Automaton

Two years ago at Gadgetoff in New York City, I saw John Abele speak about his family's search for the USS Grunion [video], a submarine his father captained that was lost during WWII. The Navy listed it as "missing, presumed lost" and never discovered what had happened to it.

The three Abele brothers pooled their resources and funded an expedition to the Aleutian Islands in the Pacific to search for the Grunion near the area it was thought to have last had radio contact (they have chronicled the entire process on their blog). They not only wanted to locate it and verify that it had sunk, but they wanted to discover what had happened -- a weapons malfunction? An enemy torpedo?

First, they had to locate it. They knew approximately where it was last located; by using a sonar towfish towed by an Alaskan crab fishing boat, they obtained a gritty sonar image and rough GPS position of an object they thought was likely to be the Grunion. A year later -- due to weather conditions in that part of the Pacific, they only had a limited window each year in August to do this work -- they decided to go in closer. To do this, they used a MAX Rover ROV from Deep Sea Systems. The MAX Rover, weighing in at nearly a ton, is a boxy underwater vehicle with two manipulator arms, video cameras, and an imaging sonar. Piloting the ROV over its tether to a spot more than a mile under water, they were able to locate the wreck and get in close for imagery of the hull and what damage they could find.

This image from the Grunion blog shows what a submarine periscope looks like after more than half a century at the bottom of the ocean. All of the photographs -- including some incredible imagery from the ROV camera -- is here.

After the August 2007 ROV expedition, based on the shape of the wreck (things like the prop guards that are characteristic of certain classes of submarines) the Abeles were confident they had found what they were looking for and began comparing the imagery to historic accounts on what might have happened during its last enemy engagement that could have caused it to sink. Though full forensic analysis is still ongoing, they have a lot more information to work with, as the image below shows.

From the blog, an artist's rendition of the Grunion wreck, based on the imagery brought back from the ROV expedition.

This week, the Navy officially confirmed that the submarine they discovered was in fact the USS Grunion. As this and other articles indicate, some of the families refused to move on during the war -- not accepting payouts from the Navy or widows deciding not to remarry -- because of the hope that the crewmen could still be alive, since the Navy had never confirmed the loss of the sub. More than six decades later, these families finally have closure.

Nothing that this ROV did is particularly new or notable in robotics, but I think it's a really wonderful application of technology normally reserved for defense or scientific use. Without the ROV, finding this needle in a haystack would have been next to impossible.

In 2009, the IEEE will celebrate its 125th anniversary.*

Some luminaries are already sending their happy anniversary messages. Watch Gordon Moore, Vint Cerf, and ... Asimo!?

Click on the image to watch the video:

* The official date is 13 May 2009, and the whole history of how the IEEE came about is quite interesting -- if a bit convoluted -- starting in 1884 with the founding of the American Institute of Electrical Engineers (AIEE), which later merged with the Institute of Radio Engineers (IRE). Don't ask me why AIEE + IRE = IEEE, for Institute of Electrical and Electronics Engineers, but it's clear that there was a clear distinction back then between those who worked with, say, power transformers and those who worked with vacuum tubes -- hence electrical and electronics engineers. For more, check Wikipedia's entry on the IEEE and IEEE's official 125th anniversary site.

There were many robot contests this past summer, both head-to-head and web-based. Yet although the competitors learn many robot-building skills, these skills do not transfer well to others. Competitors often keep what they learned secret to improve their chances, and rarely is anything documented or shared.

So back in the summer of 2007 I created what I believe was the first ever web-based robot contest. But unlike the web-based contests that followed, it wasn't a 'coolest robot wins' type contest. Instead, this bi-annual contest was judged for the 'most useful robot tutorial'. The idea was that non-competitors could also learn from the experience, and in the long term create a robot tutorial database. In return, I offered winners a choice between a $100 check or one of my Axon microcontrollers. The next contest is planned to be judged early February 2009.

August 2008 Winners:

Stanford's "Introduction to Robotics" (CS223A) -- and other popular computer science and engineering courses -- are now available free online.

The robotics intro course, taught by Professor Oussama Khatib, includes over 16 hours of lecture videos, as well as notes, handouts, assignments, and solutions.

It's essentially the same content the Stanford engineering students have access to. Well ... you can't ask questions or get feedback on the problem sets, but still, it's an amazing resource. (The site says there will be communities where users will be able to discuss coursework.)

Here's the course description:

The purpose of this course is to introduce you to basics of modeling, design, planning, and control of robot systems. In essence, the material treated in this course is a brief survey of relevant results from geometry, kinematics, statics, dynamics, and control.

The course is presented in a standard format of lectures, readings and problem sets. There will be an in-class midterm and final examination. These examinations will be open book. Lectures will be based mainly, but not exclusively, on material in the Lecture Notes book. Lectures will follow roughly the same sequence as the material presented in the book, so it can be read in anticipation of the lectures

Topics: robotics foundations in kinematics, dynamics, control, motion planning, trajectory generation, programming and design.

Prerequisites: matrix algebra.

The Stanford project, called Stanford Engineering Everywhere, follows other courseware initiatives like MIT's OpenCourseWare and, in the area of robotics, the RoboticsCourseWare.org, which we discussed early this year.

The other courses Stanford will offer:

Programming Methodology (CS106A)

Programming Abstractions (CS106B)

Programming Paradigms (CS107)

Natural Language Processing (CS224N)

Machine Learning (CS229)

The Fourier Transform and its Applications (EE261)

Introduction to Linear Dynamical Systems (EE263)

Convex Optimization I (EE364A)

Convex Optimization II (EE364B)

Um. Can't wait to finish the robotics course to start "The Fourier Transform and its Applications" -- I so needed a refresher! ;)

Thanks, Sam!

Automaton reader Davide wrote to let us know that his team at Pal Technology has developed REEM-B, a humanoid robot designed to assist humans with everyday tasks. They recently presented a teleoperated prototype at an event in Abu Dhabi, where the company is based (though much of the development took place in Barcelona, Spain). This YouTube video shows a fancy introduction of the robot (and looks something like what I hope commercials for robots will look like one day on TV), but the following video is a little more informational about what REEM-B can do:

Davide tells us:

The robot has some unique features: its batteries last more than 2 hours (twice as much as Asimo). It has 41 degrees of freedom and it has object, face and voice recognition. It's the first biped robot implementing laser-based SLAM [Simultaneous Localization and Mapping] and it is probably the strongest humanoid in the world, since it can walk carrying up to 14 kg. The first robot (REEM-A) was created in just 1 year, and REEM-B in the following 2 years.

He also points out that his team is hiring.

Like Asimo, REEM-B still hasn't achieved full autonomy, requiring a human operator to tell it where to go and what to do. "Pour me a glass of Coke" isn't understood; specific tasks must be commanded on a lower level. However, the mapping capabilities, manipulation, and facial recognition are impressive. Plenty more videos are available here.

I would like to extend my generous offer to test out any prototypes in my home. I've got some dishes that need washing.

There's been quite a bit going on with the Boston-based robotics companies lately. Here's a quick roundup:

• iRobot can't seem to go long without making news, and the last couple weeks have been no exception. Not only did CTO Rod Brooks leave to start a new company (as reported last week), but a couple new Roomba versions were released and news of another underwater robotic technology acquisition was announced. The Nekton purchase, together with the Seaglider license they acquired back in June, shows they're making an aggressive grab for a piece of the UUV market.

• A company called North End Technologies has spent the last couple years showing up at robotics-related events in Boston but they haven't shed any light on what they've been up to. A recent Mass High Tech article says they're "incorporating robotics into multi-media technology", and their website says they're going to "dramatically change how people interact with the world." Perhaps they're jumping on to the telepresence bandwagon?

• Roomba designer Joe Jones left iRobot several years ago to start a company called Q Robotics with a mission to build "low-cost, practical robots." Now Q Robotics is Harvest Automation, and their first product is a small robot that can pick up and move potted plants that may be used in greenhouses or nurseries to organize their stock.

I've just finished Edison's Eve by Gaby Wood. The book traces the conception and development of mechanical life -- better described as automata rather than robots -- but with a nod to how ideas from the Enlightenment era could have brought about some of the state-of-the-art robotics and AI research being done today.

The introduction is actually my favorite part of the book. Wood goes into the philosophy behind many early designs of artificial life and mechanical people, stemming from humanity's quest to better understand itself through creation (one of my favorite themes of sci-fi and part of the reason I'm so geeked out about robots). She also goes behind the scenes at MIT's AI lab (albeit the 2002 version -- much has changed since then). There's a great quote from then-director Rodney Brooks, describing the challenges we face in creating something like us when we don't truly understand what we are:

"[maybe] we're just too ... ignorant. We're sitting here on this flat earth, contemplating the heavens above -- if we think we're on a flat earth, we're just not asking the right questions."

SMAV platform and control electronics. © 2008 http://lis.epfl.ch

NewScientist reports on a project by the South African National Health Laboratory Service to use robotic planes as the artificial equivalent to carrier pigeons. The small autonomous UAVs could carry medical samples including blood or sputum for testing, or even deliver units of blood for transfusions and drop snake antivenom to stricken victims. The robot planes are faster than conventional couriers and may be the only option to reach remote clinics frequently inaccessible by roads after heavy rains.

Interestingly the Laboratory of Intelligent Systems at the Swiss Federal Institute of Technology (EPFL) in Lausanne (my previous lab) has been developing and using a flying platform identical to that of the African project for the past two years. The SMAVNET project aims at designing a swarm of Micro Air Vehicles (SMAVs) capable of autonomously establishing emergency wireless networks (SMAVNETs) between multiple ground-users in a disaster area. A swarm of flying robots could replace damaged, inexistent or congested networks. For more information check out their video on Bio-Inspired Flying Robots (full version here) which grabbed the Best Video Award at the AI Video Competition earlier this year.

The flying platform used in both projects is promising because it combines affordable hardware with efficient aerodynamics and easy handling. As I know from some of my own flight attempts, it's design with the plane's rotor mounted behind the wing also makes the platform extremely robust - an essential property for autonomous landing attempts among other things.

A giant piece of robot art has recently roamed the streets of Liverpool, UK. The robot spider, 50 feet high and weighing in at a stunning 37 tons, was created by a team of French theatrical engineers to mark Liverpool's "European Capital of Culture" celebration. The spider named "La Princesse" is the latest brainchild of François Delaroziÿre (who was also behind the Sultan's Elephant spotted in London a couple of years ago) and his company La Machine.

Made from steel and reclaimed poplar the robot had 50 hydraulic axes and stunned crowds on its journey through the city, spewing water, wind, smoke, flames, and even snow in the process.

The project's website has some fascinating pictures of the beast.

Image: La Machine, (c) Matthew Andrews 2008

This month's edition of Spectrum contains a wonderful article Erico put together about a subject near and dear to my heart: gliders. Gliders are a type of autonomous underwater vehicle (AUV) that is driven by a variable buoyancy system instead of a traditional propeller.

This diagram (which you'll also see in the Spectrum article) shows how the glider uses a variable buoyancy system and actuated battery packs to glide through the ocean

Back in February I talked about a thermally-driven prototype that Webb Research has developed, but the big news right now is how traditional gliders are being snapped up by defense contracting companies iRobot, Teledyne, and of course, Bluefin. (disclaimer for those who didn't read the About section: I work for Bluefin, but even if I didn't I'd think glider technology in general is just, as they say, the bee's knees) The major impetus for acquiring this IP lately is a big Navy contract that's due to be awarded soon. Erico's article describes the differences between the different glider designs and some of the challenges in meeting the next generation of evaluation and procurement system requirements:

Up to this point, however, AUVs have been a cottage industry. â''The challenge now is making the transition from this very hands-on build-and-test kind of manufacturing to commercial production mode,â'' says Tom Curtin.

And even though their original designs are alike, their makers boast of unique capabilities. iRobot claims the Seaglider has the longest range and battery lifetime, being the first glider to complete a mission of more than 3750 kilometers and lasting six months. Bluefin, which is also supplying an offshore oil and gas contractor, says the Spray glider can go about 50 percent deeperâ''1500 metersâ''than its competitors and has more durable sensors. And Teledyne says the Slocum's rudder gives it better shallow-water maneuverability and that a new system that harvests energy from temperature variations in the water could allow its gliders to stay at sea for years at a time.

Though most of my time in the working world has been spent on powerful propeller-driven AUVs, I can't help but admire the simplicity and elegance of the glider design. It has a very different history from traditional AUVs; AUV technology spun off of towed array systems for surveying -- a short-duration, high-power system -- whereas the gliders spun off of ocean monitoring float technology -- long-duration, low-power. With these very different pedigrees, it's interesting to watch them converge now as supplementary parts of the Navy's existing Unmanned Underwater Vehicle programs.