Ai pit stops

[Stevie_007]

Hi, can someone explain if the ai have pit stratergies. I setup a 25 lap race and setup my pit strategy to one pit by only putting in enough fuel to complete 50% the ai didn't pit once.

 

[Gringo]

It is possible to assign AI strategy. Keith Windsor has a good write up on this for SCE, and the same should apply to AMS.

http://www.racedepartment.com/threads/ai-pit-stop-strategies.68045/

 

[pitkin]

Problem with this is, you always know which driver will stop on which lap.
I wonder if Reiza could employ some randomisation to this feature as the track is loaded?

 

[Lorencini]

By default, AI starts with full tank and only stops when the fuel tank is empty. Said that, there is some variation in the pit lap (most AI cars stops in the same lap, but there is a few that stops 1 lap before or after).

To simulate pitstops in single race, i usually put a x4 or x5 fuel consumption and start a race with full tank. The amount of stops will depend on the number of laps and fuel consumption setting.
Note: AI does not bother about tyres, and dont seem to lose performance due to tyre degradation. So, dont spend too much time on track to the point of you lose much grip, cause you will find yourself too much slow comparing to the AI. Keep tyre consumption in normal and use fuel consumption to regulate the stint duration.

 

[Jarno Zurev]

Here's an example extracted from a results file found in the Automobilista\userdata\LOG\Results folder. It was an 18-lap race at "Buenos Aires Circuito No 15" and I had the race length set to 25% and the fuel and tire consumption set to x4.

You can see that they each stopped twice and seemed to follow a pit strategy not dictated by me editing any files. (Apparently, "only server .xml files contain fuel & tire data", so I have no tire data.)

To zoom in on one example, I've included Barrichello's lap data that shows his fuel usage per lap. At his first pit stop, he still had nearly 12 gallons in the tank and at the second stop, he had a little over 8 gallons. None of the AI drivers were ever in danger of running out of fuel and none of them started with more than 87.5% of a full tank.

In any case, whatever the problem was with AI pit-stops in SCE seems to be fixed in AMS. Maybe the original problem was just that the 25-lap race could be completed on a full tank at x1 consumption, so that's what the AI did.

.

 

[Lorencini]

Nice research, thanks for your time. Your findings kinda confirm my empirical conclusions.
Looks like if the maximum fuel tank capacity is enough for race completion, the AI will not stop. Otherwise, the AI will make the minimum amount of pit stops required to complete the race, based on the tank and the total lap count.

 

[Stevie_007]

Thanks for your responses - i ran another 25 lap race with x4 fuel and the ai came in around 8 laps.

 

[mkel1966]

I always edit the .gdb file to add my own pit strategies for the ai but after the last few updates it's not being followed that the ai goes to pit way early on a 35 minutes race. I tried the suggestion here on using "fuel usage = x4" on Copa Montana in Cascavel and we all pitted 3x. Also good thing about this is you don't know when the ai is going to pit. I lost pole position after the last pit(lost about 2 seconds when I overshoot on my pit mark) and ended in 2nd place but I did enjoy the race more. Thanks for the tip.

 

[mkel1966]

Hello again guys..how do you set up cars that cannot be re-fueled?
I did a 35mins race with Formula Retro at Jacarepagua with fuel usage=x3(25 laps) and tire wear=x4/x5. I pitted 3/4 times but the ai did not pit at all.