Last week, in Bremen, Germany, saw the start of the 10th annual RoboCup, wherein robots play soccer. If you think this is just another clever but non-useful gathering of geeks with toys, think again.
Eye on the Prize
It's easy to get caught up in the Gee Whiz factor of contests such as the RoboCup and the DARPA Grand Challenge but it obscures the real achievements of the people behind the robots. Autonomous robots need to sense and react to their environment in real time. That calls for fast sensors and very, very clever software, all in a package with size and power restraints.
In the RoboCup, robots playing soccer present the following challenges: having more than one robot, all acting in concert; requiring each robot to identify relevant objects, self-localize, dribble AND ALSO cooperate with its team mates to pass the ball or help them; and finally, a shifting environment, with moving objects, some of which are, to quote from the RoboCup site, "rational agents that play against your team."
The various leagues (small size, middle size, 4-legged, and humanoid) differ in the number of sensors and complexity of behavior. I was interested to see that the 4-legged league uses specially programmed AIBO robots. Now that SONY isn't making these anymore, will the teams be able to get replacements? There are also leagues for RoboRescue and RoboJunior. RoboRescue is about developing both software and hardware for disaster rescue situations, including information infrastructures and standard simulator and decision support systems. Teams are figuring out how to get multiple robots to work together, how to handle the data they produce, and how to standardize the systems so everything works together.
You Can't Succeed Unless You Fail
Henry Petroski, in To Engineer Is Human: The Role of Failure in Successful Design, discussed how design advances through failure. A previously proven design, when extended to a new situation, will fail in interesting ways. Those failures can help you create a new design because they teach you the drawbacks of the original, which weren't evident when everything was working properly. For instance, engineers learned all kinds of interesting things about suspension bridges when the Tacoma Narrows bridge failed so spectacularly in 1940.
That's really what these robotic challenges are about. Learning what works, learning from what fails, and creating something better.