Friday, May 29, 2015

Two ways to see this...

Two ways to see this...  Watch the video at right, then read this.

Most software engineers viewing this will have the same reaction I did: some variant of “Holy crap!” Why?  Because that robot is doing something that is incredibly challenging to program, especially in real time.  Almost every element of that robot, from sensors to actuators, poses a very difficult programming challenge.  That challenge is even harder if the solution you're seeking is a general one (i.e., let this robot run anywhere and jump over obstacles) as the MIT program is, and not a specific one (i.e., jump over this set of specific obstacles). 

Most other people will look at this and say something like “Well, this is cool and all, but what else can it do?  Just run?”

And in a very roundabout way, that's why I'm not too worried about artificial intelligence powered robots taking over the world anytime soon :)  It's just too damned hard for conventional computers and conventional programming methods, no spectacular breakthroughs are around the corner (at least, none that I know of), and consumer expectations are light-years ahead of the reality.  I suspect that the spawn of Roomba is what we can reasonably expect for quite a long time...

No comments:

Post a Comment