What Would The Theoretical Max Time Be?

  • Thread starter Thread starter don-sf
  • 14 comments
  • 3,154 views
I would imagine very close to the 1'35:000 mark. DHolland is very close to the theoretical maximum, and I believe he said he lost a tenth or two on one of the sectors in his 1'35.453 lap? And assuming he is also losing a couple of hundereths of a second on each of the 14 corners anyway (as it is near impossible to do a corner completely perfectly), the theoretical maximum would definitely be close to 1'35:0.
 
I personally believe a 1'35.000, or around that, is not possible, the lap would need to be cm perfect, and with this difficult to drive car, I think it's beyond the capabilities of a human, a very very good lap would yield a low 1'35.3, anything in the 1'35.2 would be utterly magnificent
 
i think it is the 5th corner... after the first esses and the second and third right... the long left hander i catch alittle on dhollands lap time so definitely room for improvment there. He brakes to softly and doesnt carry a great deal of speed. But his speed through the first 2 turns and the last few turns is just amazing. I would imagine a 1.35 flat should be possible... but it would have to be the perfect lap
 
lol 1 minute 35, that seems absolutely impossible...

couldnt you set up a computer software to do some sort of fastest lap?...
 
The 'perfect' lap is always something I have pondered when doing time trials. I wonder if it's possible for someone to punch in some numbers into a computer and calculate the best possible time (i.e., the lowest resulting time that would be achieved by any combination of acceleration, braking and turning).
 
The 'perfect' lap is always something I have pondered when doing time trials. I wonder if it's possible for someone to punch in some numbers into a computer and calculate the best possible time (i.e., the lowest resulting time that would be achieved by any combination of acceleration, braking and turning).

sounds like some kind of 'killer' AI in the making!
 
The 'perfect' lap is always something I have pondered when doing time trials. I wonder if it's possible for someone to punch in some numbers into a computer and calculate the best possible time (i.e., the lowest resulting time that would be achieved by any combination of acceleration, braking and turning).

That would require so much data, maths and algorithms that I think that is almost as impossible as the lap itself!
In motorsports, the engineers can only make predicted laptimes from the use of existing data - previous laps with similar cars, estimations on fuel effects, aero, etc. They could not work it out solely on the physics, or at least, it would be so massively complicated.

I guess its "possible" in the sense of yes, maths can work out anything. But the sheer amount of maths involved is a task I doubt anyone would complete.

If you just think, first you need to know the exact specifications of the car in question. Then you need to know how it reacts in the real world, how it handles, the torque it produces, etc. Then you need to know how that changes for different surfaces, different atmospheric conditions, etc. Then you need to find out what is the "perfect" line according to that data on the car's performance. This is where it becomes increasingly impossible to put into any kind of accurate mathematical quantity - because what constitutes a "perfect" line? It depends to an extent on the driver's ability and the line you can take depends on the setup of the car, the conditions and the relationship with the previous and next corner. There are just so many variables and so much external knowledge.

I would conclude its practically impossible to create such a system, its far too complex and would require such a vast amount of time to gather the data and work out how it relates to each other.
It would make a great case study for a university paper though. 👍
 
I would imagine that P.D do have an 'almost perfect' AI map for gran turismo. Its considerably less complicated than calculating it in real life.
 
It won't be an optimum line though, not mathematically. It will be a racing line that most drivers use at the circuits and it will be the same line no matter the car. This is partly why racing game AI will never be brilliant and are no substitute for real players.
For example, the line the AI takes at Fuji will be the common line most drivers use in SuperGT or whatever, but it won't be the best line for everything car combination and the AI is part of game balancing, so it will be adjusted for difficulty purposes or for improving the racing action (such as the "mistakes").
As we know, there are multiple lines around Fuji, so the line the AI takes is not necessarily the perfect line and hence not the perfect lap.
In other words, the AI lines are the same as how you would be taught the racing line in real life - from experience, not maths. (even though the AI algorithm works by maths, the line it is given is taken from what the programmer says is the best line, not what the AI works out).
 
it could be a case of some software to run laps trying ever so slightly different lines and throttle/brake combinations each time to gather a lot of data, then spit out the best combinations from those results. As already noted though, this can only give you the best possible time for the software's environment.

The only way is up.....
 

Latest Posts

Back