Trying to always be at the cutting edge of robotic technology (as we are), it's always fun to take a minute to look back and see just how much things have changed in a few short years. The PR2 and ROS, for example, have completely revolutionized robotics research (and a bunch of other things), but as with everything else, they had their origins, including Stanford's PR1. Videos of that robot, plus a lot more, in this week's Video Friday.
So, let's start off with this blast from the past; it's PR1, the precursor to the PR2:
This video showcases PR1, a robot developed by Keenan Wyrobek, Eric Berger, HFM Van der Loos, and Ken Salisbury at Stanford. It is the predecessor of the Willow Garage PR2.
Reference: K. Wyrobek, E. Berger, H.F.M. Van der Loos, and K. Salisbury. Towards a personal robotics development platform: Rationale and design of an intrinsically safe personal robot. In the Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), 2008.
It may look ancient, and we're assuming that a lot in these videos is teleoperation or pre-programmed movements, but this was only published five years ago. How crazy is that?
Via [ Pieter Abbeel ]
Being able to work with robots directly and not have to worry about accidental decapitation is a priority for industrial robot manufacturers, and thankfully, there are industry safety standards that can make it possible. Erik Nieves from Motoman talks about how it works:
Violating robots: just don't do it, people.
[ Yaskawa Motoman ]
You know who else makes industrial robots? ABB. And they've got a giant playground (that they're trying to disguise as a "headquarters and training center") out in Auburn Hills, Michigan:
[ ABB ]
Having witnessed several years of Mech Warfare, I can attest to the fact that building a (consistently) functional robot is darn hard. This robot, from the dudes at Trossen, is a beastly four-legged mech that'll be open sourced when it's complete:
Well, if you run out of ammo, you can always just crush your opponents into little plastic splinters with those stompy robot feet.
[ Trossen Robotics ]
Robotics has brought a lot of benefits to disabled people, and the next generation of prosthetic limbs will be even better, because they'll be controlled directly through existing nerves. DARPA's been working on some amazing prototypes:
A team of researchers at the Rehabilitation Institute of Chicago (RIC) demonstrated a type of peripheral interface called targeted muscle re-innervation (TMR). By rewiring nerves from amputated limbs, new interfaces allow for prosthetic control with existing muscles. Former Army Staff Sgt. Glen Lehman, injured in Iraq, recently demonstrated improved TMR technology.
The really cool bit is that the residual nerves can receive signals as well as send them, giving users some sense of touch through their robotic fingertips:
Researchers at Case Western Reserve University used a flat interface nerve electrode (FINE) to demonstrate direct sensory feedback. By interfacing with residual nerves in the patient's partial limb, some sense of touch by the fingers is restored.
[ DARPA ]
Arguably, the easiest way for a user to train a robot to do something is to just show the robot exactly what to do once, and then have the robot autonomously repeat the task. We've seen plenty of this sort of thing with ground robots and manipulators, but researchers at the University of Toronto are bringing teach and repeat to UAVs as well:
[ UTIAS ]
For some reason, roboticists seem to enjoy teaching their robots how to use swords. Yes, swords. Robopocalypse hellooooo?! The latest is a robotic lightsaber built with the new Robotis CM-900 microcontroller and Dynamixel servos, shown in the vid below during a test. And watch also two of the lightsaber robot arms in a battle at MakerFaire.
[ Martin Mason ]
Next month, RoboCup 2013 kicks off (ha!) in Eindhoven, in the Netherlands:
[ RoboCup 2013 ]
The mechatronics Class at CMU built a bunch of frisbee tossing robots. These bots might remind you of this year's FIRST competition, but the difference is that the CMU robots are autonomous (using vision), while FIRST is nearly entirely remote controlled.
[ CMU ]
Sameer Parekh from Falkor Systems (which just merged with I Heart Engineering) has been working on making aerial robots 117 percent more awesome by giving them the ability to recognize and follow people. They'll make for questionably cuddly pets, but in a practical sense, they'll be able to track you, take pictures for you, and even potentially work as safety systems:
If cuddlyness is a priority, you might want to check out Cuddly. Cuddly is a system to give life to soft, inanimate objects by stuffing your smartphone into them. The way it works is quite clever: using the LED flash and the camera, your phone can tell whether or not you're squeezing, and if so, by how much:
[ Yuta Sugiura ]
And now, this:
[ YouTube ]
Yesterday's JSConf.us featured a whole fleet of AR Drones that were programmed to do something. I don't know what they were programmed to do, but I imagine that this was not it (mild language warning):
[ NodeCopter ]
Aldebaran Robotics has just posted a bunch of very basic tutorials on how to program a NAO, which is great for beginners. Of course, if you're still saving up for a NAO of your own, a bunch of this stuff you can also do in simulation if you'd just like to check things out and learn a thing or two about programming sophisticated (and rather expensive) humanoid robots.
Check out the rest of the tutorials on Aldebaran's YouTube channel.