Wednesday, May 26, 2010

(Come On, Baby) Robot Locomotion!

I came across a video on YouTube* yesterday that was just too cool not to do a post on. (Don't believe me? Just watch it.) It seems we're progressing quite nicely on the non-human locomotion front these days—as in, we're seeing more non-bipeds, and we're seeing better learning skills and problem solving techniques from the robots themselves.

So, not only do we have robots that can navigate underwater obstacle coursesclimb walls**, and go "rock" climbing, but now we also have quadrapedal bots that can navigate any terrain, stand up when they fall, and learn from mistakes:



Add to this robots that evolve, solve mazes, and possibly exhibit predatory-prey behaviours***, the wall-climbers, and other locomotive advances, and we're looking at some awesome possibilities for the future. Also some scary ones. For instance:

  • Let's say the evolving robots were given Little Dog bodies (as in the video above) and self-assembling circuits****. Would they heal themselves? Reproduce? Be easily distinguishable from "life"? What if we added in the need to eat organic matter for fuel?
  • Could we send Little Dog bots with cameras and testing equipment to other planets, moons, and asteroids? It seems that a lot of what we send to Mars has terrain issues. If we didn't need to rely on wheels, and could climb over rock piles instead of going around them, and therefore get closer to cool geology, faster…
  • What if we made Little Dog waterproof?
  • Who says our service-bots must have wheels or be bipedal? 
  • Weaponizing a lot of these bots is pretty obvious, and the Israeli bots already getting military backing. So, how would we combat a sub that could think for itself? 
  • How would we deal with a weapon/spy bot gone rogue, smart enough to evade capture? We've had stories like that, but they've always been human-like robots, I think. And then, if there were multiple rogues and they formed a pack…
  • How would vehicles be changed if we added Little Dog's self-correction routines to the "autopilot" functions we already have?
  • Could we scale Little Dog up or down, and what could we use those for? Could we make one big enough to ride in?
Anyone want to weigh in? I'm getting all excited again…


*via, of all things, a retweet of @ebertchicago
** Engadget
*** Engadget
**** Futurismic

No comments: