I was comparing how different cars (of the original GPL cars, at least by now) behave when driven by the AI. I tested it in different tracks after copying the Jim Clark "AI" to all drivers and reducing all that I could the random features of the AI, as suggested by Lee Bowden in his AI tutorial. Following him, I also reduced the number of drivers to allow them to drive with free track. I left only one car of each brand and put them to qualify for one hour (obviously with accelerated time).
Well, my main and original question was if is there something of the track layouts that GPL takes into account to, together with features of the chassis, "calculate" how quick every car will be. This seemed to be that way, according to this table (for some tracks only)
The numbers are seconds slower that each car will be relative to the fastest car for each track in a race of 45 minutes. As you may see, in Rafaela and Avus, the AI Eagle is faster than the Lotus, and numbers of the table in general seem reasonable.
However, there are some numbers that are strange for me. For instance in Reims the AI Lotus is (marginally) faster than the Eagle and in Balcarce and The Moon (not listed here), the BRM is much slower than the Honda and Cooper.
So, my questions remain the same. Does somebody know what determines the relative speed of each AI car on every track? Is there something beyond the features of each car and (I guess) some data from the .lp files determining the speed of each car? (here I note that the .lp files are from just one car and that relative laptimes at Alta Gracia are reasonable even with my rudimentary .lp files).
Well, thanks for any clue. And I'm sorry if it was discussed in some other thread (that I didn't found).
Edited by mcmirande, Jun 23 2020 - 09:26 PM.