The Automated Foosball Table was created by a group of senior engineering students at Georgia Tech.  This project was mentioned briefly on Hack a Day last week, but we think it deserves a closer look, so here it is.  This is both a really fun project (Seriously, who doesn’t love a good game of foosball?) and an interesting experiment in human/machine interaction.  It doesn’t look like it could beat a human foosball master (so we won’t have a scene like when Garry Kasparov wept openly and cursed out Deep Blue after it beat him at chess), but the potential is there.

Four human-controlled rows, four robotic ones. At the right, you can see the custom PIC-based servo controller board.  Let’s take a closer look at the drive mechanisms:

Recognize that actuator?  That’s a Robotis AX-12 Dynamixel.  It’s the cornerstone of the wildly popular Bioloid system, and it’s one of our favorite robot servos on the market.  These actuators can be set to servo mode or continuous rotation mode.  In CR mode, you can get accurate positional feedback.  What you end up with is functionally the same as a highly accurate, powerful, serially controlled stepping motor; at a fraction of the cost.

Here’s the whole rig.  As you can see you have the table, a camera watching the table, a computer processing the video feed and sending commands to the actuator controller board, and actuators pulling and twisting the handles.

The team wrote their tracking software, which tracks the ball and maneuvers the players accordingly, in Java.  Matlab was considered, but the plan was scrapped because Matlab is too resource-hungry.

The custom actuator control board is pretty cool.  Two different types of outputs for the two styles of servos.  The kicking motion is handled by standard PWM hobby servos, while the lateral motion is handled by AX-12 Dynamixels.

Project home page
Found via Hack a Day

All images belong to the design team.  They were clipped from the project’s final report document or captured from the video.

Ok, this is just too damn cool. The Surveyor crew have integrated the SRV-1 controller with a quad-rotor helicopter.

The heli, called "X-3D-BL Scientific," is from a German company called Ascending Technologies GmbH.  I knew it was easy to use the SRV-1 controller for other wheeled vehicles, but seeing it on a UAV is pretty exciting.  They don’t have too much info up yet, but it sounds promising.  Here’s a video of the aircraft (without SRV-1 controller) in action:

Check out the original post, and keep an eye on the Surveyor Robotics Journal for updates!

The immensely popular Surveyor SRV-1 has just received some major upgrades!  The tracked base remains unchanged, but the electronics have gotten a complete overhaul.  Wi-Fi has replaced the slow and sometimes unreliable ZigBee wireless link, the camera can now grab video at a maximum resolution of 1280 x 1024, an Analog Devices Blackfin processor has replaced the comparatively puny ARM7 processor, and they've even added laser pointers for distance sensing.

Wi-Fi:

The SRV-1 now benefits from the extended range and faster data rates of Wi-Fi.  All you need is an 802.11 b/g wireless LAN card and you're ready to start spying on the neighbors.

Camera:

Speaking of spying on your neighbors, this 1.3 Megapixel camera is roughly on par with most cell phone cameras, with a maximum resolution of 1280 x 1024.  Unlike your cell phone camera, this one’s going to be riding on tank treads.

Main Processor:

Here it is, fully assembled.  The red board on top carries the 1000MIPS 500MHz Analog Devices Blackfin BF537 processor.  [Editor’s note: When I was 17, I built my first computer around the 500MHz Pentium, which was the top of the line at the time.  Now you can get all that power on a circuit board that’s smaller than a playing card.  Crazy, huh?]

Lasers:

Fact: lasers make anything cooler.  The new SRV-1 uses two parallel lasers as a low-cost range-finding system.  Try not to blind your friends/family members/pets.

Links:

Purchase the new and improved SRV-1

More information (Manufacturer’s site)

No keyboard or mouse needed! Teach your robot what to do by using motions and gestures.

Skilligent is taking a new approach to the area of human robot interactions, their press release explains:

Skilligent Robot Learning and Behavior Control System is a software product which enables robots to learn procedures and skills directly from human users. The product is a set of software components specifically designed for straightforward integration into control loops of PC-controlled robots.

The software needs to be integrated with sensors, motors, actuators and robot’s control system via a set of open APIs. On one side, the software controls every motion of the robot. On the other side, the software uses a video camera, a microphone and other sensors to receive guidance from a human user.

The software analyzes human’s gestures, looks at the objects presented by the user, listens to the sounds and tries to guess what robot is supposed to do.

Through trial and error, the robot understands what needs to be done and associates learned behaviors with stimuli used by the human. Over a few training sessions, the robot refines its understanding of the domain, the procedures and skills required to serve the user.

Robot Learning Technology is especially useful in the following domains:
1. Industrial/service robots for small and medium sized businesses
2. Defense and law enforcement
3. Health-care and Elder-care robots
4. Research and Education robotics

Skilligent is currently seeking partnerships for further sales and distribution of their software packages. Interested parties can contact them through their website.

Skilligent LLC is headquartered in Dallas, Texas and operates an off-shore research and development center.

http://www.skilligent.com

Ever wish you didn’t have to hire a babysitter to watch the kids? Well, that time is certainly near, if it isn’t already here if Korea-based KornTech (English translation) has anything to do with it:

At 1-metre tall, Rogun, a robot babysitter and security guard uses two cameras for eyes for facial recognition abilities and is able to recognize “good guys vs. bad guys”. While Rogun runs around your home playing with your kids, it can broadcast video wirelessly so that you can check up on your kids throughout the day. It can also act as a videophone and wireless Internet, so when everyone is gone or asleep, and there’s a stranger lurking around in your home that Rogun doesn’t recognize, it’ll give you a call so you can sneak back into your home and beat the stranger with a wooden spoon!

Rogun is the culmination of only three years’ worth of Korean bipedal robot research, and although prototypes have been very expensive to produce, KornTech expects that a mass-produced Rogun will reach a predicted price point of around US$5000.

Read the full article via Gizmag

Ok. I’ve seen you strutting around at the competitions, with your fancy clothes and your fancy platinum jewelry and your fancy modded KHR-2HV. I know you think you’re the O.G. of the hobby robotics scene. Well… you’re not. I’m sorry. You’re not as cool as you think you are, because you do not have this:

The Bioloid Expert Kit. Shiny.Now available from Trossen Robotics

Compared to the Comprehensive kit, the Expert kit boasts a slew of additional hardware and software. In addition to extra frames, actuators, sensors, and controllers; there are a number of new items that set this kit apart from all of the others. More programming tools have been added, as well. The manual is full of tutorials for programming your Bioloid using the Motion Editor, Behavior Editor, and C++. Now that I’ve got your attention, let’s take a look under the hood.

Open up the top compartment, and you’ll find a complete Comprehensive Kit. Eighteen AX-12+ actuators, one AX-S1 sensor module, CM-5 controller, and all the usual trimmings.

Remove the top compartment. This action requires the unbuckling of four latches, so the initial un-boxing is very dramatic. The bottom compartment is where the real fun is. You’ll find three more actuators, two more sensor modules, and another complete Comprehensive frame set; but that’s not all!

What you’re looking at here is the Robotis wireless camera set. The receiver can connect to a computer via USB, plus it has an RCA composite video output so you can connect it directly to a TV.

The camera itself is about 1 cubic inch, and it sports two AX-series Dynamixel connectors. The Expert Kit CD contains image processing tools, which you can integrate into your programs to add motion tracking, line following, path detection, and many other vision-related capabilities!

Everything’s going wireless! You can install these Zig100 modules in the CM-5 controllers for exchanging data between robots, or you can use one CM-5 as a remote to control the other. And what’s this? A Zigbee to RS232 converter! Now you can command your Bioloid creations wirelessly from your PC. (I know it’s already pretty easy to accomplish this by hacking a Bluetooth transceiver, but now there’s an off-the-shelf solution for less adventurous users.)

USB2Dynamixel adapter. I’m really excited about this particular gadget. It allows you to control a network of Dynamixels directly from your computer, bypassing the CM-5 controller entirely!

Bonus! In addition to the standard Bioloid 3-pin TTL port, the USB2Dynamixel has a 4-pin RS485 port, which would theoretically allow you to control Robotis’ shigh-end DX series of Dynamixels. I say theoretically because I haven’t tried it yet, and there’s no info in the manual about controlling RS485 Dynamixels.

In addition to all of the previously mentioned goodies, you get a pile of fuses and a nice fine-tipped screwdriver.Also, our kit contained an impressive quantity of silica gel packets, ensuring that everything is free from moisture. I know it looks delicious, but the packaging clearly states that it should be thrown away rather than eaten.

Well, thanks for reading the Bioloid Expert Kit Unboxing Extravaganza! Final word: This kit is just stacked full of awesome stuff, and I can’t wait to see all the different ideas that it will inevitably inspire.

Um, holy crap this is impressive. (As long as it’s not faking it)

Can’t tell you much other than we stumbled upon this video on Youtube. It was posted by an Alex Vogler and here is his homepage.

Scientists have been researching what “beauty” is for a long time.  We have known for some time that symmetry, smoothness, and “average” features play a large role in what we humans determine to be beautiful. The theory is that we are attracted to mate with others who show the least amount of flaws indicating their superior genetic makeup. This website lets you combine multiple faces to create a beautiful average of your liking. It’s a little like playing perverted Frankenstein and it is a bit addictive.

Two Australians at the University of Technology Sydney have invented a software program that distinguishes a beautiful face from a less attractive one.

Their invention uses facial measurements, related to ratios and images of models, actors and some 200 women from around the world.

After a photograph of a woman’s face is put into the program, it returns a beauty rating of between one and 10.

University of Technology Sydney computer scientists Hatice Gunes and Massimo Piccardi believe the software could be used in fields such as cosmetic surgery.

Read Story

There is an interesting discussion happening over at robots.net around the theory of using mirrors to create stereo vision from single cameras.

Nelson Bridwell writes, “One of the many excuses for not using stereo vision on mobile robots is the complexity and expense of arranging two synchronized cameras that can simultaneously capture image frames from different positions. However, in many cases it is possible to capture simultaneous images from two different positions, using only a single camera and a single element mirror arranged in a very simple geometry.” For more details on this idea plus diagrams and photos, see Real Time Stereo from a Single Camera

Link to discussion

related: Real Time Stereo from a Single Camera by Nelson Bridwell

For the first time, MIT scientists have applied a computer model of how the brain processes visual information to a complex, real world task: recognizing the objects in a busy street scene. The researchers were pleasantly surprised at the power of this new approach.

“People have been talking about computers imitating the brain for a long time,� said Tomaso Poggio, the Eugene McDermott Professor of Brain and Cognitive Sciences and a member of the McGovern Institute for Brain Research at MIT. “That was Alan Turing’s original motivation in the 1940s. But in the last 50 years, computer science and AI (artificial intelligence) have developed independently of neuroscience.�

keep reading article