Mass High Tech
Robots To Give a Big Hand To Humans
By William Townsend
I've seen the future of robotics. It dwarfs today's robotics market a hundredfold by taking robots out of the secretive dark corners of factories and recasting them as human collaborators in every aspect of daily life.
These robots bear no resemblance to the naively human-like robots of science fiction.
They are machines that can react to touch and "learn" on the job, thanks to advances in artificial intelligence. Evidence of this revolution surfaced only in the last few years, with uses in image-guided robotic surgery, home care for the elderly, counter-terrorism devices and manufacturing, where machines now work side-by-side with human partners.
But before considering the future, let's shatter some illusions about the present state of conventional robotics. America's gross domestic product is quickly approaching $10 trillion. How much is invested in robotics? As you think about the answer, recall recent television commercials you've seen with robots lined up on factory floors.
Today's economic expansion has been fueled by leaps in productivity, often symbolized in the media by robots at work. So, you might imagine these machines account for a good portion of the GDP. Right?
In fact, it may shock you to learn that, according to the Robotics Industries Association, the dollar value of robots sold in the United States last year barely topped $1 billion dollars – or only one-hundredth of 1 percent of the GDP. That's so small it is barely worth considering.
So what's going on here? That was my question in 1985, when MIT invited me to lead development of a more effective robot design, code-named the Whole-Arm Manipulation Project. Like most people, I had never before seen a robot first-hand. But according to popular media hype at the time, scientists had already conquered the technical challenges of robotics.
Taking the hype for granted, I imagined legions of robots already helping people do all types of tedious, dirty and dangerous tasks, both inside and outside the factory. But then, why would no-nonsense MIT waste valuable engineering resources reinventing robot technology?
It turned out that the emperor had no clothes. That parable perfectly described the disconnect between the media hype at one extreme and the underwhelming reality of robotics at the other.
In truth, to provide motion to a robot required a doctorate in complex mathematical geometry and computer science, just to prepare the computer code. To execute the code and translate it into motor commands took from minutes to hours of high-speed-computer time.
Meanwhile, the technology for controlling unanticipated motions and interactions - the kind requiring human dexterity and intelligence - did not exist. Even if 1985-era computers could have supported the necessary intelligence, the motors that drove a robot's joints used inappropriate gear and hydraulic transmission technology, which impeded the ability to detect or control forces against objects or people. Protruding linkages and appendages restricted robot mobility and introduced safety hazards.
When investors and factory managers uncovered the truth in 1987, it precipitated a sudden worldwide crash in robot purchases. By 1990, virtually every robot manufacturer in the United States had either failed or, like U.S. Robotics Inc., had stopped making robots and found another line of business.
Reacting to the severe industry contraction, the few surviving robotics companies (almost all Japanese) turned decidedly conservative, blocking badly needed innovation. For example, rather than design intrinsically safer robots (in the same way that automobiles have been made much safer since the 1960s), the robot industry dictated that fences should segregate robots from humans on the shop floor.
We have since learned that while robots are better than people in some aspects (they feel no pain and are fast number crunchers), humans are better than robots in other areas, such as intuitive reasoning and judgment.
Today, the evidence is growing that people and robots have highly complementary skills. Technology spun off from MIT's Whole-Arm Manipulator development project is now creating robots designed to cooperate literally hand-in-hand with people, rather than replace them.
The new robots are powered with advanced power transmissions that use stainless steel cables instead of gears to produce smooth and accurate arm movements. The tranmsissions were invented to match performance demands of the Whole-Arm Manipulation Project. This new class of power drives eliminates backlash, and reduces friction and limb inertia to such a degree that touch is detected intrinsically through precise electrical-current measurements in each of the robot's joints - in a manner that is amazingly similar to proprioceptive human motor control.
In the 1990s, innovative groups at NASA, Northwestern University, UMass Amherst, Harvard, General Motors, Ford, Fanuc Robotics, Z-KAT Medical Inc., Sensable Technologies and Barrett Technology began promoting this new type of machine.
As a sign of the times, I recently visited two Ford manufacturing plants in North America. I expected a cool response as I explained who I was (president of a robotics company, Barrett Technology) to United Auto Workers union members on the assembly lines.
But it turns out, the UAW came to terms with robots long ago and now understands robots better than anyone. What they can't understand is why robots aren't designed to be more useful. At the last two annual Robotic Industries Association annual forums, UAW workers joined the conference on their own initiative - not to protest, but to argue for robots to assist them with the nastier parts of their jobs.
Those machines are in the pipeline and they will be designed to be intrinsically safer than conventional ones.
The new generation of robots can "feel" forces of contact throughout their structure. They can use their inner limbs to push, grasp and manipulate things the way people do when carrying heavy bags of groceries. This gives them broader versatility, because they can handle a wide range of payloads without having to increasing the size of the machine.
And perhaps most importantly, they respond with a feather touch to a person's guiding hand, allowing a craftsman on the assembly line or a nurse in the hospital (rather than the robotics PhD who lacks crucial task experience) to collaborate closely with the next generation of robots.
The robot then adds its unusual capability to operate along virtual surfaces and virtual tracks in space to take care of the tedious details of precision motion - working with its human counterpart to accomplish difficult tasks, and letting both man and machine do what each does best.
William Townsend is president and chief executive of Barrett Technology Inc. in Cambridge. He can be reached via e-mail at email@example.com and at www.barrett.com.
Publish Date: February, 1999
Reprinted with permission. All rights reserved. Mass High Tech 1998.