POSITION & HOLD

Robotics and the future

As AI machines take on traditional human roles, they will change job assignments everywhere—including aviation and space.




RIBA will carry patients too weak to stand or walk on their own. Testing begins in 2011.

The world's second best known robot may be Hubo, a life-size walking humanoid inspired by Asimo and built by Junho Oh at the Korea Advanced Institute of Science & Technology in 2003. But it is Hubo's spinoffs that may be most interesting.

Albert Hubo, developed with Hanson Robotic in Dallas TX, carries the face of Albert Einstein. It can speak, gesture with its 5 individual fingers (per hand), and produce a wide range of natural facial expressions, often in appropriate contexts.

Mini-Hubo is a miniature version of the original, designed to cost less than $8000 for would-be researchers who cannot afford its big brother. It is intended to serve as an open-source hardware platform—a common vehicle that software developers can use for their research. Plans for Mini-Hubo can be downloaded freely from the Internet.

There is even a Virtual Hubo. It is a robot emulator program for experiments who are really short of funds. Write software to operate the Virtual Hubo, and it will work the same way on both Mini-Hubo and the human-sized original. Researchers say that having a common platform for work on software should make it easier to share their work, speeding development of future robots.

CHARLI-L shows the US's strengths and weaknesses in this field. Developed by Dennis Hong and colleagues at Virginia Tech, it cost just $20,000, some donated hardware, and the work of a dozen graduate and undergrad students over 18 months. Contrast that with the $300 million Honda is said to have put into the first Asimo! The National Science Foundation puts just $10 million per year into civilian robotics, compared with $80 million a year in Korea.

Despite its economic handicap, CHARLI-L (which stands for Cognitive Humanoid Autonomous Robot with Learning Intelligence–Lightweight) is capable of walking, at least on a good day. There are some bugs to be worked out before it will be capable of competing with Asimo or Hubo.

And a group of independent researchers calling themselves Willow Garage, in Menlo Park CA, are producing the PR2—an autonomous robot with 2 arms on a mobile base. It is not particularly human looking, but it is extremely versatile. For example, experimenters at the University of California at Berkeley recently taught a PR2 to fold laundry—or at least the towels.

For researchers, the PR2 is a significant advance, because it provides a standard platform that can be used for software development. Willow Garage has released all the software for PR2 under the BSD license, which allows anyone to commercialize it or even run it on other robots. They have even provided 11 of the robots to research teams at Stanford, UC Berkeley, the University of Southern California and other universities.

Their most recent work involves complicated object recognition, hand/eye coordination and a host of other brainy functions. When one of the Willow Garage personnel picks from a menu, the robot rolls to the refrigerator, opens the door, picks out a beer—or announces that they are out of the chosen brand—and slips it into a foam holder. Then it takes it to the thirsty user's office and hands it over. This is intelligence most robot researchers, and the rest of us, can applaud.

Back to the future

Robosoft's Kompaï can take orders by voice or touch screen to do chores like placing calls or sending e-mail.

These are just a few of the developments now spreading robots throughout industry and, eventually, our daily lives. Some experiments are replacing the cogs, wires and hydraulics that operate robot limbs with polymer gels that contract when voltage is applied to them—this should make future robots cheaper, lighter and more reliable.

Others are working on "swarms," in which groups of small robots communicate to share their intelligence and coordinate their operations, in effect behaving almost like a single "organism" distributed over many bodies. Many more are improving machine perception, situational awareness and cognition.

"The nanotech under development is enabling extremely capable robotic actuators," observes Dennis Bushnell, chief scientist at the NASA Langley Research Center. "IBM's Blue Brain project [predicts] that machine intelligence via biomimetics is currently less than 15 years out, with extremely capable systems in the runup."

Robotic Technology Pres Robert Finkelstein concurs. "The consensus is that robots will have human-level cognition by 2025 or 2030," he says.

If robotics got its start in the cockpit, it is not yet clear how future robots will aid corporate aviation. We do not, for example, foresee Angelina Hubo, carrying the face of la Jolie rather than Einstein, replacing human attendants soon on the company G650. More likely are robotic baggage handlers, refueling systems and tugs, none of them even remotely humanoid.

Carnegie-Mellon's HERB can fetch an object or steady a chair, but practical use is a long way off.

But the most important development in aircraft automation still lies a few years into the future—for which many pilots may be grateful. This is FAA's Next Generation Air Transportation System, due to come online around 2025. Operating under NextGen, aircraft will be routed automatically from departure gate to arrival gate, and it seems that any sufficiently automated airplane may inevitably become a UAV. New York City's subways have operated almost flawlessly for decades without human engineers at the controls. Under NextGen, aircraft could eventually follow their lead.

"What I am seeing is a gradual change where pilots are being 'assisted' more and more," says Universal Avionics COO Paul DeHerrera. "The next step, instead of just warning the pilot in certain situations, would be to act. Systems already exist to autofeather a propeller blade should an engine fail, so why not put the gear down or try a reset of a circuit breaker instead of just warning the pilot? When does it become robotics vs just assistance? One day, we may wake up and we are just going for a ride."

An alternative view comes from Sergei Sikorsky, lifelong slingwing consultant and observer of fixed-wing aviation. "Robotics will never replace a human in a civil flightdeck," he declares. "Judging by the growing threats in cyberwarfare, it's only a question of time before someone invents a 'death ray' to cook robotic circuits at a distance of 5–6 miles. It's harder to cook a properly protected pilot's brain."

Either way, this generation of corporate pilots will spend full, rewarding careers in the left seat before we face this ultimate implication of aircraft automation. In the final part of this article we will examine remotely operated airliners and other possible applications of aviation robotics.

Marvin Cetron is a forecaster/futurist and president of Forecasting Intl. His study for the Pentagon, Terror 2000, written in 1994, contained numerous predictions of the subsequent course of terrorism.


3


1 | 2| 3