1.10 How much better is the AI?

  • Thread starter Thread starter Smackinjuice
  • 51 comments
  • 6,608 views
S

Smackinjuice

Premium
Messages
82
http://community.eu.playstation.com...o-5-Version-1-10-Feature-Update/td-p/13135941

as we all know

-Improved Opponent AI Algorithm
-The search range of opponent AI has been expanded and their intelligence algorithm has been improved.

Well, I'm still waiting on my download; so anyone have any clue how much better it is yet? give us your feedback in here! as I know this was always one of the top complaints is the AI sucks.

Although mine would be a much needed 400hp/500 PP FF A-Spec race :D My 400 hp civic has nothing to do in this game...

in fact so many cars in the 300-500 hp range have either 1 or 2 races they can even do in this game. And that makes up a huge deal of the premiums! let alone that's most the cars we drive in real life just maxed out :(
 
In mathematics and computer science, an algorithm (Listeni /ˈælɡərɪðəm/) is an effective method expressed as a finite list[1] of well-defined instructions[2] for calculating a function.[3] Algorithms are used for calculation, data processing, and automated reasoning.

Starting from an initial state and initial input (perhaps null),[4] the instructions describe a computation that, when executed, will proceed through a finite [5] number of well-defined successive states, eventually producing "output"[6] and terminating at a final ending state. The transition from one state to the next is not necessarily deterministic; some algorithms, known as randomized algorithms, incorporate random input.[7]

A partial formalization of the concept began with attempts to solve the Entscheidungsproblem (the "decision problem") posed by David Hilbert in 1928. Subsequent formalizations were framed as attempts to define "effective calculability"[8] or "effective method";[9] those formalizations included the Gödel–Herbrand–Kleene recursive functions of 1930, 1934 and 1935, Alonzo Church's lambda calculus of 1936, Emil Post's "Formulation 1" of 1936, and Alan Turing's Turing machines of 1936–7 and 1939.
 
Just done the Fiat 500 one, came 2nd. Didn't notice any difference in AI, strict race restrictions make it a challenge though.
 
There is a distinct possibility that by having a race that is difficult by regulating your PP, and stating that the AI algo's have been updated, they were hoping for a bit more "wow effect". Its aaallllllllllmost as if they're trying to trick us?

You know what though, as crappy as every update has been (sans maybe one) this one isn't too bad, they just skipped out on in race saves. Thats really the only realistic updateable thing floating around thats even remotely realistic. We'll never see standards going to prems or anything like that.
 
The AI already improved during the last update, so it'll take a while to suss out the differences. Maybe, just maybe, they're not as timid as in the post 1.07 updates, but not as crash-happy as in GT5 1.0.
 
So good that auto A-spec grinding doesn't work anymore. Just tested it. At least I've got basically every car already. ^^
 
TorqueHappenS08
In mathematics and computer science, an algorithm (Listeni /ˈælɡərɪðəm/) is an effective method expressed as a finite list[1] of well-defined instructions[2] for calculating a function.[3] Algorithms are used for calculation, data processing, and automated reasoning.

Starting from an initial state and initial input (perhaps null),[4] the instructions describe a computation that, when executed, will proceed through a finite [5] number of well-defined successive states, eventually producing "output"[6] and terminating at a final ending state. The transition from one state to the next is not necessarily deterministic; some algorithms, known as randomized algorithms, incorporate random input.[7]

A partial formalization of the concept began with attempts to solve the Entscheidungsproblem (the "decision problem") posed by David Hilbert in 1928. Subsequent formalizations were framed as attempts to define "effective calculability"[8] or "effective method";[9] those formalizations included the Gödel–Herbrand–Kleene recursive functions of 1930, 1934 and 1935, Alonzo Church's lambda calculus of 1936, Emil Post's "Formulation 1" of 1936, and Alan Turing's Turing machines of 1936–7 and 1939.
Click to expand...
Thanks, my brain hurts now..
 
Sample size speaks for itself, but I just ran the 10-lap Super GT Suzuka Circuit A-Spec event, and I won going away in a stock Arta NSX. When I say I won going away, I mean I drove like an old lady on Sunday and hardly even saw any AI cars. Fell into second for a moment during lap 9 and retook the lead on the next corner.

So yeah, seemed exactly the same to me.
 
Turkey :D
So good that auto A-spec grinding doesn't work anymore. Just tested it. At least I've got basically every car already. ^^
Click to expand...
Physical contact with the walls Slows cars down more thats why its not working.
The X2010 would do 180MPH wall riding its now dropping to 160MPH with no change in set up.
like the wind is now a loss as it has Lemans cars.
Change to dreamcar indy as they are GT1 cars. less money but more exp.
 
Last edited:
NEOPARADIGM
Sample size speaks for itself, but I just ran the 10-lap Super GT Suzuka Circuit A-Spec event, and I won going away in a stock Arta NSX. When I say I won going away, I mean I drove like an old lady on Sunday and hardly even saw any AI cars. Fell into second for a moment during lap 9 and retook the lead on the next corner.

So yeah, seemed exactly the same to me.
Click to expand...

Were you using the racing hards they were using? The limited downforce with no suspension or Gear ratio tuning? Mind you they do brake early and pretty slow on the thtottle on the exit of a turn...
 
bullet360
Were you using the racing hards they were using? The limited downforce with no suspension or Gear ratio tuning? Mind you they do brake early and pretty slow on the thtottle on the exit of a turn...
Click to expand...

Well it was raining so I was using wets. Do they stick with hards even in the rain? But regardless I was still struck by just how easy it was. I hadn't golded that event yet, yet I fire it up after this "improved AI" patch and I win going away in the rain driving more or less like a noob?

For the record I don't have half the complaints others do about the AI. Just in the context of whether it's "better," I didn't notice anything.
 
Ya i think they're trying something psychological... But why idunno(conspiracy) haha but ya dude try run any of your garage cars that are under 580 hp in arcade mode... The ai is pretty good... Agreesive some points... Haha i seen an 11 car pile-up of the ai on the main stretch of tokyo (reverse) on the last lap before the start/finish... It was great... All y friends that were in the room watching were freaking out cause they never seen anything like that before... It was awesome.... But that was before the update... Maybe they dont crash anymore :(
 
Smackinjuice
Sigh, I was hoping they would stop driving like they were in a parade; just looking fancy following each other.

that's not racing PD!!!
Click to expand...

GT5s AI is crap.

I have a game called "Mercedes Benz World Racing" from 2003 and it's AI is simply perfect. Why can't they just buy the entire development team (Synetic/Germany) and learn how to do it right?

 
only difference i've noticed is that when you lap them they no longer move out of the way. They now seem completely unaware of me being there.
 
They were quick in the subaru seasonal i thought it would be an easy win because they drive like granny going to church, they drove allot faster than i thought they would but still not human fast.
 
EivlEvo
There is a distinct possibility that by having a race that is difficult by regulating your PP, and stating that the AI algo's have been updated, they were hoping for a bit more "wow effect". Its aaallllllllllmost as if they're trying to trick us?

You know what though, as crappy as every update has been (sans maybe one) this one isn't too bad, they just skipped out on in race saves. Thats really the only realistic updateable thing floating around thats even remotely realistic. We'll never see standards going to prems or anything like that.
Click to expand...

I don't think they could do in-race saves where you start right back up on the track. What they could do is have you pit, then save, and when you start back up you just exit the pits.
 
crispychicken49
I don't think they could do in-race saves where you start right back up on the track. What they could do is have you pit, then save, and when you start back up you just exit the pits.
Click to expand...

Yes but the problem here is all the other cars that are on the track. How can they pinpoint exactly where they are so when you start the race all looks the same?

It's tricky but surely is a solvable problem. I guess we just have to wait.
 
Smackinjuice
Sigh, I was hoping they would stop driving like they were in a parade; just looking fancy following each other.

that's not racing PD!!!
Click to expand...

Just once I recall a very funny incident with AI behavior in B Spec. It was in the British Lightweight series months ago. My Bob won a tight race, there was exchange of paint during the race. But afterwards, during the race results screen, I let it run. One of the AI and my Bob seriously went after each other on that straightaway that is displayed. It was not accidental! They rammed each other and pushed each other back and forth--after the race was over! It continued down that straight for as long as I could see them. Hilarious!
 
Since the last race I did before the update was the Turbo Challenge High Speed Ring, I did that one again.

The AI actually got WORSE here, as instead of taking 1 lap to overtake them in my current car I got them all in the first corner (the very high speed one after start/finish).
The AI slowed down a bit there before, but now they really went on the brakes there.
 
Joao71
Yes but the problem here is all the other cars that are on the track. How can they pinpoint exactly where they are so when you start the race all looks the same?

It's tricky but surely is a solvable problem. I guess we just have to wait.
Click to expand...

not a difficult problem. each car has a locator within the transform of the vehicle, each track consists of waypoints (locators) placed throughout the track. when the user saves. it saves out for example a save file with each car_name_01global pos X Y Z/ speed.velocity etc etc then when game resumes it loads the data for that track and places those cars with parameters intact.
 
creati
not a difficult problem. each car has a locator within the transform of the vehicle, each track consists of waypoints (locators) placed throughout the track. when the user saves. it saves out for example a save file with each car_name_01global pos X Y Z/ speed.velocity etc etc then when game resumes it loads the data for that track and places those cars with parameters intact.
Click to expand...

That should work. Load times would be a pain though, many people would complain. But I think they should do it this way.
 
Joao71
Yes but the problem here is all the other cars that are on the track. How can they pinpoint exactly where they are so when you start the race all looks the same?

It's tricky but surely is a solvable problem. I guess we just have to wait.
Click to expand...

It's not a problem as PS3 launch title Formula One CE has mid race saving option in the pause menu. If PD can't do it they should ask their colleagues at Sony UK on how to implement it.

Loading back to the race takes a couple of seconds.
 

Latest Posts

Back