Attack of the Robotic Cockroaches

p>

>> Email the Sensors Editors

>> Today at Sensors Archive

The Age in Australia had a recent story about the insbot, a biologically-inspired robot that can (with the addition of cockroach-specific pheromones) blend in with the nasty little insects. It moves like them, it smells like them, and other than the fact that it looks nothing like them, the cockroaches react to it as if it's just an odd-looking cousin. The insbot isn't alone. The world of robotics research is, if not rife, at least well-populated with biologically inspired robots.

Think Like a Bug, Act Like a Bug

So, why the interest?  And this isn't a short-lived interest, either.  Take a look at this list of citations for biomimetic robots and you'll see some from the early 90s.  One of the reasons is to replicate how non-humans interact with their world. Some of the simplest non-human things out there are insects, so the algorithms to navigate and explore are simpler than those we use. Theoretically, once they've got the robotic bugs down pat the researchers can move up the food chain.  Another reason is to explore learning and adaption, especially as it relates to unstructured environments.

There's also the idea that nature is really, really good at what she does and why don't we replicate it to see if it's better than anything we could come up with on our own?  Because oh, how we want better autonomous robots!

Present Day

Have the robo-bugs helped us at all?  Yes and no.  Yes, because back in 1986 Rodney Brooks suggested using a bottom-up approach (known as subsumption) to create complex behavior from a simple system. Take a look at this implementation from the Idaho National Laboratory to give you a better idea of how this actually functions. Modern autonomous robots use a combination of this approach and deliberative behaviors (creating a plan of action, essentially).  Here's an excellent primer on robot control, explaining basic approaches and tradeoffs.

The DARPA Grand Challenge is an example of what happens when these research projects grow up. It's a test of what really works and which approaches are useful rather than being interesting laboratory curiosities.  Sitting on the sidelines I can't wait to see how the DARPA Urban Challenge plays out, where the robots must navigate an urban course.

Read more on