got that covered, if you guys can quit the groucho and cheeko bits. We put the gps on a a flex dummy, cut a sunroof in the top of the jeep and hook the actuators on the dummy. So, basically hhes driving the jeep. we put some cameras in his eyes and walla , we make the race . I suggest we arm our vehicle with water sprayers to douse our competitors as we pass, so they sufer some circuitry problems. we have sort of a battlebot
Farmerman... It has to be completely automated (no remote control). From what I seen in the failed attempts, it needs to be able to continue after making mistakes. I think I got that covered if GPS will fuction through fiberglass. Do you know if it will?
no, it needs a clear up look. for best location the antenna should be to the sky.So we just screw the antenna ontop of the roof or the hood.itllll acquire some sats inside but never top performance.We need a marine radar so we dont whack into somethingr.
OK fully autonomous,, hmmmm, were also gonna need a gIS sytem , maybe we can buy a Tomahawk missile. they use a GPS/GIS ,(like a built in roadmap).
Were starting to talk about some real money here.
Farmerman: See, that's where everyone else screwed up. They mostly tried to convert 4 wheel drives and fill them with detecting equipment. I have a better plan: If the best engineering students in the country can't handle making a machine that successfully identifies obstacles, we probably can't either. So if we can't, why try? Picture an 8-foot sphere constructed of many layers of fiberglass and a Kevlar shell. Inside, sits an uncommonly beefy go-cart with a frame that reinforces the sphere via a dozen or so casters. This should provide the necessary stability to absorb an impact. With the majority of the weight being on the ground: in the event the sphere is stopped dead in its tracks the momentum of the "go-cart would simply make it spin inside of the sphere... instead of a 30 mile an hour impact. This way the G-forces would only be intense in one direction (down). If it merely hits a car it will bounce off it. A fence it should bounce over. If it hits a wall; turn 90 degrees for 20 feet, and try again. If it can't go left; go backwards 20, then left 20 and so on through preprogrammed attempted route corrections. It should handle most instructions with a simple compass since it doesn't have to pick "perfect lines". Occasionally though, it will have to access GPS to correct its coordinates. This design should prove 100 times more durable for mistakes and since it doesn't care when it makes mistakes, we don't need all that zillion-dollar equipment to avoid obstacles. What do you think?
I did wonder why I hadn't seen any entries which had "flipability". All these guys had to do is watch a few bot war shows to see the advisability of being able to go forward after flipping ass over teakettle. And every run-on-their-own windup car has the ability to back up and vector.
I think the bouncing idea has some merit but we are going to use a lot of fuel adjusting and readjusting. I think the recently developed gyro-chair with it's ability to climb and descend objects is more the way to go.
Joe
Don't forget Joe: We have 150 miles to go in 10 hours in order to qualify for the cash. That means we have to cruise at a good clip in the open stretches while being able to handle a high-speed impact.
I like the gyro-scope idea. I'm not sure if I like the spere though. Too hard to control. A
Buckyball would be easier to control, and would absorb shock.
Does it have to have wheels, could you build a hovercraft?
wHEN IN DOUBT_READ THE ARTICLE>
O.B. I see yer point, it looks like they all aftermarketed their way to destruction.It seems they all owned junkers oK, got it.
How about (a complete change of strategy).
Joe, we have a problem with flippability. Wed need an airplane engine because regular engines wont run upside down, for long .
Anew John Deere 4wd 30 hp tractor would run about 13k
Itd need some radar as well as gIS and gPS , cause wed have to be able to detect small hazards and then avoid them. Thatd all be programmed into the gIS ... gps would mostly be used for place references and it would validate the GIS autopilot. I see the problem maybe as one of sensitivity. the gps we use on our boat is tied to the radar and a chart plotter and it has an accuracy of 3 meters . In the water 3 meters is usually ok for slow speeds and open ocean , but in a marked hazard waterway, 3 meters could mean youd still pile up on a rock without the radar havin g override control. Boat navigation systems are well established and it would take a decent mechanic to hook up steering actuators and controllers to our tractor. Wed also need some better protection on our intakes and steering arms. Tractors are usually built for taking a lot of dirty air and dust in the steering arms and the engine. (They use a lot of dry knuckles and high tie bars and dust filters and high intakes).
i have an old JD 240 that I have for parts . ill put it in the mix. A decent tractor can cruise easily at 20 mph.
i still think that the answer is a fine controlling system with good crash protection afforded by the body . I think dARPA wants this as a recon and weapons platform, so the ball thing would be a rolling bomb, unless they think of this as a kamikaze weapon. (of course if its unmanned, getting back is not a prime directive0
hmmmm, many more things to think about. we need a monster truck afficienado. those things take huge hits and keep running (except when they flip, they just lay there like a beached whale)
In stead of wheels. think continuous track treads, like the very early WWI tanks where the tread went completely over the vehicle's profile. Then if it flips it simply has to revers motion and continue forward.
After a careful review of
THE RULES, After a careful reading of the rules, I suspect the sphere idea would be disqualified for lack of intelligent autonomous behavior.
darpa wrote: For example, an extremely large vehicle that simply travels on a straight line between two points by climbing over or breaking through everything in its path (and destroying what cannot support that movement) is not the type of intelligent solution that is sought. Vehicles that cannot demonstrate intelligent autonomous behavior will not be accepted as Participants.
Other portions of the rules state that a vehicle can be disqualified for "reckless behavior" (damn it). I still think the key to success will be a design that is durable enough to sustain intelligence failures. The rules are purposely vague to allow for innovative thinking. They do state that the vehicle has the option to "avoid or accommodate" obstacles... so I infer that to mean some leeway for mistakes.
The good news is:
darpa wrote:A10. DARPA will run the Grand Challenge for Autonomous Ground Vehicles approximately annually until there is a winner, or until the Congressional authority to award the cash prize expires (currently in 2007).
