really follow these and commercialize them. For a variety of reasons, that did not happen, and two of the three projects were canceled. PR2 remained, but the focus shifted from what seemed to be "commercialize this as a personal robot!" to "keep the form factor and name, but target researchers who will evolve this into something truly personal and commercializable, and focus more on our open-source ROS and enabling this research to happen."
WG is currently very much a privately-held research lab, not a robotics company in the style of iRobot or something similar. They've given away the PR2s, they provide ROS free of charge (though there are some licensing options for commercial applications), and their only other product at the moment is Texai, a prototype telepresence robot that will be competing against some far better engineered and more developed offerings from Anybots and VGO. They are a long way from really selling anything off the shelf.
That's okay, though. There's room for this in the industry. What concerns me is always the kneejerk reactions to videos like this and others about how this wouldn't sell, or it's too slow, or whatever. In the grand scheme of robotics, what these PR2s are doing is actually incredibly impressive. The video of it folding towels? Freaking ridiculous. It takes forever, true, but what it's doing is really incredible from a technological standpoint. But these are definitely dog-and-pony-show videos. So is ASIMO from Honda, and the Toyota robots that play the trumpet, and the Justin robot from Germany that opens jars, and all these others. These guys are years, sometimes decades, away from true usefulness and commercialization at an affordable price. There are plenty of robots that are already there (see: Roomba, PackBot and Talon, etc), they're just not as fun to watch and are comparatively very simple.
I think a good analogy is the biomedical industry. The time between "we did an experiment that might maybe kind of show a drug's effectiveness against certain controlled environments where cancer was growing" and "cure for cancer!" is years, many many years. The press release happens as soon as the experimental results are in and verified, and if there could be YouTube videos of something killing cancer cells I'm sure we'd be all over it. But it's not ready for prime time yet. Same with these robots.
posted by olinerd at 11:14 AM on July 11, 2010 [5 favorites]
Well, so that's an interesting question. For a little bit of history, both PR2 and to some degree Willow Garage grew out of a Stanford AI lab project called PR1 (guess what it stands for) which was sort of similarly approximately humanoid. In the industry there seems to be a general assumption that a personal robot should be humanoid. There are plenty of things to argue about that, but that's another discussion. Anyway, this research project was meant to be an early concept/prototype for a personal robot that would be commercialized. Then it spun out into the quasi-commercial entity Willow Garage, and became the PR2.
I've followed WG for a few years now, and their "business plan", such as it is, has really changed a lot over those years. They started with several projects in place -- an autonomous car based on Stanley, an autonomous surface craft (sort of jetski-ish), and the PR2. I think the intention was to really follow these and commercialize them. For a variety of reasons, that did not happen, and two of the three projects were canceled. PR2 remained, but the focus shifted from what seemed to be "commercialize this as a personal robot!" to "keep the form factor and name, but target researchers who will evolve this into something truly personal and commercializable, and focus more on our open-source ROS and enabling this research to happen."
WG is currently very much a privately-held research lab, not a robotics company in the style of iRobot or something similar. They've given away the PR2s, they provide ROS free of charge (though there are some licensing options for commercial applications), and their only other product at the moment is Texai, a prototype telepresence robot that will be competing against some far better engineered and more developed offerings from Anybots and VGO. They are a long way from really selling anything off the shelf.
That's okay, though. There's room for this in the industry. What concerns me is always the kneejerk reactions to videos like this and others about how this wouldn't sell, or it's too slow, or whatever. In the grand scheme of robotics, what these PR2s are doing is actually incredibly impressive. The video of it folding towels? Freaking ridiculous. It takes forever, true, but what it's doing is really incredible from a technological standpoint. But these are definitely dog-and-pony-show videos. So is ASIMO from Honda, and the Toyota robots that play the trumpet, and the Justin robot from Germany that opens jars, and all these others. These guys are years, sometimes decades, away from true usefulness and commercialization at an affordable price. There are plenty of robots that are already there (see: Roomba, PackBot and Talon, etc), they're just not as fun to watch and are comparatively very simple.
I think a good analogy is the biomedical industry. The time between "we did an experiment that might maybe kind of show a drug's effectiveness against certain controlled environments where cancer was growing" and "cure for cancer!" is years, many many years. The press release happens as soon as the experimental results are in and verified, and if there could be YouTube videos of something killing cancer cells I'm sure we'd be all over it. But it's not ready for prime time yet. Same with these robots.
posted by olinerd at 11:14 AM on July 11, 2010 [5 favorites]
If robots getting drinks are what you're interested in, there's a competition called RoboCup@Home for robots performing tasks like getting drinks for guests and identifying and delivering items, delivering drinks and things like that. The teams also have to publicise quite a bit of technical information, in team description papers - a lot of the equipment used is commercial off-the-shelf hardware that anyone can buy.
YouTube has plenty of videos - but not normally with such high production values as the willow garage videos have.
posted by Mike1024 at 12:03 PM on July 11, 2010 [1 favorite]
YouTube has plenty of videos - but not normally with such high production values as the willow garage videos have.
posted by Mike1024 at 12:03 PM on July 11, 2010 [1 favorite]
Well if all you want is a robot with your beer there's this
But this is crazy cool. Yeah, it's slow and it's not practical but it's the first step.
posted by lumpenprole at 1:52 PM on July 11, 2010
But this is crazy cool. Yeah, it's slow and it's not practical but it's the first step.
posted by lumpenprole at 1:52 PM on July 11, 2010
What concerns me is always the kneejerk reactions to videos like this and others about how this wouldn't sell, or it's too slow, or whatever. In the grand scheme of robotics, what these PR2s are doing is actually incredibly impressive.
I would not have had that reaction if they didn't call it "personal". That kind of implies it's for, you know, people.
posted by DU at 2:28 PM on July 11, 2010
I would not have had that reaction if they didn't call it "personal". That kind of implies it's for, you know, people.
posted by DU at 2:28 PM on July 11, 2010
The robot looks quite large and potentially dangerous. I would be a little nervous with it hulking about, swinging its claws with frightening accuracy. I will take my robots small and fairly weak until I can really be sure they can be trusted.
posted by Sukiari at 3:06 PM on July 11, 2010
posted by Sukiari at 3:06 PM on July 11, 2010
So they are "personal" Robots in that they are for research purposes?
I think they are "personal" robots because they aren't primarily designed for killing people. As in, if it isn't an anti-personal robot, it must be a personal robot.
I didn't look into this too closely, but from what I see I'm really quite impressed. Both by how damned difficult all this stuff is, but also by their technical accomplishments, and their focus on human scale life-improving applications. Nonetheless, I'm forced to wonder how far they actually are from military money. All that money has to come from somewhere, and we are talking millions of dollars per year here..
posted by Chuckles at 5:44 PM on July 11, 2010
I think they are "personal" robots because they aren't primarily designed for killing people. As in, if it isn't an anti-personal robot, it must be a personal robot.
I didn't look into this too closely, but from what I see I'm really quite impressed. Both by how damned difficult all this stuff is, but also by their technical accomplishments, and their focus on human scale life-improving applications. Nonetheless, I'm forced to wonder how far they actually are from military money. All that money has to come from somewhere, and we are talking millions of dollars per year here..
posted by Chuckles at 5:44 PM on July 11, 2010
Willow Garage, last I knew, is funded by the personal investment of an ex-Googler. They also have strong ties to the Silicon Valley venture community, in case the Google funds ever run out.
Of course, Google is, you know, the Internet, and the Internet was developed by the military, so if you really want to go six-degrees on this you don't have to hop too far.
Bit of an axe to grind there, Chuckles?
posted by olinerd at 6:09 PM on July 11, 2010
Of course, Google is, you know, the Internet, and the Internet was developed by the military, so if you really want to go six-degrees on this you don't have to hop too far.
Bit of an axe to grind there, Chuckles?
posted by olinerd at 6:09 PM on July 11, 2010
Willow Garage, last I knew, is funded by the personal investment of an ex-Googler. They also have strong ties to the Silicon Valley venture community, in case the Google funds ever run out.
Interesting.. Ex-Googlers, I suppose, have lots of money to burn. I guess it isn't only space that attracts that kind of speculative "investing". However, venture communities--nowadays even Silicon Valley venture communities--require products with profit potential before they'll give out money.
Anyway, why would I have an Axe to grind with Willow Garage? I'd never heard of them until yesterday. Actually, I'm really impressed! I bet the staff take a big pay cut to work for an ethical employer instead of one of the usual suspects.
posted by Chuckles at 12:51 AM on July 12, 2010
Interesting.. Ex-Googlers, I suppose, have lots of money to burn. I guess it isn't only space that attracts that kind of speculative "investing". However, venture communities--nowadays even Silicon Valley venture communities--require products with profit potential before they'll give out money.
Anyway, why would I have an Axe to grind with Willow Garage? I'd never heard of them until yesterday. Actually, I'm really impressed! I bet the staff take a big pay cut to work for an ethical employer instead of one of the usual suspects.
posted by Chuckles at 12:51 AM on July 12, 2010
So they are "personal" Robots in that they are for research purposes?
I believe what they mean is "this is a prototype of a robot, designed with the eventual goal of producing a personal robot"
That's to set it apart from the many other things that can be described as robots. This is not an industrial robot arm used to weld or paint cars. It's not a self-driving car. It's not a submersible for inspecting oil spils. It's not a mars rover. It's not a bomb disposal robot or a search-and-rescue robot. It's not a toy like a RoboSapien. It's not a UAV or a "smart bomb". It's not a general purpose research platform like the P3-DX. It's not an automatic guided vehicle, following a line on a factory floor.
posted by Mike1024 at 1:29 AM on July 12, 2010
I believe what they mean is "this is a prototype of a robot, designed with the eventual goal of producing a personal robot"
That's to set it apart from the many other things that can be described as robots. This is not an industrial robot arm used to weld or paint cars. It's not a self-driving car. It's not a submersible for inspecting oil spils. It's not a mars rover. It's not a bomb disposal robot or a search-and-rescue robot. It's not a toy like a RoboSapien. It's not a UAV or a "smart bomb". It's not a general purpose research platform like the P3-DX. It's not an automatic guided vehicle, following a line on a factory floor.
posted by Mike1024 at 1:29 AM on July 12, 2010
Chuckles, the axe to grind I was talking about is the anti-military research stuff. As someone working for what she considers a very ethical company (full of Cambridge, Massachusetts hippies, no less) making military robots (as well as scientific robots, like some that are helping in the Gulf oil spill, which look amazingly identical to the evil military bots), that kind of sucks. It's not all guns and ammo and bad people, and the robotics industry owes a ton of its success to initial investments and small-business research grants from the government and DoD. Just like, as I indicated earlier, technologies like the Internet.
Yeah, the Silicon Valley ex-dot-com investors have kind of entered three primary fields they feel are the "next big thing": robotics, renewable energy, and commercial space flight. The kind of key difference for all of them is that this involves... you know... hardware. It's a totally different ball game from the software products that made them rich, and it makes it kind of fascinating to watch, especially comparing their development paths to the companies that bootstrapped on government SBIRs instead of venture funding.
posted by olinerd at 5:47 AM on July 12, 2010
Yeah, the Silicon Valley ex-dot-com investors have kind of entered three primary fields they feel are the "next big thing": robotics, renewable energy, and commercial space flight. The kind of key difference for all of them is that this involves... you know... hardware. It's a totally different ball game from the software products that made them rich, and it makes it kind of fascinating to watch, especially comparing their development paths to the companies that bootstrapped on government SBIRs instead of venture funding.
posted by olinerd at 5:47 AM on July 12, 2010
The PR2 robot has two eight-core i7 Xeon system servers on-board, each with 24 GB of RAM, a 500 GB internal hard drive, and a 1.5 TB external removable log drive. The computers and most of the sensors communicate over a 16-port gigabit Ethernet hub with a 32-gigabit backplane. The robot also has an on-board, dual-radio router that can be bridged into a WLAN, as well as a secondary, stand-alone access point for laptop or smart phone access.Very impressive.
But my dog would still hump it to death and then eat it if it wandered through my living room.
posted by quin at 3:26 PM on July 12, 2010
« Older Take a Little Trip and See | People of the Stony Shore Newer »
This thread has been archived and is closed to new comments
posted by rainy at 8:23 AM on July 11, 2010