Automaton

Last Tuesday at the IEEE TePRA conference, former iRoboter and co-founder of Harvest Automation Joe Jones weighed in on the robotics design process and how a startup finds its market. Harvest Automation, as I mentioned a couple months ago, came out of stealth company Q Robotics, formed by a group of the original Roomba developers when they left iRobot. Jones is the CTO.

Harvest Automation's first and currently only product is a small mobile robot that picks up and moves potted plants in nurseries. These nurseries are extremely large -- much like fields in commercial farms -- and the potted plants have to be spaced evenly on a black plastic surface. If you imagine a worker picking up and moving these pots for ten hours on a hot summer day on a black surface, you can see why nursery owners have difficulty finding workers to do this. Enter the robots.

A view of the fields of potted plants at commercial nurseries. Photo courtesy Harvest Automation.

As this cnet article rightly points out, who wakes up one morning and decides, "Why don't I build a robot that moves plants?" The answer, said Jones, is "no one." The design process doesn't start with the robot; it starts with identification of a market. Q Robotics (as they were at the time) started by searching out several untapped markets that still utilized manual labor in undesirable jobs. By going out on numerous on-site visits and researching the size of the markets, they found about 15 different leads before narrowing down their focus to these nurseries.

Next they had to establish the actual task. They knew how big the pots were and how they needed to be spaced; from there, it was relatively simple to design the mechanism that would allow a small mobile robot to find, pick up, move, and set down a potted plant. They also needed to determine what they *wouldn't* be able to do: move the plants from, say, a flatbed truck onto the ground. Humans can continue to do that, or perhaps another automated system; but limiting the tasks was just as important as determining what they were in the first place.

Harvest Automation's prototype robot. Photo courtesy Harvest Automation.

A number of problems still face Harvest Automation's design. For example, Jones mentioned that a single robot would work a full "shift" and that it would move at roughly the same speed as a walking human. While he didn't indicate how long a shift is, if we assume it to be like a human's -- eight hours -- these robots will have some demanding power requirements that will only be fulfilled by heavy and probably expensive batteries. However, these are technical challenges that can be overcome; Jones and his engineers already know they have a market and they know these folks are willing to buy (in fact, he mentioned that some of these nurseries have already put up small amounts of funding to continue the development).

Jones brought this process up again on a panel the next day. When the panel was asked what technologies are limiting practical robotics development, Jones said that the technology is there -- the question at this point is how to put it together the right way for new applications. If wannabe-startups can focus on the applications instead of the technology itself, they could have a lot of success.

In 1924 George Antheil, a "bad boy" of music, composed a piece of music called the Ballet Mecanique which, due to the complex synchronization between player pianos that it required, Antheil never heard performed as composed during his lifetime. Robotic technology has enabled music technologist Paul Lehrman to complete the Ballet and allow it to be performed as originally written.

Lehrman is a professor at Tufts University where he teaches a course called Electronic Musical Instrument Design. It's a combination of music and engineering where students work on projects like musical shoes and MIDI ties. In the 90s, Lehrman was approached to use MIDI technology to automate the synchronized player pianos required to perform the Ballet.

The Ballet was finally performed several times when, in 2006, the National Gallery in DC decided they wanted a full installation that could perform the piece daily. Since comissioning human musicians to play daily would have been prohibitively expensive, Lehrman teamed up with Eric Singer of LEMUR, the League of Electronic Musical Urban Robots. Together they roboticized the entire orchestra for the Ballet and the installation became an extremely popular attraction at the National Gallery. Today at the TePRA Conference Lehrman described how earlier this year, the robotic orchestra provided the musical score to a theatrical production called Frequency Hopping, about Hedy Lemarr and George Antheil. (Note: If you don't know about Hedy Lamarr's contributions to spread spectrum radio technology, read the Wikipedia article I linked there.)

Anyway, below is the video. It is ... noisy. I'm loving the automation, though -- how great to see technology so integral to a modern art installation!

I'm currently at the inaugural IEEE TePRA conference in Woburn, Mass. The TePRA conference allows researchers to present on the technologies they've developed that are enabling low-cost practical robots to be commercialized. The technical program covers vision systems, manipulators, multi-robot applications, navigation and control, and more. Check back for reports on the cool stuff being shown (I just heard something about a robotic musical installation...)

Two sister projects, both funded under EU's Seventh Framework Programme spanning from January 2008 to 2013, are investing heavily in swarm robotics research.

Although not yet realized, swarm robotics has numerous advantages such as:

1) system failure resistance by high redundancy

2) the ability to conquer goals only possible through teamwork

3) mass manufacturability with low per unit costs, etc.

It is the goal of these two projects to research and develop the technology needed to make each advantage a reality.

The Symbrion project has four distinct goals: to create a "platform for exploring artificial evolution and pervasive evolve-ability," develop the required computational on-board resources, support artificial immunology and embryology research, and manufacture a "large number of light modules."

The Replicator project also has four distinct goals: to create an "intelligent, reconfigurable and adaptable 'carrier' of sensors," build a "sensors- and communication-rich platform," ensure high-reliability in an open-ended environment, and manufacture a "medium number of heavy modules."

For a quick summary of the various prototype platforms and current research, check out this video of their past work:

One major challenge the researchers point out is miniaturized power sources. As batteries are scaled down, so is the voltage and charge capacity. Their smallest robot, the I-SWARM bot, draws power from a tiny solar cell, while their larger Jasmine machines use a battery.

Being incredibly small also limits computational capacity. Their current onboard processor has just eight kilobytes of program memory and two kilobytes of RAM, although typical of modern microcontrollers used by most hobbyists.

For more info, see the image gallery, many informative videos, and their publications.

Two new items and one I should have mentioned a long time ago:

• Boston Dynamics, of BigDog fame, looks likely to get a new contract from DARPA to build the next gen of their creepy yet awesome quadruped. Though the proposals aren't due till next year (and take it from a defense contractor, these things tend to be delayed even more), the specs appear to be tailored for the Big Dog design and Boston Dynamics is certainly the best contender. The solicitation, titled "Legged Squad Support System (LS3)", specifies improvements to terrain handling features, increased payload capabilities, and as I'm sure anyone who's seen that video will be pleased to hear, a decrease in sound. Less lawnmower, more stealth! (Thanks, Joe!)

• I mentioned Willow Garage earlier this year when they started getting press for the open-source PR-2 personal robot. At the time, WG had two other concurrent research projects: an autonomous surface vehicle and an autonomous car. Over the summer it appears that those last two projects have been dropped and they're concentrating exclusively on PR-2. Their blog has an interesting piece-by-piece reveal of some of the subsystems as they're being built up.

• The biggest bombshell is that everyone's favorite company iRobot has lost two of its three iconic founders. Rod Brooks, as previously reported, left his position as CTO to start a new company, Heartland Robotics. Just last week, chairman of the board Helen Greiner announced her resignation. CEO Colin Angle will add chairman duties to his plate, and both Brooks and Greiner will remain involved as a technical adviser and director, respectively. There's quite a bit of speculation and buzz online about the motivation for this, but both Brooks and Greiner have emphasized that they remain committed to iRobot's mission and their continued support of the company in their new capacities.

Carnegie Mellon is well-known for its robotics program and contributions to the industry. The Field Robotics Center is celebrating its 25th anniversary this weekend with a special symposium on how the field robotics industry has developed and what the latest and greatest is today. The live webcast of the speakers is here, and hopefully the video will be archived (I know this is late notice; sorry!).

Speakers include some big names like Dave Lavery (NASA), Dana Yoerger (Woods Hole Oceanographic Institute), Hugh Durrant-Whyte (Australia's best-known robotics researcher), Sebastian Thrun (Stanford), and Dave Akin (University of Maryland Space Systems Laboratory).

Via Chief Delphi

The technology for wireless power transmission via induction goes back to a patent by Nikola Tesla in 1900 (US649621). Now Travis Deyle at Georgia Tech and Matt Reynolds from Duke University have created a wirelessly powered robot swarm.

Their work presented at ICRA 2008 earlier this year describes how robots with receiving coils positioned between their wheel bases can draw power from an arena equipped with transmit coils. The picture above shows 5 robots in such an arena, with their yellow and black receiving coils clearly visible (although I can only spot the coils on four of the robots...).

Best of all, the setup looks fairly straight-forward. For more details have a look at their ICRA publication and the video below. I wish the little Alice robots had had such a wireless power option when I did my research on robot swarms ...

Wirelessly Powered Robot Swarm from Travis on Vimeo.

The latest issue of World Robotics is out!

As we've said before, this fantastic report, produced by the good folks at the statistical department of the International Federation of Robotics, is a must-read if you want to understand the global robotics market.

So on to the hot-off-the-press numbers, which refer to the market at the end 2007. First, industrial robots: there are now nearly 1 million units toiling in factories around the world. As for service robots -- which range from military bomb-disposal bots to home vacuum cleaners -- their number grew to about 5.5 million. All of which brings the world's robot population to 6.5 million. (That's bigger than Papua New Guinea but smaller than Tajikistan, if you're wondering.)

According to World Robotics forecasts, 1.2 million industrial robots and more than 17 million service robots will populate the world by 2011. (See chart above, which Automaton put together by adding up totals for industrial and service robots. See also the note at the end of this post.)

Some other highlights from the report:

* In 2007, 114,365 new industrial robots were installed worldwide, a growth of 3 percent over the previous year.

* These new robots are worth some US $6 billion. If you factor in the cost of software, peripherals, and systems engineering, the total triples to $18 billion.

* Asian countries installed 59,300 new robots in 2007. That's far more than any other region in the world, but a decline over the previous year.

* Sales of industrial robots in Europe grew by 15 percent to 34,900 units, the highest

number of robots ever installed in one year.

* In the Americas, sales of industrial robots reached 19,600 units, 9 percent more than in 2006.

* As for forecasts, the World Robotics folks are optimistic about the market for industrial robots over the next few years. Here's what they say:

So we expect a moderate increase worldwide of about 4% over the course of 2008. The dark clouds of a worsening global economic outlook will affect robot installations in 2009 and probably also in 2010. But we do not anticipate a sharp decrease. Why? Industrial robots are a key component in automating processes. Productivity, labour shortages caused by demographic shift, high quality standards, environmental regulations, reduction of tedious and even dangerous jobs, energy and infrastructure costs, inflexible production etc. will pose new challenges to automation technology and will stimulate the demand for robots in all manufacturing sectors. Last but not least, investments in the emerging markets will continue apace. A strong global recovery can be expected by 2011 at the latest.

Finally, some service robots highlights:

* Some 12,000 professional service robots (heavy-duty units, as opposed to home units) were sold up to 2007. Most of them (25 percent) are in defense, rescue, and security applications, followed by field robots (mainly milking robots) with 20 percent, and cleaning robots and underwater systems with 12 percent each.

* As for personal service robots, sales up to 2007 reached 5.4 million units -- about 3.4 million units for domestic applications and about 2.0 million units for entertainment applications. Most of the units for domestic applications are vacuum cleaners (3.3 million units) and more than 110,000 are lawn mowers. Sales of these robots amount to about $1.3 billion!

Note: Keep in mind that the 2007 and 2011 totals shown in the chart are a mixed bag of things, from $9.99 toy robots to multimillion dollar factory manipulators. What's more, the number of industrial robots refers to operational units (it considers obsolescence), whereas the number of service robots refers to units sold (it still counts robots no longer in operation, like that first-generation Roomba you harvested for parts.)

The tech experts on Ars Technica are compiling a series of reports on autonomous cars, called "The Future of Driving." The three-part series talks about the DARPA Grand Challenge and the technologies it has generated, the potential for integration of autonomous cars into industry and society, and the many challenges that researches and developers still face in really getting this technology going.

There's still a lot of work to be done before these vehicles will be mature enough to be let loose on our streets. Real streets contain significantly more obstacles and complexities than the simplified environment found in the 2007 contest, and the most successful vehicles in the Urban Challenge were not designed to cope with pedestrians, bicyclists, traffic lights, or icy roads. And even when the technology is ready for prime time, a variety of economic and political obstacles could delay widespread adoption of self-driving automotive technologies.

Nonetheless, the success of the Challenge series has made it clear that it is only a question of "when," not "if," self-driving automotive technologies will make their way into car showrooms and onto our streets. It's impossible to know exactly when such will arrive, but it's a good bet that a college student entering the workforce today will have a car that can drive him to work before he retires.

Part 1

Part 2

I was tipped off to the recent publishing of a new book, "Robotics: State Of The Art And Future Challenges" thanks to posters on Chief Delphi, the unofficial FIRST Robotics forums.

FIRST favorite Dave Lavery contributed to this review of the robotics industry around the world -- what the cutting edge is within each of the markets and who the leaders are in those fields. Specifically, the book is meant to identify where the US is lagging in robotics R&D. Other contributors include George Bekey from USC, Rob Ambrose from NASA Johnson's robotics lab, Vijay Kumar from the University of Pennyslvania, and other big players in research.

Thought it looks like it's published and priced like a textbook, it appears to be more of an industry guide that a company or research lab may be interested in. In any case, with the contributor list, it's got to be a great resource for anyone interested in a global picture of the latest and greatest in robotics.