Thursday, January 21, 2010

Financing the Strawberry Project

Robotic projects get funded for strange reasons. Many are at the whim (in scientific circles they call it "inspiration") of researchers attempting to conquer an interesting scientific challenge. Others are funded to solve specific as-yet-undiscovered methods of doing something that needs to be done. The space and defense programs (NASA and DARPA) are perfect examples of the latter and are referred to as "strategic funding".

In Europe, strategic funding happens when a need (or series of needs) is identified that has a potential robotic solution. A consortium is formed combining industry with educational research facilities.  A roadmap is developed. Then the plan is funded by the government.  Upon completion all the members of the consortium are privy to the resulting solutions. Similar strategic public-private partnerships exist in Korea and Japan. As these consortiums succeed, more are added to the pipeline.  But not in the U.S.

There isn't a corresponding public-private process in most areas of American scientific discovery although the NSF and NIST have awards programs for targeted research in areas such as health care, energy, bio and nano technologies, and communications. Also, last month Pres. Obama outlined his forthcoming manufacturing initiative which may include some funding for robotics, but this initiative is an exception. All of these fundings are small in comparison to the need, thus the American funding solution tends to be through venture capital and charismatic entrepreneurs getting loans from family, friends and the SBA. This combo has worked in many, many areas of science. But robotics these days is multi-disciplined. Every robot is a collection of computers, engineering, electrical, mechanical, psychological, motion and vision algorithms, and software. They are a marvel, processing formulae and moving precision parts to do wondrous things, sometimes autonomously and other times interacting with computer networks, humans or other robots.  The complexity and multi-disciplinary aspects of robotics cry out for project and funding management.

There are core metrics involved in worthwhile robotic projects. Metrics that ask necessary questions like whether there is a real business need; whether costs (including amortization of initial research costs) equals what would willingly be paid to satisfy that need; whether the solution proposed is one that correctly and fully solves the need better than other solutions; whether the research can be translated into a long-term commercially feasible solution; and defining the main obstacles and how can they be solved.

The latter (what are the main obstacles and how can they be solved?) is really interesting to a scientist; the former isn't. But the former is necessary to commercialize and pay for robotic science development. Which brings up CMU's strawberry project.

Farming co-ops are commonplace and they are generally different from industry associations. They often have a service function like providing harvesting and distribution assistance or shared production resources. It is not uncommon for a co-op to fund research to solve one of their problems. Thus it came about that five California strawberry nurseries (representing 85% of the California strawberry nursery market) grouped together to fund research to solve their immediate problem: the tedious, costly and labor-sensitive sorting of millions of plants each year. They turned to the National Robotics Engineering Center (NREC) at Carnegie Mellon's Robotics Institute.

To maintain good strawberry yields, growers must replace their plants every year using manual labor to sort several hundred million nursery-grown plants into "good" and "bad" categories. They needed a precise, mechanized solution combining many sciences to accomplish their goal - and they were willing to fund that effort.

CMU's solution is a plant sorter that uses computer vision and software to classify strawberry plants into groups beyond the previous good and bad. Groups of sizes, varieties and stages of growth enable new efficiencies for co-op members helping them improve quality, streamline production and deliver better plants to growers (which, in turn, produce better strawberries for consumers).
During a 10-day field test in October, 2009 NREC engineers tested the strawberry plant sorter under realistic conditions, where rain and frost change plants’ appearance and roots may contain mud and debris. The prototype system had to sort plants of different varieties and levels of maturity. While in the field, it sorted over 75,000 strawberry plants. On average, it sorted 5,000 plants per hour, several times faster than human sorting. NREC hopes to achieve sorting rates of 20,000-30,000 plants per hour with the final system. While the sorter’s overall error rate was close to that of human workers, it inspected and sorted plants more consistently.

“That’s the beauty of it,” said one grower. “Hand sorting varies more and has more drift in quality.”
There are three more phases to complete the strawberry project which include better methods to separate harvested strawberry plants, improve the equipment's robustness and ease of use, and integrate it into the nurseries' harvesting and packaging processes.

To me, this project is unique in that there are few American robotic projects initiated by corporations or business groups to solve a problem with robotics.  UPS, GM and Ford, Deere, Boeing and Lockheed - have funded university research to solve their problems in the past.  MIT has a small program enabling directed research.  And NIST and the SBA fund selected projects.  But NREC is to be congratulated for making Carnegie Mellon's Robotics Institute rich set of core capabilities available to industry to solve their strategic problems.

Wednesday, January 13, 2010

2010 Robotics Predictions; Reality Check for 2009

Trick math question: If you have $100 and lose 50% and then gain back 50% of that, how much do you have?
  1. $100
  2. $50
  3. $75
  4. None of the above.
Stock markets around the world rose dramatically in 2009. Robotic stocks did as well or better than the major tracking indexes, particularly service robotic stocks. But the real story isn't the gains of 2009 but the lack of recovery from 2007.

An example: U.S. publicly-traded industrial robotic companies saw their stocks rise 40% in 2009. But those same stocks lost 53% in 2008. Thus their year-to-date rise in 2009 of 40% really only recovered 16 points of the 53 lost the year before. U.S. industrial robotic stocks are still down 37% from their close at the end of 2007 as are almost all robotic stocks worldwide. That is what these Robo-Stox™ charts - one for industrial robotic companies and another for service companies - attempt to show. Click to enlarge.

Most countries' robotics stocks didn't fare as well as the American NASDAQ Index with the exception of Canada, India, Israel, Taiwan and a very few individual stocks. Thus although 2009 was a significant up year for stocks, and robotic stocks in particular, robotic stocks have yet to recover their highs of 2007 and have a long way to go to do so.

The seriousness of the recent worldwide stock market and economy crash - of the drop in market value of the companies - of the loss of jobs and orders, and revenue and profits - is a long way from recovery. Although jobs in the robotics sector are available for qualified takers, particularly in the service sector, unemployment in general is dramatically high and most economists are predicting that it will be well into 2012 before any real gains occur.

That is not to say that all is pessimistic, particularly for robotic businesses. 2010 looks to be a good year with definite "drivers" effecting selected marketplaces.

Worldwide military, police and security agencies are continuing to purchase and invest in R&D for all types of unmanned, remote-operated aerial, underwater and ground robotic devices. More jobs - with the likelihood of continued growth over the next few years. There will be some budget cutbacks. EG; the US Army just announced cancellation of an autonomous ground vehicle (Lockheed Martin's MULE) and a pilotless helicopter (Northrop Grumman's FireScout) - both are being replaced by upgrading existing equipment to provide the new functionalities.

Medical robotics (included in the services sector) are poised for many years of rapid growth propelled by:
  1. Growing patient demand for non-invasive surgery,
  2. The current effort to reduce hospital costs by increasing productivity through a variety of robotic activities (non-invasive surgery, pill dispensing, materials transfer, lab assistance, etc.),
  3. Hospitals, which have held back capital purchases (such as Intuitive Surgical's million dollar da Vinci devices) for the past two years, are beginning to reinvest in these types of equipment.

With the return of small amounts of discretionary income back into the economy, consumers are once again interested in robotic toys and kits as can be seen by 2009's Christmas rush to buy millions of robotic hamsters (Zhu Zhu) and thousands robotic penguins. And the hit of CES was an indoor-flying iPhone controlled quad copter by Parrot that will sell for $129.

For industrial robot manufacturers, orders will stay down for quite a while (photo: used robots for sale from closed Chrysler plant). For those vendors that have switched or are making inroads into the services sector, the horrendous spate of bankruptcies and buy-outs has stopped and the future is looking brighter especially in new markets including the SME market.

Small and Medium-sized Enterprises (SMEs) are now being offered affordable robotic products that haven't been available to them before, first in Europe and Asia, and later in the U.S. Lightweight and easily trainable, these flexible robots are enabling these smaller manufacturers to increase productivity and not have to go off-shore to produce their products.