The future of intelligent robots
Special to CNET News.com
February 14, 2004, 6:00 AM PST
With the rovers Spirit and Opportunity sending messages from Mars and rekindling discussion about the future of U.S. space discovery programs, what is the state of the robotics industry in general?While we tend to imagine robots in terms of science fiction adventures and movies (think R2D2 and the California governor), or as legions of automatons replacing human workers in the job force, the truth may be more mundane and less sinister. Vijay Kumar, University of Pennsylvania mechanical engineering and applied mechanics professor and director of Penn's General Robotics, Automation, Sensing and Perception (GRASP) Laboratory, recently discussed the future of robotics with students in the Executive Master's in Technology Management program, offered by Penn Engineering and co-sponsored by Wharton.
According to Kumar, an old Webster's definition turns out to be a pretty good one for a robot: an automatic apparatus or device that performs functions ordinarily ascribed to humans or operates with what appears to be almost human intelligence.
The rise of robotics in the manufacturing industry--and increasingly, the service industry--continues to create the greatest value where it can perform tasks that might typically be done by humans or that require something close to human intelligence to complete.
The robotics industry posted strong numbers in 2003, with the Robotic Industries Association (RIA) reporting North American orders up 28 percent in units and 15 percent in dollars, following a two-year fall-off from the peak years of 1999 and 2000. Overall, the growth of the $10 billion worldwide ($3 billion U.S.) robotics industry can be tied directly to the fall in the price/performance ratio of microchips since the early 1980s. The number of microcontrollers used in industrial applications rose from 185 million to 778 million units between 1993 and 1999. This reflected a parallel rise in sophistication, from motor and energy control to emissions sensors and factory automation to highly sophisticated security systems and medical applications.
Japan stands out as the clear leader in the use of industrial robots, with the United States and Germany closely tied for a distant second place. According to statistics reported by the International Federation of Robotics and the United Nations Economic Commission for Europe, the stock of multipurpose industrial robots in 2003 was projected to reach 344,000 in Japan, 111,300 in Germany and 111,100 in the United States. (A recent RIA report estimates that the U.S. number is 132,000 robots, second to Japan.) Other major users include Italy, the Republic of Korea and France. Among the world's largest robot manufacturers are Fanuc, headquartered in Japan, ABB of Europe and Adept Technology in the United States.
The rise has not been limited to industrial uses, however. In fact, the typical U.S. household today has more than 225 microcontrollers, handling everything from thermostat control and entertainment centers to smart appliances and carbon monoxide detectors. In 1980, that number was in the single digits, and garage door openers were in the high-tech gadget category.
Manufacturing has dominated the use of robotics so far, the automotive industry being the major player. Robotics here are used for tasks such as materials handling, arc welding and coating (paints and finishes).
Within manufacturing, applications range from advanced mechanized systems to the more flexible automation and robotics systems we tend to associate with our image of robots. For example, says Kumar, the U.S. Postal Service processes about 165 billion pieces of mail per year. "Fixed automation"--custom-engineered, special-purpose equipment designed to automate a fixed sequence of operations--enables distribution centers such as those in Philadelphia and Washington, D.C., to sort 20 million pieces of mail per night with an amazingly low error rate that Kumar says is "stunning"--despite common perceptions to the contrary. High production rate and quality are clear gains, but the limitation is inflexibility of product design. You can't ask mail-sorting equipment to switch gears and start batching cookies (or even unusual sizes and shapes of mail).
"Programmable automation" adds greater flexibility--but still within a highly structured environment. Essentially, the equipment is designed to allow a range of product changes within a specific class. Best uses of programmable automation are with batch production for medium-volume products--manufacturing similar parts with different dimensions, such as different-size bolts, brackets or doorknobs.
Finally, "flexible automation" introduces more highly advanced robotics systems, creating equipment designed to manufacture a variety of products or parts. A painting robot in the auto industry "recognizes" the car coming through the assembly line, knows the unique geometry of that model car and modifies its program so that the paint is applied appropriately.
But we aren't ready to sit back and leave the work to the robots. "Flexible automation and the 'lights-out factory' concept may be incompatible," says Kumar, because "typically, robots are a small part of the puzzle." And when it comes to tasks that involve controlled interaction with the physical environment, automation is even more difficult. Kumar cites the example of finding a light switch in the dark. The non-rigid human arm and hand are far more efficient and potentially less destructive than the powerful but rigid arm of a robot. (It turns out that screwing in a light bulb may be underrated, after all.)
Service industries: The next frontier
The entertainment industry has also entered the fray, with Sony's AIBO dog and Honda's Asimo humanoid robot. Lifelike animal robots are now capturing the attention of visitors in theme parks. Look for expanded uses in movie production as well as retail toys.
For biotechnology and medicine, potential looms in all directions. In drug discovery research, for example, automation allows for the small-scale manipulation and observation of fluids and solutions at a level of precision humans simply cannot match. Robots are also tireless when it comes to the exacting and exhausting work of chemistry experimentation.
Perhaps the most high-profile uses will be seen in military applications, in emergency situations such as fire fighting and search-and-rescue missions, and in exploration in space or under water. In recent military operations in Bosnia, Afghanistan and Iraq, soldiers were able to see around corners and examine tunnels using remote-controlled robots installed with cameras and sensors that let the operators observe what the robot was seeing--but from a much safer distance. Similar types of robotic technology can be used to enter and explore burning buildings, to search for human life after earthquakes, or to venture into other dangerous environments for emergency or exploration purposes.
Indeed, if NASA's "Robonaut" unit had been ready last year, the damaged space shuttle Columbia might have been able to be repaired in flight. Looking like Star Wars droid CP30 on a jointed stick (no need for legs in space), the Robonaut is controlled by a human counterpart, located inside the craft or back on earth, who is able to 'see' what the Robonaut sees and to guide its actions with considerable precision.
Early assumptions were that developments in robotics would proceed along two dimensions--automation (typical industrial uses) and autonomy (the "smart" robot). While there has been substantial progress in industrial automation, autonomy gains have been minimal. As it turns out, maybe we weren't thinking about the right dimensions.
The greatest potential for robotics development most likely lies along a third dimension--augmentation. As seen in the surgical, military and disaster applications, robots can produce dramatic results when they are used to extend or augment human capabilities. And Kumar defines the most promising areas for application as "the 4D environments: dangerous, dirty, dull and difficult." Consider, for example, the use of robot technology to enter burning buildings, take on household chores like vacuuming or perform high-precision biochemical experiments.
So, in the end, will robots replace jobs? "Yes," says Kumar, "but robots replace jobs that are the kind of jobs that nobody wants to do or are good at doing. They are the 4D types. Through robotics, these jobs will keep disappearing, and we should be happy that they are."
And what about Spirit, at this point experiencing difficulties, after an initial four-star landing? For Kumar, it's not a question of if, but when, the problem is fixed, despite the challenges of testing and repairing technology based in outer space. As far as the robotics are concerned, Kumar suggests that one measure of success is how well Spirit can make independent decisions based on its own observations of life on Mars. If NASA tells Spirit to go to a certain location, for example, will it be able to navigate the terrain using its internally programmed systems to make observations, process information and adjust for new evidence, like a sheer drop-off that was hidden from view? Or will it remain dependent on a guiding hand from Earth? As with all good space adventures, stay tuned for the next exciting episode.
All materials copyright © 2004 of the Wharton School of the University of Pennsylvania.
Interested in more research studies like these? If so, sign up for the Knowledge@Wharton Newsletter, a free service of the Wharton School of the University of Pennsylvania.