The inventor of the floppy disk turns his mind to “behavior programming” robots and talks to Wilson da Silva.
DICK MORLEY reckons there’s a right way and a wrong way to automate manufacturing, and most companies do it the wrong way.
The American physicist-turned-venture-capitalist – inventor of the floppy disk and 20 other patented technologies – believes automation has been dominated by “the control freaks”: engineers who try to pre-program every possible contingency into a robot, or who strictly limit an automated system with a ponderous rule book of dos and don’ts.
Today’s manufacturing environment is far too complex. Robots break down and assembly parts fail constantly, halting production.
Yet each individual incident might be so rare, or so novel, that robot reaction cannot be pre-programmed. Human intervention is always required.
In artificial intelligence (AI) circles, this is called “state programming” and is the traditional approach to automation control.
“You don’t have a hope in hell of understanding factories. You really don’t have control. By striving to get control, you only make it worse,” Morley said.
Morley - and a growing number of others in the AI game - says the solution in today’s complex manufacturing process is not top-down programming for every possible eventuality, but “behavior programming”: giving robots rudimentary brains and letting them make their own choices.
“You give them some degree of intelligence; not very much, just enough to react as an ant would,” he told said from Ann Arbor, near Detroit, where he was attending a National Centre of Manufacturing Sciences meeting.
“Ants themselves are simple organisms. They just want to do very simple things. But when they get together, what arises is seemingly very complex behavior.”
Similarly, Morley argues that for complex problems, a simple instruction set is all a robot needs. But when they act in concert with other robots programmed in the same way, they start exhibiting complex behavior.
“Manufacturing is an implosion: many objects come together to make a single object,” he said. “You can give those objects ‘behaviors’, rather than programming explicit instructions.
“General Motors, convinced by Morley’s ideas, programmed its robot spray painters, known as paint booths, with very simple instructions: paint as many trucks as possible. Be selfish: try to paint only the color in your tanks. If another booth breaks down, then change your paint tanks to their color and go back to painting as many trucks as you can with the new color.
“Quite literally, there’s only four lines of software code, so we’ve reduced the code by a factor of 10:100,” Morley said. “What we are doing is empowering the paint booth to make decisions that keep production going. “It ran beautifully.
And this stuff saved money - $1 million a year in paint alone.
“The paradox of this approach - known as “chaos computing” - is that the simplest of programs work best in such complex situations, while simple but information-intensive situations work best with complex programming.
“There are certain classes of problems that seem to respond well to this kind of programming, and that’s complex problems, “ said Morley. “For complex problems, ‘state programming’ doesn’t work well. For simple problems, ‘state programming’ is ideal. Like any tool in the toolbox, they are meant for different tasks,” said the 63-year-old New Hampshire resident.
Another example of how simple rules can guide complex behavior is taxis. “They only have three rules they work by: if the dispatcher calls, bid on it. Go to the shortest line, like at a hotel, not the longest line. And if there’s someone on the side of the road, pick them up.”
Three simple rules are enough to guide complex and self- regulatory behavior, such as reacting to supply and demand: a state programming approach would need thousands of lines of software code dealing with a multitude of eventualities.
“The system automatically adapts. If the line-up is short, taxi cabs swarm that hotel. If it’s long, fewer cabs join the line, and the line is eventually shortened. Taxi cabs act more like a termite mound,” he said.
This is just one example of better technology management that Morley preaches: while the technology may be great, smart management is the key to competitiveness.
While automation is his big thing, Morley also has a sideline as a futurist: he convenes annual Delphi studies with leaders in their fields, and claims 80 per cent success with his forecasts.
Morley’s life one of chaos and kids
If Dick Morley’s chaos computing sounds a little off-the-wall, so is the man himself.
Morley, who is visiting Australia this week for a series of seminars sponsored by BHP, is 63 and loves his Harley Davidson.
His company, R. Morley Inc, is based in a red two-level barnhouse in rural New Hampshire, off the beaten track and with only one dirt road leading in and out.
He owns 11 pairs of skis, which he usually keeps just outside the door when the snows come. And he has 25 foster “kids”, ranging in age from 13 to 33.
“They live in our house anywhere from six months to six or 10 years,” said Morley. “They’re battered wives, or ‘rights of passage’ kids having problems with their parents.”
Why 25 of them? “Because we couldn’t have 50,” he says, chuckling. “Life is a rental deal, you know. You gitta give something back.”
Morley trained as a physicist at the Massachusetts Institute of Technology and worked as a machinist to support himself.
He eventually combined the two interests: working with his hands, and number-crunching, and turned his attention to computing in the 1960s.
At the time, squeezing just 1k of RAM from a room-sized computer was a difficult and expensive process. Industry was keen to use computers in manufacturing, but they were too impractical.
So Morley solved the problem with the invention of task- specific “black box” - the programmable logic controller.
This controlling device is still in use today, the basis of most industrial machinery and a business worth $5.3 billion a year.
With it, Morley launched Modicon, a $600 million-a-year concern. He is also credited with inventing floppy disks, bar-code readers and the thin magnetic strips used in credit cards.
These days, Morley and a few friends act as venture capitalists, funding bright new ideas that occasionally pay off. They call themselves “The Breakfast Club”, and they have been responsible for some 15 successful start-ups, he said.
“It’s just four of us old farts who invest our own money, “ he said. “We look at two or three business plans a week, and invest in about four or five a year. Once in about five years, one of them pays off.”