DPRG
DPRG List  



[DPRG] DARPA Grand Challenge

Subject: [DPRG] DARPA Grand Challenge
From: David Anderson davida at smu.edu
Date: Fri Dec 20 14:19:58 CST 2013

On 12/20/2013 02:15 PM, Rud Merriam wrote:
> Paul & David,
>
> Right here in Texas at TAMU College Station  there is a group using 
> robots for in disaster situations.

Humanoid robots?

That's all we're talking about here.  Not whether competitions are a 
good thing nor whether robots are useful in disaster situations --- they 
obviously are.

best
dpa




> The DARPA Challenge is not so far fetched (or silly). Granted the 
> CRASAR (http://crasar.org/) robots are not autonomous but they are 
> leading the way toward autonomous operations. Robots from CRASAR have 
> been deployed to incidents.
>
>  After my experience with the NASA SRR, I think semi-autonomous is the 
> next big step for planetary rovers and disaster incidents. This would 
> have robots who can be assigned some tasks that are autonomous with 
> over all control done remotely. Curiostiy actually did this a few 
> months ago. It autonomously navigated through an are that was not 
> easily visible from the Earth remote controllers. For the SRR, an 
> example would be for remote control to position the rover in front of 
> the sample but the sample would be picked up autonomously.
>
> There is a long history of challenges advancing technology. Lindbergh 
> crossed the Atlantic and won a prize. The X-Prize for repeated 
> reusable space flight is another example.
>
> The SRR prize pool is $1.5 million. Collectively there has probably 
> been $500,000 spent by the teams. (That is a really wild guess.) No, I 
> did not contribute significantly to that total,  primarily because I 
> used all off-the-shelf hardware.
>
> The SRR last June was an awesome experience. The 11 teams were all 
> friendly and cooperative. We were happy to see a success by anyone no 
> matter how small it might have been. One team blew a power supply the 
> night before the challenge. Other teams contributed a supply and 
> worked with them to get it running. That team was the first to leave 
> the  and return, unfortunately empty handed but it thought it had 
> picked up a sample.
>
>
> - 73 -
> *Rud Merriam K5RUD
> * /Mystic Lake Software <http://mysticlakesoftware.com/>
> /
>
> On 12/20/2013 12:57 PM, Paul Bouchier wrote:
>> David,
>>
>> At the lowest level, I have to align with your assertion of silliness, but
>> at a higher level, I think you're ignoring a big vision and know-how aspect
>> (perhaps deliberately).
>>
>> The idea that robots should be humanoid has little merit IMHO for the
>> reasons you state, and furthermore, disasters that are appropriate for
>> application of relatively inflexible, unintelligent robots are few and far
>> between. (Japan's reactor is the only one I can really think of - even the
>> World Trade Center was done better IMHO with humans and dumb machines).
>> Investing a huge amount of money preparing for some disaster that surely
>> won't play out the way someone anticipates and that will render most robots
>> inadequate would be flagrantly wasteful, if that was all that was going on.
>>
>> One of the popular collections of economic lessons (maybe The Naked
>> Economist) has a story demonstrating that competitions with prizes are a
>> very effective way of stimulating investment in an area. I think the example
>> was rocketry, where a $1M prize stimulated $20M in investment; some teams
>> invested twice the prize value in their entry. What was really at stake was
>> knowhow and a market. The prize pulled together investment in a particular
>> goal that was considered worthy. That investment included directing the
>> research of PhD students and postgrads, from which much know-how resulted.
>>
>> In a similar way, the DARPA driving challenges stimulated research and
>> investment that was going on anyway, but focused it on a particular goal and
>> brought many teams together to work on the same goal at the same time. It
>> was a concrete expression of a vision. I'm working with some folks who
>> participated in those competitions, and it's clear that they gained a lot of
>> know-how through that activity, and are propagating it into their work
>> today.  Yes, companies and universities were working on self-driving cars
>> before and after, but there wasn't the breadth of research and academic
>> involvement that those challenges brought together and which resulted in
>> algorithms and know-how in that area. You point out that Thrun's algorithms
>> with which he won aren't in use anymore, but that's missing the point - they
>> were a stepping stone to better things - a marker against which improvements
>> could be measured. Without the focus that a competition brings, the markers
>> would likely not exist or be well known. Now look at the time lag from the
>> DARPA driving challenges (2004 - 2007) until autonomous vehicles are
>> expected as products (2020) - 15 years. And we now have a concept for how
>> the evolution will happen: gradually unloading the driver through driver
>> assist features leading to mostly autonomous operation with high-level human
>> supervision. These competitions are stimulants for long-term developments.
>>
>> You will surely argue that this would have happened anyway. Perhaps. Perhaps
>> in the same timeframe, perhaps not. But it seems incontestable that the
>> challenges brought together a lot of smart people focused on a specific
>> problem at one time, from which they, and they community at large, learned a
>> lot. Through this, I argue it was an accelerant.
>>
>> I think the present robot competition is that kind of thing. The specific
>> goals are not as concrete or focused as self-driving cars, but neither are
>> the tasks that robots working in human environments would have to do. Much
>> must be discovered and developed, and a robot that can work in a simulated
>> disaster (however hokey the scenario) will have to incorporate know-how that
>> will have much broader applicability. The current grand challenge is
>> bringing together researchers and engineers at one point in time who will go
>> forth and apply that in broader contexts, hopefully many of a peaceful
>> nature. As you would know, the national robotic roadmap calls for domestic
>> robots to be reality in the 2030's (15 - 20 years away). That's the big
>> vision requiring much know-how that I think is behind the present
>> competition.
>>
>> Great topic and stimulating discussion!
>>
>> Regards
>>
>> Paul
>>
>>
>
>
>
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://list.dprg.org/mailman/listinfo/dprglist

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.dprg.org/pipermail/dprglist/attachments/20131220/a2cb7b8b/attachment.html 

More information about the DPRG mailing list