By Doug Peterson
November 2006
Some of the surgeons doing the finest work today have not attended a single day of medical school. Instead, they have come from a factory and have hands made of steel and cameras for eyes.
They are robots, guided by the movements of human surgeons, and they can make incredibly fine incisions without the natural tremors that afflict hands made of flesh and blood. Already, robotic systems are showing up in surgical centers to perform thousands of operations each year, including heart bypass surgeries and hysterectomies.
When patients literally put their lives in the hands of a robotic device, it dramatically underscores the importance of building trust into human-robot interactions.
CSL is known worldwide for its work in reliable, trustworthy systems, so it is a natural place to explore issues of trust with robots. Many of its researchers are involved with the University of Illinois Information Trust Institute (ITI) and the new Center for Autonomous Engineering Systems and Robotics, or CAESAR.
“More and more, we are entrusting critical tasks to machines -- and not only robots but automobiles and airplanes with their embedded controls,” says Mark Spong, a CSL professor and director of CAESAR. He says the new center draws together expertise in robotics that is now spread across campus. The Center has a core of 10 scientists, with close to 50 affiliate researchers.
As examples of the work going on in CAESAR, Spong’s own research projects include one on walking robots and one on multi-robot teleoperation, which is the control of many robots from a remote location.
“For example, let’s say you want to build a habitat on the Moon,” Spong says. “You’re not going to send a construction crew there because of the expense. But you’re also not going to do it purely with robots.”
What is most likely, he says, is that a small number of humans will direct the robots, either from the surface of the Moon, from an orbit around the Moon or from the Earth. These robots will have to work together, cooperating to lift and carry objects.
With this goal in mind, Spong’s research team is learning how to reliably control a fleet of “segbots” -- small robots that balance themselves on two wheels. He says they have developed manipulator arms for one of the segbots, and they plan to eventually be able to control the fleet with a joystick.
Spong is also exploring the possibility of bipedal, or two-legged robots, which can play an important role on rough terrains. As he puts it, “Most of the Earth’s surface is not paved. You can’t use wheels everywhere.”
For instance, a robot might need legs to step over obstructions at a disaster site. Legs also might be much easier for navigating through a forested area if robots are used one day to cut down trees.
According to Spong, two-legged robots have received a lot of attention in Japan. Honda, for instance, brings along its humanoid “Asimo” to trade shows. Although Asimo can carry a tray, greet people and push a cart, it’s not very energy-efficient. Spong’s team is aiming for a more energy-efficient robot that takes advantage of a leg’s natural swinging pendulum motion to generate walking motion.
Science fiction has often portrayed robots running amok and breaking free from their human controllers. But these far-fetched scenarios are not the issues of “trust” that Spong is talking about. In CAESAR, the major trust issues are stability, reliability, and availability. By “availability,” he means the robotic system must be available at the precise moment it is needed. And “stability” means the robot must be able to function under specific conditions that could pose problems.
For instance, a robotic surgeon has to be able to overcome the time delay that results whenever humans control a machine over a network. Time delays can cause the system to become unstable, which is not what you’re looking for when a robot surgeon is making a fine incision.
“Robots are so complex that sometimes you can’t predict what they’re going to do,” Spong says. “Trust is all about making their actions more predictable.”