Robots Revolution

The arrival of robots at General Motors Corp. in 1961 brought the promise of flexible automation. Today's advances in research offer robots the chance to reach their full industrial potential.

The time: 1923, and the first performance of Karel Capek's play, "R.U.R (Rossum's Universal Robots)", depicts a future where millions of intelligent mechanical workers have taken over the earth. Capek called them "robots" borrowing from the Czech word for work. But reality still lags science fiction. Only now, 41 years after those automatons of science fiction became a factory tool, are robots gaining the intelligence to approach their industrial potential. Today's industrial automatons deliver dramatic value even though most of them are deaf, blind and without the sense of touch. Just adding vision guidance would revolutionize manufacturing, some experts believe. Even so, manufacturers are using robots today in a variety of applications. At auto-systems manufacturer Visteon Corp., Dearborn, Mich., employees get help maneuvering large heavy parts. General Motors Corp. (GM) is employing the devices to locate tools for workers as vehicles come down the assembly line. And TI Automotive Ltd., Warren, Mich., is using robots that can "see" to identify dimensional variations in plastic parts. About 800,000 robots populate global manufacturing with almost half working in Japan. About 121,000 industrial robots work in the U.S., says Donald A. Vincent, executive vice president, Robotic Industries Association, Ann Arbor, Mich. Tapping The Potential Even with their primitive intellects, robots have demonstrated an ability to capture substantial gains in productivity, efficiency and quality, notes robot pioneer Joe Engelberger. Some of the "smartest" robots work outside manufacturing if one counts their implementation as space laborers, surgeons and pets such as Sony's AIBO mechanical dog. In some ways, they're bellwethers of what might be possible on production floors if manufacturers realize that industrial robots don't have to be bolted to the floor or constrained by the limitations of yesterday's control concepts and outmoded software algorithms. Big advances have occurred in the strategic value of robotics since GM implemented its first robot, a Unimate made by Engelberger's Unimation Corp. Installed at a GM Ternstedt, N.J., plant, the hydraulically powered robots were relegated to a task in the 3-D category -- dull, dirty and dangerous. The Unimates unloaded hot, heavy parts from die-casting machines. A few years later, GM installed the first spot-welding robots, and painting by robot was next followed by parts assembly, says Steve Holland, director of controls, robotics and welding, General Motors Corp.'s Technical Center, Warren, Mich. (GM's first assembly robot, Unimation's PUMA, is now on display at the Smithsonian.) The technological watersheds since that first industry implementation have completely revised the capability, performance and strategic benefits of robots. For example, by the 1980s robots transitioned from being hydraulically powered to become electrically driven units. Accuracy and performance improved. Then with the rapidly increasing power of the microprocessor and new computer control algorithms, robots have dramatically increased their potential as a flexible automation tool. The new fundamental is intelligence-robotic technology converging with a wide variety of complementary technologies, says senior analyst Dick Slansky, ARC Advisory Group, Dedham, Mass. He cites machine vision, force sensing (touch), speech recognition and advanced mechanics. The result: exciting new levels of functionality for areas never before considered practical for robots, adds Slansky. "Another growth driver is Intelligent Assist where operators manipulate the robot as though it were a bionic extension of their limbs with increased reach and strength [Intelligent Assist Devices, or IAD]. The introduction of robots with integrated vision and touch will dramatically change the speed and efficiency of new production and delivery systems." Robots have become so accurate that they can be applied where manual operations are no longer a viable option, observes Slansky. Semiconductor manufacturing is one example. The same consistent high level of output, power and quality cannot be achieved with humans and simple mechanization, Slansky says. Brian Carlisle, CEO, Adept Technologies Inc., a Livermore, Calif., factory automation provider, dramatizes that point: "Even if labor costs were eliminated, a strong case can still be made for automating with robots and other flexible automation. In addition to quality and throughput, users gain by enabling rapid product changeover and evolution that can't be matched with hard tooling." Boosting Competitiveness Robotic applications originated in the automotive industry, and GM, with 25,000 mechanical workers, continues to utilize and develop new approaches to the technology. "The ability to bring more intelligence to the robot is giving us new strategic options," Holland says. GM, in effect, is pursuing technology investments to add shareholder value, observes GM's Holland. "Generally speaking, car pricing has actually declined over the last two to three years, so really the only way manufacturers are going to continue generating money for shareholders is to cut structural costs. "When we convert a plant to a new product, hundreds of millions of dollars are put into the facility. Our manufacturing technology focus minimizes the capital investment by increasing equipment flexibility. When we're looking for new robot applications now, we're often seeking alternatives to operations that are already automated with dedicated equipment. We want to take advantage of increases in robot flexibility to perform those same automated operations more consistently with more common equipment and at lower cost." Holland cites simulation tools as another way that advances in computer technology are benefiting the application of robots. "For example, a decade ago computer power often was inadequate to properly simulate robot implementations." In contrast, Holland says assembly operations at GM's new Lansing Grand River plant were easily and quickly simulated during the early planning stages. The simulation step today is an important engineering tool for reducing risks in deploying automation. Assisting And Fixturing Another sign of the increasing convergence of computer power with assembly tasks is the robot-like technology of IADs. IADs are not replacements for humans or robots, but rather a new class of ergonomic assist technology that helps human partners in a wide variety of ways, including power assist, motion guidance, line tracking and process automation. Holland says IADs use robot control technology to help production employees with ergonomic challenges in part handling. Using a human-machine interface, the operator and IAD work in tandem to optimize lifting, guiding and positioning movements. Sensors, computer power and control algorithms translate the operator's hand movements into the lifting power required. A current example is an implementation at Visteon's Chassis Division, Dearborn, Mich. The IAD helps employees maneuver unwieldy 4-foot-by-3-foot, 42-pound catalytic converters from a turntable to a gage fixture for testing. Before the IAD was installed, an operator would have manually lifted 62,200 pounds over an eight-hour shift. At an equipment cost of just 25% higher than traditional lift assist equipment, Visteon reports a 100% gain in labor productivity with no risk of ergonomic injuries. Other users include Ford Motor Co., GM Honda, Toyota and the U.S. Postal Service. IADs leverage the recent quantum leaps in microprocessor, sensor and control technologies, says supplier Paul Decker, president, Cobotics Inc., Evanston, Ill. Decker equates the control challenges of designing IADs with those of developing the fly- and drive-by-wire concepts and the Segway two-wheel personal transport device. "All those applications awaited low-cost compute power." Expect more innovations in robot configurations as the economic implications of Moore's law continues to shift computing power to a device orientation from its traditional service emphasis, says Holland. "I think the biggest change that we're going to see in industrial robots is that they will evolve into a broader variety of structures and mechanisms. In many cases, configurations that evolve into new automation systems won't be immediately recognizable as robots. For example, robots that automate semiconductor manufacturing already look quite different from those used in automotive plants." Another new robot configuration is revising GM's approach to parts fixturing at the company's Lansing Grand River assembly plant. Most people still think of a robot as a programmable device intended to perform some kind of operation on a part -- such as dispensing sealants or adhesives or performing spot welding. Instead, GM engineers have designed a novel robot whose primary function is to hold a wide range of part sizes for machining. The device functions as a reconfigurable fixture -- a flexible alternative to conventional dedicated parts-holding units. About the size of a basketball, it serves as a locating device to position tools in exactly the right place for different models as they come down the line, explains Holland. The intent is to replace the traditional system that requires cumbersome tooling stations that index back and forth for different models, Holland adds. Like conventional robots, it's easily reprogrammable and replaces a conventional dedicated solution. Holland predicts that "we will see the day when there's more of these programmable tooling kind of robots than all of the traditional robots that exist in the world today. It is an enormous sea change that is coming, and it has a lot of potential because it greatly increases fixturing flexibility -- an advantage that wasn't possible before." Eventually, Holland predicts, "GM will source these kinds of devices from mainstream companies much like we buy conventional robot systems." Envisioning Vision Despite all the anthropomorphic desires of leading robot researchers to emulate the human appearance and intelligence of Capek's characters, it hasn't happened. Most industrial robots still can't see, and there are few examples of bipedal, upright walking research robots such as Honda Motor Co.'s P3. A minority of the installed base of industrial robots are integrated with machine vision systems, says ARC's Slansky. That's why it's called machine vision rather than robot vision. Machine vision, probably the ultimate computer peripheral, was first championed in the late 1950s by researchers at the Massachusetts Institute of Technology, Carnegie Mellon University and Stanford University. The early machine vision adopters paid steep prices because of the technical expertise needed to implement such systems. For example, in the mid-1980s, a flexible manufacturing system from Cincinnati Milacron included a $900,000 3-D vision guidance system. By 1998 average costs had fallen to $40,000, and declining prices are the rule. The major tasks include inspection, identification gauging and guidance (eyes for the robot). Today, simple pattern matching vision sensors can be purchased for under $2,000 from Cognex Corp., Natick, Mass., Omron Corp., Schaumburg, Ill., and others. That reduced price reflects both today's reduced computing costs and astute packaging of vision systems for specific jobs such as inspection. A remaining opportunity for machine vision suppliers is to challenge the standard presumption that robots are already flexible automation, says Babak Habibi, president, Braintech Inc., North Vancouver, British Columbia, provider of vision-guided robots. He asks a question that a traditional robotmaker might not emphasize: "How can sightless robots be flexible if they can't quickly accept and adjust to information feedback?" With traditional so-called flexible robot, precision fixtures are required to properly present the part to the end effectors. Vision-guided robots remove that rigid constraint by adding true flexibility, he emphasizes. "Our attention was first drawn to vision-challenged robots in 1998 by a major robotmaker, ABB Global Engineering Solutions, [Brampton, Ontario]," says Habibi. "Until that time, our business focused on inspection solutions for food and medical products -- developing and applying algorithms for image processing and object detection and identification. ABB's customer was seeking a vision-guided robot system for deburring aluminum wheels." The company's second team effort with ABB was for a plant of TI Automotive Ltd. A vision-guided system was mandated because the ABB robot needed to be able to see and react to changing conditions within a factory environment. Habibi explains that the vision-guided robot system had to continuously adjust to unpredictable variations in the dimensions and orientation of blow-molded plastic fuel tanks. "Plastic is affected by variables such as temperature and humidity, so an accurate, fully integrated vision system to guide the robot is critical to the process," says ABB general manager Barry Mitchell. "Now the robot can quickly identify dimensional variations." To Habibi, vision-guided robots represent more than a new interesting application area for vision software suppliers. "We think that the acceptance of vision-guided robots will enable a significant shift in manufacturing process design. With this added capability, manufacturing lines can be designed with a whole new mentality." With a sense of irony evident, he explains that manufacturers will finally be able to do away with all the hard automation that traditionally supports "flexible" robots. "It will no longer be important that parts always show up in exactly the same position."

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.