While it's not yet ready to explore inner space in a real patient, a "colon robot" being designed in the Artificial Intelligence Laboratory has been getting smaller and more practical in the past year.
Students in MIT's micro-robotics group (part of the AI Lab) have been working on building a robot that would some day be capable of moving through a patient's colon to the site of a polyp or other problem and surgically removing it with help from a tiny camera and other tools. The robot would be controlled by a surgeon who could perform procedures without having to cut open the patient or use a long flexible endoscope (MIT Tech Talk, May 18, 1994).
Arthur Shectman, a senior in mechanical engineering, has advanced the project from James McLurkin's Cleo, a robot that's small but lacking the necessary shape and design for colonic navigation, to one that's approaching the right dimensions. The cylindrical device is about three and a half inches long and just over an inch in diameter, and its electronics are sealed against contact with the large intestine's hostile environment.
The biggest challenge in designing a colon robot is inventing a means of locomotion that can handle the terrain (which is flexible and offers little purchase) without injuring the patient. Mr. Shectman experimented with a tube studded with a series of small "paddlewheels," but he concluded that this design would damage the intestinal lining.
Around the circumference of his latest prototype are longitudinal elastic treads powered by 10,000-rpm motors and worm drives, which reduce the speed but boost the torque. The result: a robot that moves forward at about two feet a minute.
Still to be solved is the problem of how to mount implements that the surgeon will control from outside the patient's body: a camera, sources of light and air (to distend the colon and allow room to work), and suction for surgical debris. Further downsizing is also necessary; surgeons with the Advanced Research Projects Agency, which is helping to fund the research, are hoping for a thumb-sized device, said Mr. Shectman. His faculty advisor for the project is Rodney Brooks, professor of electrical engineering and computer science and associate director of the Artificial Intelligence Laboratory.
The elements limiting miniaturization at the moment are the motor and its accompanying reduction gears, but doctoral student Anita Flynn and Dean Franck, a senior in mechanical engineering, are working on a new and much smaller type of motor that could solve this problem. The ring-shaped ultrasonic piezoelectric motor is just 8 mm in outer diameter and 2 mm thick. A piezoelectric ceramic material in bulk form, lead zirconate titanate (PZT), is bonded onto a ring of aluminum or steel. Appropriately applied electrical fields induce a traveling wave of mechanical bending in the ring, causing surface points on the ring to move in a retrograde elliptical motion. When a rotor is pressed down against the ring, it is propelled through friction.
The advantage of piezoelectric ultrasonic motors is that they are lightweight and compact because they run at low speeds with high torques, thus eliminating the need for gears. The technology is already in commercial use on a larger scale, principally in the motors that operate autofocus camera lenses, Mr. Franck noted. Ms. Flynn's doctoral research, in a joint project with Lincoln Laboratory and the Penn State Materials Research Lab, also encompasses scaling down the actuators using thin-film PZT that can be deposited directly onto silicon for production of micromotors, which can be much smaller and driven at significantly lower voltages than the bulk motors.
Mr. Shectman has tested the latest prototype on a surface approximating that of the colon: a chicken skin stapled into a tubular shape and lying on a tray of red Jell-O. The robot is years away from testing inside people, and with Mr. Shectman's impending graduation, its future is unclear. "It's kind of sad to leave the project when we're finally starting to make some progress," he said.
A version of this article appeared in MIT Tech Talk on April 26, 1995.
