Automaton

Worcester Polytechnic Institute in Worcester, Mass is a small science and engineering university more in the company of flexible, innovative institutions like Harvey Mudd College and Olin College than the tech behemoths MIT or CalTech. Focusing on interdisciplinary and practical education, last fall they kicked off the country's first undergraduate robotics engineering program. I recently had the pleasure of speaking with Professor Michael Gennert, the current director of the Robotics Engineering Department. Prof. Gennert was kind enough to answer many of the questions I've had about the major since I learned about it last year. Read on for the interview -- and high school students, take note! An application to WPI may be in your future!

How is this curriculum different from the more traditional engineering majors like mechanical, electrical, or aero/astro? What specifically will make robotics engineers better suited to their careers than with a different background?

Robotics Engineering is similar to Aero/Astro in that it covers a wide range of topics and is an integrative, systems discipline. It is no coincidence that our core curriculum consists of Introduction to Robotics followed by Unified Robotics 1-4, similar in philosophy to the MIT Aero and Astro Unified Engineering 1-4 sequence. I want us to graduate engineers who are comfortable machining parts, wiring a board, or writing code, as well as conducting rigorous modeling, analysis, and design.

Each of our robotics courses includes elements of CS, ECE, and ME. The courses are team-taught by faculty from these departments. Consequently, students see that there are no boundaries in engineering. We don't just talk about it; we practice it. What matters is what do you need to solve the problem. It is never "not my job". In this sense, Robotics Engineering is good old-fashioned engineering, not overly compartmentalized like so many fields are today.

Let me add that we did manage to include a one course requirement in Entrepreneurship. Although it is a small part of the program, we believe that it will be immensely valuable to graduates' understanding of the business contexts within which they will have to operate. Interestingly, although this was the hardest part of the curriculum to win approval from the faculty, industry is very excited about it.

Why was the major introduced? What led to it? Was it feedback from industry, from academia, from prospective students?

We saw that robotics was going to become an area of increasing importance. The rapidly decreasing cost of robotics components - computers, memory, communications, sensors - creates a lot of "push" on the supply side. Couple that with the "pull" on the demand side - increasing national needs for manufacturing, defense, health, consumer products, elder care, and entertainment - and we get an industry that is on the verge of exploding. But no one field provides a broad enough background for robotics. One has to know a bit of everything and be willing and able to learn whatever you need that you don't know yet. Clearly the intellectual basis for robotics lies in Computer Science, Electrical and Computer Engineering, and Mechanical Engineering. We are very fortunate to have extremely solid programs in all three at WPI, in fact, one might call them our flagship departments. WPI has a tradition of close interaction among departments and minimal barriers to collaboration. So members of each department just got together and figured out what a robotics program would look like. We quickly realized that robotics - we ended up calling our program Robotics Engineering (RBE) - was a perfect fit for WPI and our project-based curriculum. Also, the opportunity to be leaders with the nation's first robotics undergraduate degree was pretty attractive.

The path to Robotics Engineering was paved by the launch two years earlier of our Interactive Media and Game Development program. IMGD combines Computer Science and Humanities & Arts, two very different departments. So we had already figured out how to assemble multi-disciplinary majors. IMGD went from conceptualization to final approval in about 14 months; Robotics Engineering, a more difficult major to assemble, took only 10.

The response from industry, academia, and students has been overwhelming. Industry craves excellent talent and is very eager to get in on the ground floor of this program. Everyone I asked to join the RBE Advisory Board has accepted. My colleagues at other universities sometimes wonder how we did it and how we did it so fast. The simple answer is that it took a lot of hard work and constant communication - and still does! - aided by the fact that most WPI faculty members actually know and like each other. Many alumni wish that we had a robotics major when they attended. Student response has been great. Our business plan for the program identified 20-25 students each year as the break-even point. We counted 80 RBE majors last year, our very first year and enrollment so far this year is way ahead of last year's.

Carnegie Mellon University of course was a pioneer in offering a robotics engineering graduate degree. Did they influence the way the WPI major was developed? Were you influenced by any other academic programs, organizations, or corporations in framing the coursework requirements?

We looked at the CMU graduate curriculum, Georgia Tech's Threads model, and others. CMU has a very impressive graduate robotics program, but its goals are different from ours. We are aiming to give the best undergraduate engineering education possible. Schools that have robotics minors, concentrations, and similar programs will not be able to give students as rich an experience in robotics as we will, although there are many fine engineering programs that include robotics.

We were influenced by accreditation and designed the program to be accreditable. The national engineering accreditation body, ABET, has criteria for accrediting engineering programs, including specific criteria for each engineering field. Of course, they do not have criteria (yet!) for Robotics Engineering, so we have tried to make the program accreditable under General Engineering. We can go up for accreditation after we've graduated our first student; with four senior RBE majors (they changed majors last year) that should happen this Spring.

Speaking of graduate degrees, I'm looking at grad school some time soon (as are, I'm sure, many of us who were too late for a robotics undergrad degree and are now intensely jealous). Any plans for WPI to offer a graduate program in robotics?

We are planning M.S. and Ph.D. programs right now. We plan to finalize the M.S. program this fall. Although we have only one year of experience with the robotics B.S. degree, we would like the cohort of RBE sophomores to be able to stay for our 5-year B.S. / M.S program. We are looking at several tracks for the M.S. degree: a thesis-based track that is especially appropriate for continuing on for the Ph.D., and non-thesis tracks for B.S. / M.S. students and working professionals. But the non-thesis tracks will still have significant design experiences - Robotics Engineering is definitely not a purely theoretical program at WPI, no matter what level degree. One nice thing about the M.S. is that we can fill in the subjects that did not fit into the 4-year degree, while adding Systems Engineering and other advanced topics. After that, the Ph.D. curriculum almost writes itself.

Looking at entry-level openings in industry, they're still hiring EEs, MEs, and software engineers, not robotics engineers. Do you anticipate any challenges for these students meeting industry expectations while not having the subject-specific depth of their peers? Will industry have to change their expectations to accommodate these students?

We certainly will have to educate employers about this major. Companies such as iRobot, Foster-Miller, BAE, Brooks Automation, and others, are very supportive from the top down and are likely to target robotics majors. When we see that a company is advertising for CS, ECE, and ME majors on campus, we can ask if they want to include RBE majors. RBE majors might not have quite as much depth as others, but they should make up for that with their breadth of knowledge, Where else can you find a fresh graduate who can talk intelligently about Object-Oriented Design, Fourier Transforms, Nyquist criteria, Reynolds numbers and Young's modulus?

While our graduates will be well-suited for the growing robotics industry, they will also be highly qualified to work in fields that are not traditionally thought of as robotic. For example, the principles and technologies involved in detecting incoming projectiles, assessing their threat levels, and deploying countermeasures follow the same sense-compute-actuate paradigm as robotics.

What high school preparation do you recommend for students interested in robotics engineering that may be different from other engineering tracks?

Honestly, it is the same as any other engineering discipline. The most important thing is a passion for the subject and that is perhaps the one thing we cannot teach. Other than that, I look for students who have strong math skills, can think analytically and creatively, and are highly literate and communicate well. Come here with that, and we'll teach the rest. Of course, being well-read and knowing a lot of math and science helps, too.

Thanks, Prof. Gennert! And be sure to check out WPI's Robotics Engineering homepage for more information.

I've had my eye on Aldebaran Robotics, a company based in Paris, France, for a while (and not just because of my love for pain au chocolat). In addition to raising € 5M in venture capital earlier this year, Aldebaran has spent the last three years developing an affordable autonomous humanoid robot called Nao.

Nao speaks, emotes, and moves. It runs on Linux and it's powered by a rechargeable battery -- nothing particularly revolutionary there. It sounds like the real novelty will be in the software interaction and mobility (it has 25 degrees of freedom). Initially it's pitched as a research and development platform for the next generation of applications, but eventually they want Nao to be a household robot to assist with tasks. (At only 23 inches tall, though I suspect these capabilities will be relatively limited.)

From Aldebaran's website:

"Eventually, with many improved behaviors, it will become an autonomous family companion. Finally, with more sophisticated functions, it will adopt a new role, assisting with daily tasks (monitoring, etc.)

Featured with an intuitive programming interface, the entire family will be able to enjoy the robot experience. Yet, full of new technologies, our robot will also satisfy the demanding techno-addict's expectations."

What interests me the most is the idea that the plan is for Nao to always be user-programmable, even ultimately by the average consumer end-user. Given the problems most people I know have resetting VCR clocks and setting automatic functions on air conditioners, I'm a little skeptical that consumers will want something they have to "train" with any sort of programming interface, rather than just buying a specialized end-product with the necessary behaviors already built in. However, I can't argue with how beneficial it could be to have even a basic understanding of programming in a graphical "block" environment be so commonplace. As with most recent technology, though, I anticipate it'll be the kids doing this while the parents look on in wonder.

This puts Aldebaran in competition with Willow Garage's PR-2 robot; they're also designing it as a university development platform to encourage research into next gen applications. PR-2 doesn't seem to be nearly as close to production as Nao, but with greater size and dexterity it may have more potential for a broad range of household tasks.

Aldebaran is planning general public release at the end of 2008. The press release on their website doesn't indicate pricing, but this blog suggests it'll go for € 10k initially with a target price of € 4k when they can ramp up production.

Via Slashdot

Researchers at the Technical University Delft will present the smallest flying flapping robot carrying a camera today. With a wing span of just 10 centimeters and weighing in at a mere 3.07 grams, the remote controlled dragonfly is half the size of Borneo's Tetracanthagyna plagiata dragon flies.

The video below shows the DelFly Micro's first test flight indoors. The smaller picture in the bottom right corner shows video data transmitted from the robot in-flight to a ground station. Using image recognition software developed by the DelFly team, objects can then be recognized automatically. This may allow the robot be operated from - or by - a computer.

Guido de Croon, developer of the vision-based control system of the DelFly, already has applications in mind. When a university building burnt-out recently he thought of his team's MAVs: "Since there was some risk of collapse, people could not enter it, and we proposed to attempt to fly into it with the DelFly or a quad-rotor." With the new DelFly Micro still in development, a quad-rotor was deployed to survey the site, but was found too large to enter the building. "Unfortunately, we did not succeed in getting in," De Croon explains, "However, we did gain some experience in what problems one can encounter in such a situation."

Thanks Guido!

Photo: Takanishi Lab

Toyota's Partner plays the violin and trumpet, and Honda's Asimo has even conducted the Detroit Symphony.

Atsuo Takanishi wants to build an entire humanoid robot orchestra. Takanishi, a professor of mechanical engineering at Waseda University, in Tokyo, started with a flutist.

From John Boyd's "This Robot Toots Its Own Flute" story in Spectrum:

Getting the robot to produce a melody turned out to be a monumental task. First, the researchers worked with professional players to create a performance index of what constitutes the best flute sounds. They translated these sounds into mathematical formulations, to which the robot refers. The researchers then programmed the robotâ''s organs to create a sound. Once a sound was produced, they used the parameters controlling the organs that produced the sound as a base and then adjusted those parameters repeatedly until the sound improved and eventually approximated a target sound in the performance index.

â''We had to teach it everything,â'' says Takanishi. â''The different positions of the lips and fingers, the strength of the air pressure, everything. There are any number of parameters [making it] almost impossible to engineer.â'¿ It was a very slow process.â''

The result? See for yourself -- here's a video of the bot in action: http://spectrum.ieee.org/ns/video/flutebot.mp4

Another creation of robot maker and animator Crabfu. I'm a big fan of this guy! His robots are not only simple and beautiful -- they have a lot of "personality" as well.

Description from his site:

PutterBot uses 2 standard size servos for the tracks, each with the potentiometer popped out, and servo taped to the top of the servo. This is an easy way to make the servos continuous, and you can adjust the trim with the potentiometer or with the radio. The tank chassis is a kit from Tamiya, and directly driven from the servo horns. The tread mesh is not perfect, but it works ok. One micro servo is used for the putter, and another one used for the head, and mounted in the back of the Putter Bot, driven through wire linkage. The reason using a linkage system, instead of directly mounting the head/light to the servo, is to lower the head for a better center of gravity... and it just looks cuter with the head down low. Radio mixing is used, so that the right stick controls the tank movements, and the left controls the putter and head.

Students from Singapore's Nanyang Polytechnic university have won this year's Embedded Development invitational challenge of Microsoft's Imagine Cup 2008. Their solar powered robot, dubbed EMS (Environmental Monitoring System), processes air quality on the fly while autonomously navigating indoor and outdoor environments. It then transmits the air's Pollutant Standards Index (PSI), temperature and humidity wirelessly to a remote user.

This year's finals in Paris saw 370 finalists chosen from a pool of more than 200,000 students from over 100 countries and regions competing in nine categories centered around the motto: "Imagine a world where technology enables a sustainable environment." The yearly event organized by Microsoft and endorsed by the United Nations is one of the largest student technology competitions.

With this year's winners announced, registration has opened for next year's Imagine Cup to be held in Egypt.

The IEEE Women in Engineering group has just put out a new issue of their magazine that does a fantastic job of showing many aspects of the robotics industry and the women who have contributed to it. Fortunately, even for those who don't have a paper subscription, there's an online version here. You can download a PDF for easier reading as well.

It profiles people like Ellen Purdy (a part of the Army's Future Combat Systems), Helen Greiner (cofounder and chairman of iRobot), and a DARPA Grand Challenge team member, as well as fun stuff like Disney Imagineering and the intersection of robotics, engineering, and music (who knew Guitar Hero was so important to the industry?).

I'm pretty tempted to get paper copies of this.

UPDATE: There's no better way to understand Kiva's systems than seeing it in action. Here's a video Josh Romero and I prepared:

There's been a lot of press about Kiva Systems, the Boston-area startup that developed mobile robots to automate pick-and-pack warehouse operations. No article, however, has really explained the technology that lets the Kiva robots do what they do -- swarm a warehouse by the hundreds and in a highly coordinated bot ballet deliver inventory to workers, racks of products arriving one after another in seconds, flawlessly.

Photo: Joel Eden Photography/Kiva Systems

In other words, there have been few or no details about the robots' control system, their mechanical design, and the overall resource-allocation algorithms. Until now. Spectrum has filled this gap with an in-depth article ("Three Engineers, Hundreds of Robots, One Warehouse") by yours truly in the July issue.

The two things that most impressed me about Kiva's technology were the distributed control and the robots' mechanical design.

The robots don't just follow orders from a central, know-it-all computer. They have software agents that interact with agents on a warehouse-management server and on PCs at pick-and-pack stations. All the agents act independently, each trying to optimize its own tasks using heuristic methods like greedy algorithms. What's even more interesting is that the robots, which navigate by reading barcode stickers on the floor, detect how far off their bodies are from the center of the stickers and report these readings to the warehouse server. By sharing this information -- by using a "wisdom of the crowd" approach (the proper term is distributed estimation) -- the robots can improve their navigation capabilities.

As for the bots' mechanical design, they have a lifting mechanism capable of jacking up half a ton of stuff that's one of the most beautiful pieces of machinery I've seen recently. It's a custom-machined hard-anodized aluminum ball screw powered by a single dc motor (Kiva was very forthcoming with technical details, but sorry, they didn't allow us to include photos of this piece). The really cool thing is not the screw itself, but what Kiva did to prevent the inventory rack from rotating while the screw turns: the robot's control system makes its wheels rotate in the opposite direction at the exact speed required to keep the rack motionless. Neat!

The article also has details on Kiva's business side and its three founders and early (and cold) startup days in Boston. The whole story is online, or if you're an IEEE member you can download the PDF at Xplore.

LEGO today announced the WeDo kit, which is meant to be not just a younger kid version of the Mindstorms but is designed to appeal to "emerging markets" -- developing countries -- and even interfaces with the OLPC XO computer to do so, which I think is pretty sweet.

Like the Mindstorms kit it comes with a kit of LEGO bricks and sensors and has a similar building-block software interfaced based on LabView, though it's simpler and aimed for a younger audience. The press release describes something of the curriculum they have in mind for it:

Working in teams, children invent their own solution by building a LEGO model and programming it to perform a certain task. Cause and effect learning is enhanced by the models remaining tethered to a computer; similar to scientists in working labs, children can test and adjust their programming in real time. After reflecting on what did and did not work, students can consult with peers, adapt programming, adjust models or begin again.

Unfortunately it's not up on the LEGO website up, and there's no hint on how much it'll cost. Only info at this point is that it will be available in January 2009.

Thanks, Trisha!

I saw "WALL-E" last night with some folks from work and it was universally agreed to be an excellent movie. The characters are fantastic, the animation is as always impressive, and the story is fun. Fans of Apple products will find a few winks to Macs and iPods, and I'm pretty sure EVE is exactly what an Apple robot would look like. For his part, WALL-E bears a striking resemblance to Johnny 5.

I can't wait till I can have my own. For now I'll have to do with the LEGO version, I guess.

Go see it. Cute WALL-E wants you to.

Image from Disney/Pixar