Selasa, 19 Februari 2008

Bug Robotic

I don't show many people on this site, but there is no discussion of bug robots without Rodney Brooks. He pretty much started the bug robot idea. There is so much information online about Rodney, his work at MIT and with iRobot that I am not even going to bother with that. I will tell you about a time I had lunch with him in the mid-1990's just off the Stanford University campus. I'm sure he doesn't remember. It was a group of ten or twenty people presenting research work and I was lucky enough to sit at the same table as Rodney. We talked about our kids, our wives, the weather and none about robots. He's a nice guy; with an unassuming, down-to-earth personality and an Australian accent. I do find it interesting that Rodney's robotics research has progressed from bugs, to horses, to dogs and now on to humanoid robots. A kind of evolution in a lifetime.

The basic idea behind the bug robots was to try to understand very simple biologic creatures and create corresponding robots before trying to build highly complex robots that try to mimic human reasoning. Even a fruit fly with less than twenty neural connections can fly, avoid obstacles, find food and mate (however it is that fruit flies mate). This idea makes a lot of sense, but in practicality building very tiny robots is quite difficult. It seems like the people who work on bug robots spend more time developing techniques for building tiny robots than they do on studying bug behaviors and ways to mimic them.

If you want to give bug robots a try, you might consider the little BugBrain by Yost Engineering on the left. Those big whiskers on the front give the bug the ability to sense contacts with objects in its environment and you can program the onboard computer to make decisions about how to react. That's a true robot. You can add other sensors to it too. Maybe a phototransistor so the bug can "run for the shadows" like real bugs do? According to the manufacturer you can also add wireless RF and an ultrasonic range finder to the bug. That could give it sensing and decision making capabilities on-par with University research robots.

The exploration of Mars is one application that has been proposed for bug robots. Instead of sending one or two big robots, send one or two thousand bug robots equipped with small cameras and chemical sensors. One of the advantages of this approach is fault tolerance. If a few of the bugs break or get lost, it is no big deal. Another is the ability of the small bugs to get into small places such as cracks or fissures in rocks. Or course the small size of the robots also limits the scale of tasks they can accomplish. For example even a thousand bugs working together are not going to drill a core sample ten feet into the Martian crust.