DPRG
DPRG List  



[DPRG] DARPA Grand Challenge

Subject: [DPRG] DARPA Grand Challenge
From: Steve Rainwater srainwater at ncc.com
Date: Tue Dec 17 11:36:03 CST 2013

On Tue, 2013-12-17 at 10:17 -0600, David P. Anderson wrote:
> Does anyone else think this is silly? 

Definitely not practical but fun to watch. ;) This is just DARPA trying
to push the technology forward a little more. I haven't seen any video
of the Virginia Tech THOR robots in action yet, so no idea if they
really work. Based on the first Grand Challenge, I'd expect at least
half of these robots to immediately fall over at the starting line. But
the Boston Dynamics humanoids that will be in the challenge are making
pretty good progress and you can see YouTube vids, so it seems likely to
be the real thing. 

Boston Dynamics petman
http://www.youtube.com/watch?v=tFrjrgBV8K0

Boston Dynamics Atlas
http://www.youtube.com/watch?v=zkBnFPBV3f0
http://www.youtube.com/watch?v=SD6Okylclb8

> As we've discussed before, why have a humanoid robot open the door? Why 
> not just have the door open itself?  (see your  local grocery store).  

Agreed. But DARPA's scenario is looking for a general purpose robot that
can complete a series of actions in an environment designed for humans.
I can't remember the specifics but I believe the robots have to enter a
simulated disaster site, find an available vehicle (designed for human
drivers), navigate a rubble-filled course to a specific building, kick
down the door, climb stairs/ladders, and break through some obstacles
using power tools found onsite (designed for humans), and then diddle
with some kind of equipment - shut down a simulated reactor before it
destroys the planet or something.

For day to day to use, I still agree with you that automating doors,
houses, and cars is far easier than building a humanoid robot to
manipulate them. But in these specialized rescue robot scenarios? I
dunno, seems impractical now but in another 10 years, or for
super-important disaster areas, it might make sense.

On the other hand, those CMU snake robots look like they'd have a much
easier time than a bulky humanoid robot getting into and out of broken
down nuclear reactors to examine and fix things.

> Steve, remember how the Aibo was going to replace real dogs for old 
> people so they didn't have to feed and cleanup after them, but they 
> would otherwise have all the joy and affection of a real puppy dog? 
> Didn't happen.   I wonder why?

In part because they never got cheap enough to seem worthwhile. And I
think the folks pitching the business model didn't make the robots
practical enough to be worth the money. There are still plenty of
initiatives working on old-geezer-companion-bots but now they seem more
oriented toward practical applications like following people around,
reminding them to take their pills and relaying evaluations of their
medical status to the doctor as needed.

-Steve


More information about the DPRG mailing list