Killbots on the Rampage (19 October 2007)

by Iain Dooley | Permalink

explore : Iain Dooley, Iain Dooley, Home Page, Opinion, Robotics

Robotics is all around us. We rely on software at every turn, but software is so error prone, do we really thinking of the consequences of things going awry? This article in Wired magazine describes a scene reminiscent of the prototypical Robocop in the movie of the same name:

http://blog.wired.com/defense/2007/10/robot-cannon-ki.html

My mechatronics professor always used to say the first thing you should think about when designing a machine is what happens when it fails. If something in your software fails, does the machine stop moving? or does it accelerate into a crowd of small children?

If your gun target register overflows and you get a bogus sensor reading, does your gun power down and wait for further instructions or does it start firing wildly into the air like Yosemite Sam?

We should really think of robots as people with very bad autism. They might be amazing at picking out moving targets, but what happens when they miss jepoardy? Would you put them in control of a self-reloading high calibre anti-aircraft rifle?