This should prove a laugh.
DARPA applied a firm hand when narrowing down the list of robot vehicles that can take place in the weekend's $3.5m Urban Challenge. The organization - an arm of the US Defense Department - has deemed only 11 out of 35 competing teams as worthy to take place in the final event. The teams have spent the past week showing off …
Doing those is difficult for us mere mortals. If a robot can do it, I'll congratulate it with open arms.
For the next test, we sit the cars down in some REAL locale. I'd pick: San Francisco (lots of hills and traffic), London England (does it have "the knowledge"?). Rome, Italy (I was there this summer, and on foot I couldn't find my way "The Trevi fountain is which way?"), maybe Paris France.
My bet is on the Stanford team (they did the Mojave desert thing) to work it out. Nice team, had a chance to see the car after the run.
Does this include gunfire and that annoying "hopping" action by the front suspension as seen on many rap videos?
If they ever produce an anitpodean module for the robot, they need to include a mannequin that will lean out of the car and yell "phwoaaarrrr" at any women of breeding age and a < 20 BMI.
This will be funny to watch, I'm just waiting for the cardboard cutout meant to represent little suzy on her way to get some sweets to get mown down by one of them!
Well aint that the way, I find myself agreeing with Toms comments above. Instead of designing these machines to operate in ANY environment, they'll have been especially designed to compete on this course, and no other.
The reasoning for this - the premise of the challenge is about the first team to come up with a system that completes the course - all for a big cash prize.
Well if I were designing something like this for moolah, I'd make damned sure it gets around this course better than any other team, the only way of realistically doing this is to design it to specialise instead of generalise.
Meaning that all the research that comes out of this will be for nothing - that is unless the enemy (in the eyes of DARPA) lives in an EXACT replica of their challenge course!
But then again I could be wrong - but to prove the point, I sure hope DARPA change some of the 'parameters' of the course secretly before it begins.
I recall that one of the teams, during the previous desert challenge, actually programmed in a complete, high accuracy path for their robot to take, as given to them shortly before the course began - instead of letting the car figure it out with road network processing and only a general set of nav points.
Not only that - what the hell is it with all the cameras, radar, laser, gps sensors etc??? If they really were 'intelligent' then all they'd need is a pair of eyes (normal visible spectrum only), and maybe a GPS satnav (in my case).
DARPA should introduce this restriction in a following challenge, after things improve significantly.
I'm a Brit, but I am given to understand that Paris, France represents a more difficult challenge than Paris, Texas (AI vehicles probably don't need to swerve to avoid tumbleweed) and that Rome, NY is probably far too boring (traffic wise) than the capital city of Italy. The latter does, of course, represent Stage IV of the DARPA challenge (vehicles to ignore all traffic controls but stop dead on a dime as una bella ragazza crosses the road NOT at a crossing place).
Obviously, you haven't seen Ashley drive. I feel relatively confident that the expansion of 'merging' to 'merging with oncoming traffic' was merely due to the extreme driving skills of the author, and nothing to do with the actual contest, per say.
Of course, if I'm wrong, it will be a *very* interesting race to watch. :)
The robot software is actually very generalized. CMU's efforts (massively planning after being given the course data) at the 2005 challenge still didn't win them the race. Stanford (the eventual winner)'s implementation for the 2005 challenge was extremely general, and Stanford didn't have to do any data entry after receiving the race data.
I'm not sure where you get the idea that they've been optimized for only this course. The announcement of its general location only happened in September (before that, all teams knew was "somewhere in the southwest USA". Teams will only receive their 'mission' very shortly before the start on Saturday. Yes, it is possible that the winning team will only work well on the DARPA course and will fail elsewhere, but it is much more likely that the winning team's vehicle has fairly generalized sensing and driving skills.
As far as unethical things to do to win $2 million, I'd say it'd be much more likely for a team to be controlling their bot remotely from a substantial distance away. Hopefully DARPA has some countermeasures in place against a scheme like that.
Biting the hand that feeds IT © 1998–2019