Question about professionals and the Porsche GT

[sanjuro]

In the end, the process of determining the code would just be too complicated. Remember those huge computers used to study nuclear weapons? This might be equally or more complicated. One lap on a bike in a race is a stream of constantly changing data with minute decisions made without even thought involved. Enter a corner three inches over from the last time and everything changes accordingly. I'm not making out to be a superhuman sport, just a very complicated one.

Those huge computers were using analog technology back then.

Cool article and review about the Cray-1 the first really high end production supercomputer.

It cost $8 million and performed at blistering 80 MFLOP/s. For comparison, a Pentium 4 2.8ghz can hit about 2.5 GFLOP/s or about 31 times faster. The current supercomputer champ can handle 280 TFLOP/s or about 350,000 times faster.

http://blog.modernmechanix.com/1979-review-of-the-cray-1-supercomputer/

You can't say computers cannot control a motorcycle or a car because of technology limitations. There is a lot of compute power used right now:

http://articles.economictimes.indiatimes.com/2011-10-28/news/30332526_1_drivers-race-garage-pit

Don't get me wrong: I don't want to see a robot sitting on a motorcycle, or more likely, a liquid weight distribution system with mechanical arms controlling the fork. But there will be robot controlled cars on the road.

 

[DReg350]

In the end, the process of determining the code would just be too complicated. Remember those huge computers used to study nuclear weapons? This might be equally or more complicated. One lap on a bike in a race is a stream of constantly changing data with minute decisions made without even thought involved. Enter a corner three inches over from the last time and everything changes accordingly. I'm not making out to be a superhuman sport, just a very complicated one.

Well, you could make it out to be a superhuman sport and get no disagreement from me. Stating the obvious, pros have an awareness, a consciousness that processes at a WHOLE nuther level when compared to mere mortals.

Question is, will computers, sensors, controllers and power supplies ever be small enough and powerful enough to process the same inputs and outputs as fast or faster than a pro. Challenging yes, but I wonder how many thought we'd get as far as we have.

 

[afm199]

Those huge computers were using analog technology back then.



http://blog.modernmechanix.com/1979-review-of-the-cray-1-supercomputer/

You can't say computers cannot control a motorcycle or a car because of technology limitations. There is a lot of compute power used right now:

http://articles.economictimes.indiatimes.com/2011-10-28/news/30332526_1_drivers-race-garage-pit

Don't get me wrong: I don't want to see a robot sitting on a motorcycle, or more likely, a liquid weight distribution system with mechanical arms controlling the fork. But there will be robot controlled cars on the road.

The computers I am speaking about are today's. Livermore has 100,000 small computers networked to do the calculations. http://io9.com/5916320/supercomputer-simulates-nuclear-explosion-down-to-the-molecular-level

I'm not discussing autos on roads that are already mapped with GPS coordinates, have speed limits, and where safety is the first issue.

I am discussing going faster than Lorenzo or Marquez on a GP bike on a race track. That's not even the same stadium. It's another country. Get back to me when you can find a link

 

[Asphaultnaut]

The computers I am speaking about are today's. Livermore has 100,000 small computers networked to do the calculations.

I'm not discussing autos on roads that are already mapped with GPS coordinates, have speed limits, and where safety is the first issue.

I am discussing going faster than Lorenzo or Marquez on a GP bike. Get back to me when you can find a link

you do understand your objections are identical to those made prior to DARPA's Autonomous Vehicle Challenge?

Where vehicles were not given a route but a destination, and not only calculate the route on their own, but have to navigate local conditions as well, both off-road and on?

it's only a matter of time until track ready autonomy is deployed and you'll probably live to see a Chinese toy version of it knocking you over on the sidewalk some day. maybe even delivering something from Amazon LOL

 

[Yakoo752]

You do realize that most of the really fast guys came out of dirttrack racing, where no rider aids are involved, don't you?

There are no lower classes where they utilise TC, so every one out there came from a class with no aids. If they're at the MotoGp level, they are extremely good riders, very, very high level of skill, talent and balls.

The aids just make things a bit safer for them, they're still drifting them sideways, they're still the best riders in the world.

Stoner did 8 years in MX
Rossi did karts
Marquez has 2 years in MX
Colin has 10 years in MX
Spies started CMRA on a ysr80
Simoncelli did minimoto
Hayden did GNC but started CMRA (only one in 10 years that came out of dirttrack racing)
Tony Elias raced scooters (granted his dad was a 10x mx champion)
Pedrosa started on minis
Roberts had 1 year in MX
that goes back 10 years of champions and includes some other talent... with only 1 champion having quite a bunch of years in the dirt.

yes all forms without traction control

I am not arguing that these guys aren't the best at what they do and I'm not saying Doohan, Rainey, or Lawson could come in and do it better. I'm just not a fan of their rider aids is all.

and I'm not arguing your opinion vs mine. My opinion is the rider aids have lowered the excitement of MotoGP. I understand this to be prototype racing and I think everyone is currently invested in rider aids/electronics. I personally would much rather watch BSB or an LMP race (if we're really talking prototypes).

 

[Abyss]

Cos you have money and can buy whatever you want to satisfy your juvenile delusions.

 

[Ducky_Fresh]

Jeremey Clarkson may have driven a whole lot of cars and therefore have a broad understanding of the Carrera GT but he's a bit of a ham fisted driver. A better judge of this car is someone intimately connected with its development and a few world rally championships under his belt.

Walter Rohl, Porsche's own test driver was scared of this car at the 'ring. He's about as bad assed as it gets. IIRC the early cars came without TC and he convinced the engineers to at least add a minimal version but they fought him on it.

Not only do these cars have an ungodly amount of power, both torque and HP but Porsche had to commission Michelin to make special tires just for the car. They're still considered not enough by the engineers. Itll break the tires loose with very little provocation and with so much power on tap the tires don't just break loose, they try to come around on the driver.

Being a mid engined car, once a spin is past a certain point, the car keeps going in circles and the driver is now a passenger. It's a strange charastic of mid-engined cars. It's called a low polar moment. There is no mass outside the wheel base for the car to right itself around. The best analogy I can come up with is when a figure skater starts a spin and gradually pulls her arms in and her rotation gets faster until she extends her arms o slow it down. A mid engined car can't recover from a spin very easily. It's just turns all it forward momentum into a spin and until the tires have scrubbed off enough energy, it keeps spinning.

 

[tzrider]

Take a look at 600 times VS 1K times in local racing…most races we're at, the top 600's would podium in the 1K classes. It's not about HP to get to the limit of a bike...

Years ago, moto magazine testers noticed a similar thing. We've found that on average, riders of high powered bikes tend to open the throttle a little later than those of lower powered bikes. It suggests they're leery of getting on the gas too soon.

 

[Holeshot]

Interesting Andy. They also seem to enter a corner with too much speed and thus, can't change direction as fast, which becomes a never ending circle.

 

[afm199]

for nerds and tech heads, yeah, they'll watch. But the point you were making was that it can't be done..not that it wasn't marketable or entertaining.

More that the cost and complexity would be enough to fund social security for a year or three. Livermore is spending fortunes to build a computer network that simulates nuclear explosions and it's not really a success yet. And that is a very simple process, in the sense that almost all of the variables, effects and fixed properties are known. Fuck they got fission with slide rules.

you do understand your objections are identical to those made prior to DARPA's Autonomous Vehicle Challenge?

Where vehicles were not given a route but a destination, and not only calculate the route on their own, but have to navigate local conditions as well, both off-road and on?

it's only a matter of time until track ready autonomy is deployed and you'll probably live to see a Chinese toy version of it knocking you over on the sidewalk some day. maybe even delivering something from Amazon LOL

Not even remotely. A Roomba fulfills most of those functions.

 

[UDRider]

Basically what your premise would entail would be a construct capable of computation AND decision making at the level of the .15 quadrillion synapses of the human brain. That's what seems to be missed here. Pardon me if I don't expect to see this in my lifetime and doubtfully in yours.

The top racers ( or even the slow trackday guys) are making huge numbers of decisions and computations simultaneously while riding on the track. Frankly the idea that a robot is going to be built to mimic this tells me someone is not considering the sheer complexity of the process. In order to program a robot to mimic human behavior, you have to be able to program the human behavior. The incredibly complexity of that is the problem.

I have no doubt that TODAY a robot could be built to run in a straight line faster than a human. That's one set of fairly simple movements. Racing introduces a level of difficulty several degrees of magnitude higher.

If we can get beyond silicon there is a good chance it would be possible in my lifetime. That's the key. We are at 14nm, now. Can't shrink much further, even with fancy shmancy tri-gates. A lot of brain power is dedicated to interpreting signals and controlling motor functions. Not to mention although brain i fast it's not particularly efficient and not all of it is dedicated to tasks required to ride and race bikes.

 

[frozenuts]

More that the cost and complexity would be enough to fund social security for a year or three. Livermore is spending fortunes to build a computer network that simulates nuclear explosions and it's not really a success yet. And that is a very simple process, in the sense that almost all of the variables, effects and fixed properties are known. Fuck they got fission with slide rules.



Not even remotely. A Roomba fulfills most of those functions.

But can it take Homie for a walk?

 

[afm199]

But can it take Homie for a walk?

Homie would follow a Roomba to the moon!

 

[aminalmutha]

Was told by an acquaintance yesterday that he knows a guy who was close friends with the guy driving the Porsche GT in question and had actually ridden with him in siad car thru the same area at speeds of over 150mph. He said they did that kinda shit all the time.

I know this is total second hand info, but I have no reason to not believe him, tho I don't think anyone is really wondering what happened.


What I don't understand is that if you have that kinda money, WTF are you doing pulling that kinda crap in that area? I'd have my own tractor trailer and would be taking the thing to B-Willow, Willow Springs, etc. whenever I wanted. Not showing off to fools in Honda Civics.

 

[fawndog]

Stoner did 8 years in MX
Rossi did karts
Marquez has 2 years in MX
Colin has 10 years in MX
Spies started CMRA on a ysr80
Simoncelli did minimoto
Hayden did GNC but started CMRA (only one in 10 years that came out of dirttrack racing)
Tony Elias raced scooters (granted his dad was a 10x mx champion)
Pedrosa started on minis
Roberts had 1 year in MX
that goes back 10 years of champions and includes some other talent... with only 1 champion having quite a bunch of years in the dirt.

yes all forms without traction control

I am not arguing that these guys aren't the best at what they do and I'm not saying Doohan, Rainey, or Lawson could come in and do it better. I'm just not a fan of their rider aids is all.

and I'm not arguing your opinion vs mine. My opinion is the rider aids have lowered the excitement of MotoGP. I understand this to be prototype racing and I think everyone is currently invested in rider aids/electronics. I personally would much rather watch BSB or an LMP race (if we're really talking prototypes).

Sometimes the race is won in the pit, I gotta give the crews thier due. Electronicly, mechanically, aerodynamically. More so in F1 than GP.

 

[SM610]

Sometimes the race is won in the pit, I gotta give the crews thier due. Electronicly, mechanically, aerodynamically. More so in F1 than GP.

Yea, that's why MotoGP is a bit more interresting to me than formula1. Technically, the cars are amazing and that interests me. The racing is less so because it seems to have little dicing and slicing, save for the first 2 laps.

 

[morthrane]

As to robots racing . People go to races to see people race. Do you seriously propose sport will become robotic?

We sorta already have that in MGP vs CRT...

A spec moto series that allowed for robotic or human rider could be cool, so long as there's a networking airgap between the robot and the moto.

You mentioned the complexity of one track turn, but someone will start breaking down each aspect for a computer to analyze, then to react.

Eventually there will be a machine that will be able to perform on these calculations.

You're making light of the sheer complexity of one turn at physical asymptotic limits. There's no doubt that with enough time, brute force analysis and calculation can be done. To do it in real time... that's the mother of non-trivial right there.

Try this as an exercise: think about how many sensory inputs a human uses on a racing motorcycle. A computer will have to make sense of all them and coherently make real time snap judgements based on prior experience. Onboard processing power is by nature limited, and Siri-like benefits might not work out so well with the inherent lag in remote communications.

It probably doesn't help much that AI is kinda limited in some respects... because it will only be as intelligent as its creators.

With ONE totally programmed lap, with no changes, you could possibly set up a faster lap than a human could by crashing a lot and storing the info.

Isn't that what human racers do? There seems to be a strong correlation between great fast racers and how much crashing they do early in their career.

you do understand your objections are identical to those made prior to DARPA's Autonomous Vehicle Challenge?

Where vehicles were not given a route but a destination, and not only calculate the route on their own, but have to navigate local conditions as well, both off-road and on?

it's only a matter of time until track ready autonomy is deployed and you'll probably live to see a Chinese toy version of it knocking you over on the sidewalk some day. maybe even delivering something from Amazon LOL

DARPA AVC is hardly motorcycle racing.

Those contests, drones, google cars, and other such robotic devices have left no doubt that it can be done on the mild side. Whether they can hack it at competition/performance levels, still untested.

that goes back 10 years of champions and includes some other talent... with only 1 champion having quite a bunch of years in the dirt.

yes all forms without traction control

What I wonder: what happens when growing new generations of riders start with traction control and never learn manual mechanical control?

 

[ohio]

Not sure how this thread got so derailed, but I love a good nerd out opportunity...

In a theoretical future, with infinite processing power and infinite speed, infinite software engineers, and perfect sensing technology AND perfect servo mechanism technology, all that weighs less than a human, an android could beat a human around a track.

Presently? Not a fucking chance, not with an infinite budget and the most advanced aerospace or military tech.

The human brain as we know it has the processing power of about 80,000 modern processors and processing speeds about 2500 times faster than that system. Additionally, there is evidence that we are underestimating the processing power of the human brain by orders of magnitude.

The human subconscious and reflex system, especially on the superhumans in motoGP, has a capacity to integrate a massive volume of incredibly subtle inputs and generate an equally complex collection of outputs that we are only beginning to understand. To imagine that an electromechanical system would be able to "feel" the clipons, tank, seat, and pegs with the same subtlety, process that information properly and at equal speed, and be able to respond properly with inputs to bar, tank, pegs and weight distribution is fantasy. A computer beating a human in Forza 4 is NOT the same as a computer beating a human in a real car on a real track (witness the Audi autonomous pikes peak car... pretty good, but not yet human), and DEFINITELY not the same as a motorcycle on which the rider is an integral part of the vehicle dynamics.

There are things that a computer can do better, right now, which is why the fastest way around the track is a human on a bike with pretty considerable rider aids, but those things are discrete and relatively simple and coarse, compared to what it takes to replicate the feel and response of a human fingertip.

 

[sanjuro]

Was told by an acquaintance yesterday that he knows a guy who was close friends with the guy driving the Porsche GT in question and had actually ridden with him in siad car thru the same area at speeds of over 150mph. He said they did that kinda shit all the time.

I know this is total second hand info, but I have no reason to not believe him, tho I don't think anyone is really wondering what happened.


What I don't understand is that if you have that kinda money, WTF are you doing pulling that kinda crap in that area? I'd have my own tractor trailer and would be taking the thing to B-Willow, Willow Springs, etc. whenever I wanted. Not showing off to fools in Honda Civics.

Because you can.

 

[sanjuro]

You're making light of the sheer complexity of one turn at physical asymptotic limits. There's no doubt that with enough time, brute force analysis and calculation can be done. To do it in real time... that's the mother of non-trivial right there.

Try this as an exercise: think about how many sensory inputs a human uses on a racing motorcycle. A computer will have to make sense of all them and coherently make real time snap judgements based on prior experience. Onboard processing power is by nature limited, and Siri-like benefits might not work out so well with the inherent lag in remote communications.

I feel like one of those futurists who predicts flying cars and telepathy without providing one bit of science.

Right now, there is real-time telemetry being analyzed, from speed and gps location, to engine and suspension performance.

That data will translate into onboard adjustments, likely in 5 years, assuming there is no rule changes preventing this.

In the meanwhile, several sources are developing self driving vehicles. The last DARPA challenge, 6 vehicles were able to travel 55 miles on a closed course. http://en.wikipedia.org/wiki/DARPA_Grand_Challenge_(2007)

My guess is with 10 years, you will see self-driving cars on the road.

If there was no path to the robot racer, then I would be "no f'ing way". But at least we can see the start line.