Ralph Hollis is headed to Santa Cruz, Calif., to attend what is considered the premier annual conference in robotics. He's worried. This invitation-only event brings together select researchers who debate potential breakthroughs in the field. On this international stage, Hollis is slated to give a talk about a new technology he has been working on at IBM-a robot wrist that could perform many of the same functions as the human hand.
Problem is, even as he finishes his paper for the Fourth International Symposium on Robotics Research, he still can't get his prototype to work. "We tried day and night, but nothing," recalls Hollis, now a research professor at The Robotics Institute at Carnegie Mellon. "But I had to go and give the talk, right?"
It had been a tradition at the symposium to bestow an award to the best paper. At a conference of such repute, this award represents the best of the best, and it's the kind of recognition any presenting author would covet. Hollis, though, harbors no such illusions. He just doesn't want to embarrass himself. After he delivers his presentation, he warily fields some questions. Then, thinking he's in the clear, he gathers his notes. But as he exits the stage, Matthew Mason, a professor in Carnegie Mellon's Department of Computer Science, voices one final comment: "This paper should get an award for the idea least likely to succeed." Hollis can appreciate the humor of the remark, but he realizes, too, as he walks off the stage that it's almost as if doubts are being cast on his life's work.
Robots were a calling Hollis first discovered in 1957 as a high school student in Kansas. While others his age were listening to Elvis Presley, he was absorbing the latest issue of Scientific American. He built his first mobile robot that year after reading an article in the magazine by British neuroscientist William Grey Walter, who constructed some of the earliest autonomous robots. "Walter had made some small little creatures that had vacuum tubes in them so they could find their way around the room and sense obstacles," Hollis says. "I thought that was the coolest thing ever, and I really got inspired."
It led to a career. "I've probably done robotics longer than anyone here at Carnegie Mellon," admits the 69-year-old professor, pointing to his shock of white hair. "I pursued it as a hobby first, because there was no such field as robotics back then."
Hollis built robots throughout his college years at Kansas State University, where he studied physics. In 1975, he obtained his PhD in solid state physics from the University of Colorado, where, just for fun, he constructed one of the first robots with an on-board computer. The robot, named Newt, was featured in a cover story in the computing magazine Byte-a framed copy still hangs in his office in Smith Hall today. Three years later, he joined IBM, where robotics finally became his job, not just a pastime.
At IBM, Hollis developed technologies for automated assembly and testing of the company's increasing number of products. One of his inventions coupled a large robot arm with a hand-like function that could fit together tiny parts in an assembly process. These robot arms achieved their precision by moving in three ways-up and down, left and right, and backward and forward at the point of contact. But Hollis wasn't satisfied. He wanted to create a device that could move in six ways, also tilting forward and backward, turning left and right, and rolling side to side. This would give it even greater flexibility for taking on tasks best suited for the human hand. Unfortunately, creating the "six degrees of freedom" would entail adding a slew of motors, cables, and other parts that would encumber, not enhance, the motion. "I was baffled about how to design the linkages," he recalls.
He stewed over the problem at the breakfast table. On his way to work. In the lab. Throughout dinner. Even at bedtime. One evening, while sitting on his living room couch, he had an idea: "Eliminate the mechanisms that support everything and have the device just float in magnetic fields." Such technology would be ideal for precision assembly of computers and other machines since they would have full range of motion without any mechanical complexity. The floating devices would also be isolated from any environmental disturbances, such as a forklift rumbling across a factory floor.
Hollis excitedly talked it through with his wife, Beth, who isn't a trained scientist but has always served as a sounding board for her husband's theories. That night, he grabbed a pencil and pad to scribble down some preliminary drawings and a few calculations. His wife looked at his sketches, but she didn't really understand where they might lead. Still, she bugged him to sign and date his papers-just in case.
Back at work, Hollis enlisted the help of Peter Allan, an intern, and Tim Salcudean, a new hire, to put his idea into action. The trio cobbled together a crude instrument using surplus power amplifiers and a $9,000 digital signal processor hooked up to an IBM PC/XT.
But there were no magical results to report when Hollis spoke at that 1987 conference in Santa Cruz. Returning home after that nightmarish presentation, he resumed modifying his design, motivated by "perseverance and stubbornness." A few months later, there were shouts erupting from his lab. "Look at this! It's impossible!" Hollis and his colleagues went and grabbed others in their offices, imploring them to hurry to the lab and see the device floating in magnetic fields on the bench top.
Thrilled at their achievement, the researchers quickly planned a demo for their department head. When they began the presentation for him, it didn't go exactly as planned. Instead of the device levitating, it started to emit smoke.
"Very interesting," the boss said sarcastically as he headed for the door.
"It turns out there was a bug in the software," says Hollis.
Undeterred by the setback, Hollis and Salcudean, now a professor at the University of British Columbia, continued to refine their magnetic levitation-or maglev-technology and got it to float smoke-free. Ultimately, they used the technology to develop a robot called the Magic Wrist, capable of delicate precision assembly by moving in six degrees of freedom. However, IBM chose not to equip its plants with Magic Wrists. On top of that, IBM was downsizing its research division. Hollis decided it was time for a career move. In 1993, he joined the faculty at Carnegie Mellon's Robotics Institute, bringing four of his maglev devices with him to his new Microdynamic Systems Laboratory.
At the university, he became intrigued with haptics. The word haptics arises from the Greek root haptikos, meaning "able to grasp or perceive." Haptic interaction with the world is manipulation using touch-our most important sense, says Carnegie Mellon psychology professor Roberta Klatzky, an expert on the science of touch who collaborates with Hollis. "If you lost the sense of touch, you would die quickly," she explains. "You couldn't stand up because you couldn't feel your feet. You couldn't masticate food without biting your tongue. You couldn't handle any tools. A newborn couldn't suckle. Touch is the key to all our interactions with the world."
Imagine if a computer could accurately replicate touch:
A dental student could learn the difference in texture between real and diseased gums without working in an actual mouth. Industrial designers could feel and shape mock-ups of car parts or home appliances without having to build costly samples. An online shopper could touch a sweater before deciding whether to buy it. Surgeons could experience what it's like to cut through an organ without having to practice on real patients. Physics students could feel electromagnetic forces between atoms. A soldier could decide how much pressure a remotely operated robot should apply to diffuse a landmine while safely located miles away.
With support from the National Science Foundation, Hollis and a PhD student, Peter Berkelman (CS'99), began to repurpose his maglev technology for these kinds of haptics applications. Based on the work he began at IBM, he developed a system that consists of a single moving part-a bowl-shaped device called a "flotor" that is embedded with six coils of wire. Electric current flowing through the coils interacts with powerful magnets underneath, causing the flotor to levitate. A user sits in front of a computer, holding a handle attached to the flotor. The handle works like a freewheeling computer mouse. With optical sensors and digital processors, forces and torques are sent to the handle through the flotor's electrical coils, creating the sensation of touching what's on the screen.
Other haptic devices exist, but none use magnetic levitation, which means the touch signals are muddled by the mechanics of the equipment's own motion. Hollis says the beauty of his device is that it has no distortion, "what you feel is more pure." Klatzky agrees. "It's almost a magical kind of thing. The forces you feel are very real, yet you don't have any mechanical interactions."
For his work in maglev haptics, Hollis was awarded a highly competitive $300,000 Major Research Instrumentation grant from the NSF in 2004. He formed an international consortium of haptic researchers and distributed six maglev haptic interface devices to its members, among them Hong Tan, a professor of electrical and computer engineering at Purdue University. "Ralph's device is the first one with the resolution to simulate a surface with that much precision," says Tan, a haptics expert who studies human perception of fine surface textures, which requires simulation resolution at the micron level. "None of the commercially available devices come close."
In 2007, Hollis formed Butterfly Haptics LLC, which produces Maglev 200, a commercial version of his device. He chose the company's name because butterflies seem to float in the air like the product's flotor. Support from Carnegie Mellon and the seed-stage investor Innovation Works (plus a sizable home equity loan) enabled him to put his device through another round of engineering and bring the product to market, where it retails for about $48,000.
The potential for Maglev 200 is vast, but Hollis acknowledges a gap between the device and its applications. For instance, Maglev 200 could produce the forces that show an army medic what it feels like to pull a sliver of glass out of an eye. But that application would require a 3D graphic of an eyeball and a physical model describing those forces based on the behavior of actual eye tissue, and those don't currently exist in an advanced enough state to take full advantage of what Maglev 200 offers. "What we have is a general purpose tool that is a solution looking for a problem," Hollis says. "But if ways to fill that gap can be developed, it's going to be gangbusters."
Butterfly Haptics is very much a family affair. Beth Hollis, who has a background in accounting and 35 years of experience as her husband's unofficial research assistant, serves as president. Their son, Benjamin, who graduated from Carnegie Mellon (CS'05) and works at Amazon, designed the company's Web site. And some of the complex parts of Maglev 200 were built in the family's basement workshop. It's a business model that's working. Today Butterfly Haptics has customers throughout the United States and Canada, with plans to expand to Japan and Europe, and the company is "in the black," Hollis says.
What's more, Maglev 200 won the 2010 R&D 100 award from R&D magazine, which recognizes the 100 most technologically significant products of the year. Past recipients have included the ATM machine (1973), fax machine (1975), liquid crystal display (1980), Nicoderm anti-smoking patch (1992), and high-definition television (1998).
Not bad company for an "idea least likely to succeed." As for that early detractor at the robotics symposium, Matthew Mason is still on the Carnegie Mellon faculty. He is now Hollis' friend and colleague-and director of The Robotics Institute. "At the time, I let it go," laughs Hollis, recalling Mason's jab. "But today, I won't let him forget it."
Jennifer Bails is an award-winning freelance writer. She is a regular contributor to this magazine.
Related Links:
Butterfly Haptics, LLC in the News
Microdynamic Systems Laboratory
Psychophysics of Haptic Interaction
