Johns Hopkins Magazine - April 1994 Issue

At the Slithering Frontier of Robotics

By Melissa Hendricks

The snake-like robots that Gregory Chirikjian invents can squirm their way into nooks and crevices where no human or tool could ever hope to reach.

Hasta la vista, baby." Our Austrian hero fires a shot directly at the policeman.

The policeman, actually an evil Terminator, shatters like a glass thermometer. Pieces of its body scatter over the floor of the factory, which is now a crumbling inferno. But all is not over, for the fragments melt into shiny metallic puddles that meld together like liquid mercury. Slowly, the glistening pool stretches upward, transforming once again into the evil Terminator, which turns to face our hero and resume its deadly chase.

For several days after Gregory Chirikjian '88 saw Terminator 2 two years ago, he daydreamed about the possibility of building a robot like the Terminator. "Wouldn't it be great if a person could have modules on them, sort of an exoskeleton?" thought the assistant professor of mechanical engineering. When you wanted to reach something from afar, you would point your finger. The modules would grow out from the end of your finger in a line reaching toward the object. A "hand" at the end of the line would grab the object, and the modules would retract back toward your body.

Chirikjian knew that the technology to build a liquid robot does not exist. But if the robot were made out of discrete modules, and if the modules were small, and if there were enough of them, he thought, "the robot would approximate a liquid, or continuous medium."

Chirikjian, a tall and confident man with a firm handshake, was no kid having a pipedream. Though only 27, he has already proved himself in the world of robotics by designing and constructing a snake-like robotic arm that colleagues say is at the cutting edge of the field. The arm can configure itself into 33,000 different positions and has potential to slither into nooks and crevices that no human or tool could reach. To support his work in robotics, Chirikjian has won a National Science Foundation Young Investigator Award--$25,000 per year for five years, plus extra federal funds to match support by non-government sponsors.

Squirming around the floor of Chirikjian's basement laboratory at Hopkins is the 6-foot-long robotic arm. It looks something like a metal snake and is, in fact, of a breed known as "snake-like robots." Rakesh Malik, one of Chirikjian's students, sits at the controls of a computer that is wired up to the wire-and-metal arm. The arm lies perfectly straight, with no bends or kinks. Malik taps a few keys, and the arm curls to the left so that its top almost touches its bottom. The "gripper," a plastic hand at the end of the arm, grabs a roll of plastic packaging material and lifts it up off the floor. Malik taps a few more keys, and with a percussive clang, the arm curves to the right and deposits the roll several feet from where it had been sitting.

Chirikjian laid the foundation for his work on snake-like robots as an undergraduate at Hopkins (he completed undergraduate degrees in both engineering mechanics and mathematics as well as a master's degree in mechanical engineering in only five years). With Nick Jones, associate professor of engineering, he designed mechanical devices that have many moving joints, and he wrote his master's thesis on the dynamics of a satellite with an extendable arm.

Chirikjian then went to CalTech for his doctorate, where he worked with Joel Burdick, studying robots that have many joints, and from there came back to Hopkins, where he is currently one of the university's youngest faculty members. So young, in fact, that his graduate students could easily be older than he is. His students give him high marks on teaching evaluations--perhaps in part because, since Chirikjian is close to their age, they can communicate with him. Indeed, teacher and students share at least one aspect of popular culture: they have grown up soaking up some of the same science fiction robots on television.

As a kid, Chirikjian watched "Star Trek" and "The Six Million Dollar Man," the story of a man who could bend bars and heft enormous bad guys with his bionic arm. Could those programs somehow have influenced his robot designs the way Terminator 2 has? "They consciously did," says Chirikjian. "They sparked my imagination."

Science imitating art and art imitating science are familiar to robotics. The whole notion of a robot was first proposed by a playwright, Karel Capek, in his 1917 play RUR (Rossum's Universal Robots). Rossum builds artificial people for use as servants, the Czech word for servitude being robota. Rossum's robots eventually turn on their masters and kill people. (Thus Capek began a theme- -of robots becoming murderers--that has recurred throughout science fiction, in films such as 2001 and Terminator.) Since Capek, robots have populated movies and films, usually as anthropomorphic structures that walk and talk, and often perform human feats better than people can.

But movies and science fiction novels can raise false hopes, says Burdick, Chirikjian's doctoral advisor. If Hollywood can make robots like the lovable R2D2 of Star Wars, then why shouldn't people suppose that very soon we will all have personal robots who can wash floors, make dinner, and take the trash out? In reality, the technology just isn't there, and it won't be for a long time, says Burdick. He points out that R2D2 was not a real robot: much of the time, it had a midget inside.

Likewise, some industrial leaders have also had to lower their expectations. "In the early to mid-'80s, lots of people thought robots were the ticket to the future," says Chirikjian. "General Motors, especially, piled on the robots to try to lower cost." Robots had appeal because, in principle, they could reduce the number of workers required. They also could take over boring jobs and repetitive tasks that can lead to stress injuries, as well as tasks that require higher accuracy than most humans are capable of.

In practice, however, while robots do not unionize, they have their own risk: becoming outmoded. "A hand can feel and people can learn," says Chirikjian. "But a robot is generally suitable only for one task." Suppose, for example, that GM drastically changes its line of cars in 1995. If the robots that built the 1994 cars cannot do the jobs required to produce cars for 1995, they end up in the junkheap, at considerable cost to GM. Indeed, Chirikjian has seen ads in shopping circulars selling used robots.

Chirikjian's robotic snake arm may be a solution to the problem of high cost. While conventional robotic arms cost hundreds of thousands of dollars to construct, Chirikjian built his robotic arm for just $1,000. An arm good enough to market would cost only about $10,000, he says.

Most conventional robotic arms are called "continuous range-of-motion robotic manipulators" because they can reach an infinite number of points within a given territory. However, these manipulators also require sophisticated motors and computers that make them prohibitively expensive to use, says Chirikjian. "One reason that robot arms are not now used in industry is that you don't need sophisticated hardware to pick up a load and put it down."

Chirikjian's robotic arm can reach to only a finite number of places--33,000 of them. But that is enough for practical purposes, he says. And it is far simpler to build: it is powered by compressed air and controlled by an inexpensive computer printer port. The robot looks something like a metal ladder, five rungs long. Three pneumatic pistons connect each rung to the one below it: two are at the sides of the "ladder" and one runs at a diagonal between each of the rungs. As the pistons change length, the robotic arm moves.

The arm is binary, meaning that each piston can be in only one of two states at any given time: retracted or extended. There are 15 pistons, each of which has two possible states, so the robot can configure itself into 215, or 33,000, different positions.

Chirikjian compares the difference between continuous and binary robotic arms to the difference between analog and digital devices. Digital devices attain a fixed number of discrete states. A digital clock shows only a finite number of times, while the hands on a conventional clock, an analog device, sweep around the clock face reaching an infinite number of positions.

Because digital devices are cheaper and more reliable, they have replaced many analog ones in the computing and electronic industries. Likewise, says Chirikjian, inexpensive binary arms may replace continuous manipulator arms.

Chirikjian sees snake-like robots as part of the new direction in robotics. With its ability to bend in any number of ways, a snake-like robot could wend its way through the pipes and fuel rods of a nuclear reactor to look for leaks or defects. Chirikjian is even designing a micro version--possibly made from a rubber or plastic material--that could travel through a patient's intestines to help surgeons in life-saving procedures. "This disposable proboscis could inch-worm its way through arteries or help remove sections of bowel," he suggests.

Just the opposite of a Terminator.

Chirikjian's latest project actually is a type of Terminator robot. During the few weeks after he saw Terminator 2, he began seriously to ponder the idea of building a metamorphic robot--one that changes form. He sketched plans for a robot consisting of two dozen hexagons, each about six inches in diameter, made of aluminum and plastic. In a resting position, the hexagons lie flush against one another in honeycomb style, so that there are no spaces between them. They look something like floor tiles. The robot reconfigures itself when the hexagons "walk" over each other within a plane--to form a circle, for instance, or a straight line, or a diamond shape. The effect is as if a mason has rearranged the floor tiles.

In the past two years, Chirikjian and his students have developed algorithms for the movement of the robot, and they have a computer graphics program that plays out the robot's movement on the screen. Now they are building components for the first actual device, which they expect to have "walking" by next year. It will contain 25 hexagons. In its most compact form, the robot will be about as big as an average office desk top.

The metamorphic robot will be wired up to a main computer. In addition, each hexagon will have an on-board microprocessor, and every other vertex will contain a motor. A hexagon will "walk" when the computers signal the motors to change the angle of a vertex. Sensors, perhaps light-sensitive ones, will detect how close each hexagon is to a neighboring hexagon or other object. The robot could be programmed to move autonomously, or an engineer could control its movement by clicking on hexagons on the computer screen.

This metamorphic robot could enter environments not suited to people, points out Chirikjian. Suppose, for example, that an earthquake causes an apartment building to collapse, sandwiching people in between its floors.

Send in the metamorphic robot! Stretching out a length of hexagons, the robot could buttress a collapsed ceiling, allowing people to escape. Or it could pick its way through the rubble to search for survivors. A camera on the robot would relay the scene to engineers outside the building, who would direct the robot's movement through remote control.

Or suppose a flood washed out a road: The robot could unfurl over the road to serve as a bridge. Metamorphic robots might even be used in space, Chirikjian suggests--for example, to reach out and retrieve a wayward satellite.

Could these robots become so powerful that, like the Terminator or Rossum's robots, they one day start to destroy their creators? The young investigator smiles. "They could not do any physical harm like attacking people," he says. "The only harm they could do would be to eliminate the need for large sections of the workforce. They would diminish the certain level of satisfaction and self-esteem people get from their work."

Admitting that that is true, he jokes, may put him out of a job. But he is not worried. "I justify what I do by the fact that I create robots for jobs you would not want people to go into, like in the nuclear environment and outerspace," Chirikjian explains. Furthermore, "people will always be much more capable than machines. You're never going to get craftsmanship from a machine."

Melissa Hendricks is the magazine's senior science writer.

Send EMail to Johns Hopkins Magazine

Return to table of contents.