Robotics: Today & Tomorrow

[VelesHomais]

Here we will discuss new exciting breakthroughs in robotics, both experimental and practical. The field of robotics is considered to be controversial by some, let's all demonstrate respect for everyone's opinions.

Let's start with this short, but wonderful, introduction into some of DARPA commissioned Boston Dynamics latest robots:


And eLEGS, exoskeleton for the disabled:


As well as the military application exoskeleton:


This is just a hypothetical vision, created two years ago, showing an autonomous semi-android that is dispatched to assist countries struggling with high crime rates.

[ramanujann]

I like robotics and I have build and programmed few simple robots in past.
Now I don't have enough time and money (it's quite expensive hobby) so unfortunately I have restricted myself to reading about robotics.

Anyway, here's an interesting humanoid robot called Roboy which is currently under construction (in uni of Zurich). It's expected to be finished and presented in March 2013 (construction time - 9 months). Everyone can donate to help build this robot!

http://www.roboy.org

http://phys.org/news/2012-12-zurich-...-humanoid.html

[Ulpia-Serdica]

ALDEBARAN Robotics announces the launch of the new V4 generation of its NAO robot, equipped with an Intel Atom microprocessor. The first NAO sold over 2000 units worldwide. The NAO robots were mainly acquired by universities for further robotics developments, including Tokyo U which have acquired several of these little robots for studying and further developing since NAO has re-programmable OS.

In parallel with hardware improvements, NAO V4 features a new voice recognition algorithm, developed by Nuance; faster and more reliable, it also incorporates word-spotting capacities, allowing NAO isolate and recognize a defined word in a sentence or in a conversation.

[Ulpia-Serdica]

FESTO has a program aimed at using biology in order to learn about the fluidity of movements. They first make prototypes mimicking animals and try to integrate it in their products used for industrial purposes.

[Ulpia-Serdica]

Here is some more info about the Roboy project You can donate for R&D on their website.

If robots are going to be part of our everyday lives, they’ll need to fit into our homes rather than the factory floor. Few people would be comfortable living with a metal spider on tank treads, so the University of Zurich’s Artificial Intelligence Laboratory (AI Lab) is building a robot toddler called “Roboy.” Using “soft robotics” technology that mimics the human body, the 1.2 meter (3 ft, 11 in) tall humanoid robot is part of an effort to make robots that people are more comfortable with in day-to-day situations.





The University of Zurich’s Artificial Intelligence Laboratory is building a robot toddler called “Roboy"

Roboy doesn't look very endearing at the moment. In fact, it looks more like a cyborg skeleton than a charming child, but it’s still a work in progress. The laboratory’s goal is to build Roboy in only nine months. Work began last June with 15 project partners and over 40 engineers and scientists. These parties are providing expertise and funds through sponsorship and crowdfunding that includes auctioning space on the robot for logos, and hiring it out for business functions when completed.





Roboy's chest assembly

Roboy is based on the laboratory’s previous project, the humanoid, frighteningly cycloptic Eccerobot. Built out of plastic, Roboy is modeled on the human musculoskeletal system, but this mimicry goes beyond the aesthetic. Instead of motors in its joints, Roboy uses motor assemblies that pull elastic cables, so the system operates in a way similar to muscles and tendons. AI Lab claims that this will allow Roboy to move “almost as elegantly as a human.”



Roboyboy's neck assembly

Currently, Roboy is more of a research project than an engineering enterprise. The team is developing new technologies with an eye toward scalable production using CAD and 3D printing to allow for full production of robots within days of development.



Roboy's chest assembly

The purpose of Roboy is to push for the acceptance of service robots by making people more comfortable having them around all the time. With an aging population, AI Lab believes that such service robots will be increasingly important in helping the elderly to continue independent lives.



Roboy muscle unit

Roboy is currently getting a new face chosen by a Facebook contest, and can move its arms. Later, the robot will be covered with a soft skin. Roboy will make its first public appearance at the “Robots on Tour” exhibition on March 9, to celebrate AI Lab’s 25th anniversary.



Roboy's arm



Roboy's features



Roboy's neck assembly

http://www.gizmag.com/roboy/25571/

[Ulpia-Serdica]

An here is some info about the ECCE robot, on which the Roboy project is based on.

The ECCEROBOT project is a spin-off of the CRONOS1 project conducted at the University of Essex. The goal of this project was to investigate machine consciousness through internal modelling. For this purpose the first anthropomimetic robot torso was built (see figure on the right).



Within the ECCEROBOT project we will further enhance this torso, develop a controller for it, and investigate the development of human-like cognitive abilities.

ECCE is being advanced to Roboy since 2011 by a project in cooperation with internationally leading research institutions and partners in the industry at the forefront of developments in mechanics and electronics. One of the first breakthroughs is the anthropomorphic tendon driven arm ANTHROB.

http://www.roboy.org/index.php?cmd=i...ages=1&lang=en

http://eccerobot.org/

[Ulpia-Serdica]

A Viennese page turner: Quidenus's robots are helping build custom e-libraries



Sofie Quidenus uses robots to build digital libraries. "I knew nothing about robotics when I started, but I felt that there would be a huge demand for book digitisation," the Austrian entrepreneur says. Her robot scanners have helped to digitise 80 per cent of all European national-library books. Her next chapter: the ScanGuru, a book-scanner that allows home users to turn books into ebooks.



Vienna-based Quidenus founded her company in June 2004. Her first design was a robotic finger to flip sheet-music pages so performing musicians could keep their hands on their instruments. "Turning pages is the major bottleneck for book digitisation," says Quidenus, 30. "So we started building a robot to make ebooks." Qidenus Technology now sells a dozen models -- the RBS Full costs from €70,000 (£50,000).



The book rests in a cradle and the glass "V" descends to flatten the pages, which are photographed by two cameras and digitised. The "V" then ascends, and the robot finger turns the page. It can digitise up to 3,000 pages per hour at 500dpi resolution.



Qidenus counts the Google Book Project, the Library of Alexandria in Egypt and several national libraries in Europe among its clients, and it has received €1 million from Austrian investors. The next step is to bring the technology to the masses. "The ScanGuru will be smart, but very simple," says Quidenus. "We want people to use it at home."

http://www.wired.co.uk/magazine/arch...se-page-turner

[Ulpia-Serdica]

Robotic Rehab Helps Paralyzed Rats Walk Again



By employing a combination of drugs, electrical stimulation, and robot-assisted rehabilitation, researchers have restored a remarkable degree of voluntary movement in rats paralyzed by a spinal cord injury. After several weeks of treatment, the rodents were able to walk—with some assistance—to retrieve a piece of food, even going up stairs or climbing over a small barrier to get it. The rats' recovery raises hopes that a similar combination strategy could help restore movement in some people with spinal injuries. Indeed, such efforts are already underway.

Spinal injuries cause paralysis because they sever or crush nerve fibers that connect the brain to neurons in the spinal cord that move muscles throughout the body. These fibers, or axons, are the long extensions that convey signals from one end of a neuron to another, and unfortunately, they don't regrow in adults. That's why paralysis from a spinal injury is a lifelong disability. Restoring axons' ability to regrow using growth factors, stem cells, or other therapies has been a longstanding—but frustratingly elusive—goal for researchers.

The new study, which appears in Science today, takes a different approach. Instead of trying to repair the main information superhighway from the brain to the body, Grégoire Courtine, of the Swiss Federal Institute of Technology in Lausanne, and colleagues focused on alternative routes. Most spinal injuries in people do not sever the spinal cord completely, explains Courtine. To approximate this situation in rats, his team made two surgical cuts in the spinal cord, severing all of the direct connections from the brain, but leaving some tissue intact in between the cuts. Then they had the rodents begin a rehab regime intended to bypass the fractured freeway, as it were, by pushing more traffic onto neural back roads and building more of them.

This regime, which began about a week after the rats were injured, lasted about 30 minutes a day. During each session, the researchers injected the animals with a cocktail of drugs to improve the function of rats' neural circuits in the part of the spinal cord involved in leg movements, and they stimulated this area with electrodes. With its spinal cord thus primed for action, a rat was fitted into a harness attached to a robotic device that supported its weight and allowed it to walk forward on its hind legs to the extent that it was able. At first, the rats could not move their legs at all, let alone walk.

But after 2 or 3 weeks, the rodents began taking steps toward a piece of food after a gentle nudge from the robot. By 5 or 6 weeks, they were able to initiate movement on their own and walk to get the food. And after a few additional weeks of intensified rehab, they were able to walk up rat-sized stairs and climb over a small barrier placed in their path. Rats that did not undergo rehab, in contrast, showed no improvement at all. Rats suspended over a moving treadmill that elicited reflex-like stepping movement, did not improve either, suggesting that full recovery depends on making intentional movements, not just any movement.

"It's a really remarkable finding," says Michael Beattie, a neuroscientist at the Brain and Spinal Injury Center at the University of California, San Francisco. Additional experiments in the paper make a compelling case that the rats' recovery is due to new neural connections forming to create a detour around the injury, he says. Beattie notes that Courtine's work suggests that all three components of the rehab strategy—the drugs, the electrical stimulation, and the robot-assisted physical therapy—seem to be necessary to maximize recovery. "I think that it actually provides a lot of hope that this kind of strategy will have a big payoff" in people, Beattie concludes.

A case study published last year reported some recovery of voluntary movements in a man paralyzed in a vehicle accident, after he underwent a combination of electrical stimulation and physical therapy. The new rodent research provides a potential explanation for that patient's recovery, says one of the lead authors of the case study, neuroscientist V. Reggie Edgerton at the University of California, Los Angeles. Edgerton says two more patients are undergoing similar rehab now, and his group hopes to add drug therapy to enhance nerve repair in the future. "We're not there yet," he says. "But the bottom line is, things are still looking good."

As encouraging as the new findings are, Courtine is careful to note the strategy's limitations. For one thing, it wouldn't work if the spinal cord were completely severed. In addition, treated rats could only make voluntary movements while the electrical stimulation was turned on, and the same was mostly true of the patient Edgerton and colleagues worked with. "This is not a cure for spinal cord injury," Courtine says. "It's a promising proof of principle."

http://news.sciencemag.org/scienceno...paralyzed.html

[Ulpia-Serdica]

AirBurr MAV Can Now Self-Right, Is Utterly Unstoppable



In October of 2009, we wrote about the very first version of EPFL's AirBurr micro air vehicle, called HoverMouse. It was an innovative design: a roll cage protected the MAV's engine and flight surfaces, enabling it to crash into walls and floors without damage and then take off again, provided it had enough room to get airborne. Seven iterations later, the AirBurr V8 Samurai includes an active self-righting mechanism that allows it to crash and take off again even in rugged and cluttered environments.

AirBurr's latest trick involves first crashing into something and falling to the ground, which I imagine was a pretty easy thing to get it to do. Second, the MAV rolls over onto its side thanks to a clever arrangement of carbon fiber caging plus a carefully designed center of gravity. Third, AirBurr activates an "Active Recovery System" consisting of carbon fiber legs that deploy out from the body, pushing the body of the MAV into a vertical position from which it can lift off straight up.

Overall, this system is sort of like a cross between a WeebleCopter and that nifty little jumping robot that can self-right after landing.

The big advantage that AirBurr has is that it doesn't have to be super clever apart from the structural design. This isn't to say that the robot isn't clever, but what makes it clever is that it doesn't have to localize, it doesn't have to detect objects, and it doesn't have to rely on all kinds of sensors and onboard computing power. Instead, it can just smash into things and recover and keep on going. As the video points out, this is a very insect-like behavior, and AirBurr is ideal for confined and cluttered environments that are bad for traditional sensors, like disaster areas or anywhere indoors or underground.

According to a forthcoming paper in IEEE Transactions on Robotics (authored by Adam Klaptocz, Ludovic Daler, Adrien Briod, Jean-Christophe Zufferey, and Dario Floreano), the AirBurr Samurai was able to autonomously right itself within 25 seconds in 100% of the time after being manually knocked over on a flat surface. The robot had a little bit of trouble trying to get back up after landing on slopes greater than 10 degrees, and gravel and rocks proved tricky as well. Part of the problem is the autonomous controller, and EPFL

researchers are already working on the AirBurr V9, which will be even smarter:

The active uprighting mechanism described in this paper is a first mechanical implementation that could be extended through improved sensing and control. Initial investigations show that lightweight strain gauges integrated into the legs can measure the shape of the leg and, thus, the force it provides at its tip. Further, strain gauges can measure tension in the string and, thus, contact with obstacles. Such information can be leveraged to create a more intelligent controller that can extract the platform from ever more complex situations.

Now, as we've mentioned, the Samurai is the 8th generation of the AirBurr. Just for fun, here's the whole lineup, dating back to 2009:



AirBurr V1 "HoverMouse": The first prototype of the AirBurr, weighing only 25g, and featuring a gravity-based self-recovery system which allows it to return to a takeoff position after a collision.



AirBurr V1 "HoverMouse": The first prototype of the AirBurr, weighing only 25g, and featuring a gravity-based self-recovery system which allows it to return to a takeoff position after a collision.



AirBurr V3 "Flying Stick": "Flying Stick" is the first AirBurr prototype to feature embedded sensing in the form of the BurrSens V3 Inertial Measurement Unit (IMU). It is capable of autonomous hovering flight stabilization.



AirBurr V4 "Crashy": "Crashy" was designed to test improved collision energy absorption techniques. It uses deflecting carbon-fibre beams and rods to absorb the energy of head-on collisions.



AirBurr V5 "Bumpy": "Bumpy" is the first prototype to have a sense of touch thanks to force sensors strategically placed inside the carbon fibre structure. The signals from these sensors are analyzed using an embedded algorithm and can detect the position and amplitude of a force on its outer ring to an accuracy of around 0.5 N and 10 degrees.



AirBurr V6 "Sticky": "Sticky" is the first attempt at using gecko-inspired dry adhesives to attach to smooth surfaces such as windows.



AirBurr V7 "Multitasky": "Multitasky" is a platform used for experiments in robust orientation stabilization. It is the first AirBurr platform to have altitude stabilization using an ultrasonic distance sensor.



AirBurr V8 "Samurai": "Samurai" is the first flying robot to feature an active self-recovery mechanism, allowing it to return to an upright position and take off after a collision, no matter how it lands on the ground. The mechanism has been successfully tested in a variety of environments ranging from hardwood to gravel to rocks.

http://spectrum.ieee.org/automaton/r...ly-unstoppable

[Cujas]

The futur of robotic will be the biomechanics which is the study of the structure and fonction of biological system to translate him into engineering.

They are many devices in nature which have proven since millenium by the natural selection, and who could be reproduce in technology.

I have ever saw a documentary about it, there is a engineer school in germany who work on it and who got some very interesting result and sometime suprising.

For instance the study of structure of exosquelette of fish was use to built the hand of a robot.

I will try to found this documentary to show you it.

[Ulpia-Serdica]

Spain's Agrobot to manufacture more robots for strawberry harvest



The firm Agrobot from Huelva, Spain, which presented the first prototype capable of harvesting strawberries, will manufacture more units of the robot in the United States.

The engineer who designed the robot, Juan Bravo, said that the tests conducted in American soil have been successful, and that the firm believes that it may even be possible to transfer some of the staff to work there directly. The robot has been used since the summer in the fields of Watsonville, in California, where 40% of the State's strawberries are produced, once California's entrepreneurs saw it working, still as a prototype, and were able to test it in experimental orchards.

The robot reduces harvesting costs by 50% for fresh strawberries and up to 90% for industrial strawberries, which are the ones used for the manufacturing of purees and yoghurts.

The prototype was created at Huelva's Agro-industry Technology Innovation Centre (Adesva), in the town of Lepe, member of the Technological Network of Andalusia (RETA), supported by the Region's Council of Innovation, Science and Business.

http://www.freshplaza.com/news_detail.asp?id=104477

[Spartan_X]

All this effort to make the robots more human in appearance, where does it lead to?



Not to that, i hope.

[brightside.]

I built a bomb disposal robot as my Senior Design Project in university, and was offered Government of Pakistan funds to start a company for the manufacture of such robots. But I was not ready to be an entrepreneur. I know jack about business, and it would be too much work and I could risk spending 5 years trying to build a company that eventually fails.

Anyway, robots can become a huge part of our lives, if only the artificial intelligence aspect improves.

[Atmosphere]

Originally Posted by Spartan_X

All this effort to make the robots more human in appearance, where does it lead to?



Not to that, i hope.

I was actually wondering this as well, in the opening post there is this robocop idea robot. But why does it look like a human. It needs to look around to see whats going on for example which is quite....weird. Why not give it a 360 degree view with 10 camera's. Why does it need a front and a back side even....

[Spartan_X]

Who can say that the military isn't secretly funding this? Maybe they have in mind exactly that ( a terminator-style robot ), to use it as a kind of super-soldier, ideal for covert operations, infiltrations etc.

Of course we are decades away from making Jim Cameron's vision a reality. Most important things we're still lacking are .. first a sufficient power source ( lithium-ion batteries are not enough :p ) and secondly, a true A.I - and that's especially something that may never happen. What scientist will be so idiot to give complete free will to a machine? I don't see that happening.

However, my opinion is that, if not we, our children ( or grandchildren ) will for sure see real androids. For what uses will they be used, it still is open for debate - but seeing what humanity did historically with other inventions, i think is logical that they will be used for war.

[ramanujann]

Development of Artificial Nervous System for Humanoid Robot (Avatar)

http://2045.com/experts/29099.html

[Сталин]

Awesome thread!

[VelesHomais]

Originally Posted by Spartan_X

Who can say that the military isn't secretly funding this?

Nothing secret about it ...

[Atmosphere]

Can't wait to see some updates on PETMAN. It's been a while since the last videos.

Last version from almost a year ago was already able to go up stairs, climb a little bit and jump.

[Spartan_X]

Originally Posted by VelesHomais

Nothing secret about it ...

Yes, true, it is obvious that they are researching it. However, the question about what kind of a power source a completely autonomous robot could use that would allow it to go on for several hours, or days, or even longer without a need for recharge, still remains. Unless of course, they have already solved it somehow.