List of good AI & AI testing.

Discuss the new AI features ("NoAI") introduced into OpenTTD 0.7, allowing you to implement custom AIs, and the new Game Scripts available in OpenTTD 1.2 and higher.

Moderator: OpenTTD Developers

User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

List of good AI & AI testing.

Post by Redirect Left »

Update / tldr; here is the page of all AI tested and results, for anyone just wanting to find a known good one - click me!

Well hello there!
OpenTTD AI's, some of them are great, some of them (purposefully) suck, and some of them just suck and crash. Ultimately though I've spent the past few days testing a load of AI's on stock OpenTTD 13.0, with no GRFs and generic map settings.
The game was then played on full speed, whilst I only had some basic passenger trains to keep my company from dying. The only stops was when the AI crashed, i then carried on the game without resetting the AI, as for me a single crash means the AI failed, it sucks, 'literally unplayable' as gamers say.

Most of the AIs i saw and tested were from developers who are no longer around, even the better ones (the AI) have authors who (Based on last Github commit, or last forum usage) have moved on to other things. This leaves us with a lot of AIs that are either broken, or will work for a while and then crash with errors, usually the crashes are because it used too much CPU time (...i mean i have 24 cores an on i9-12900k, I'm sure you'd have found time somewhere, i'm not going anywhere), or it had an error in an array (most of these I've not debugged specifically to date), some of them also are unstable if you do anything a bit too interesting, such as build a town within the game, which is a perfectly not-cheats only thing you can do in single player, but at least one AI crashed at this, whilst some require very precise settings, or something basic (like airplanes) to not'be disabled. The other big issue is, how long to leave an AI before marking it as abandoned and a free for all (again where licences allow) for updating by anyone who wants to commit time to an otherwise good AI with a repairable bug?
Some AIs seem to be basically indestructable, even if they haven't got an update in quite some time. So I'm not sure what causes AI instability, is it OpenTTD updates that alter behaviour that need adapting to in AI code, or does it suggest the AIs in question were always a bit prone to crashing? For me the most crushing part (of this game i wasn't playing, only fast-forwarding anyway), some AI's survived a loooong time, but then randomly errored, which of course forces a reset and it undoes its beautiful massive network, which may be crushing for any long term players of games who use AI.

So, why did I make a thread? Good question! As weeks go by, the AI situation seems to be... slowly getting worse. Because of long-term illness I now find myself at home, a lot. So I'm considering taking on an AI as a project, taking a working but perhaps slightly flawed AI that has been long abandoned (as far as we can tell) by its original author, but is still actively used by people. What is the protocol for taking on an AI like that and just patching it, keeping it ticking over, fixing bugs. Is it a case of just cloning things (where licences allow, of course) and taking it on, or do I need to fork it as a specifically separate thing, with a new name in BaNaNaS which also means people who want to keep updates need to know that has happened, and abandon any current games with a ticking AI timebomb (potentially), and start a new one, with other stuff.

I can also fairly easily make a list, on a web page (if orudge still does it, maybe a page on TT-F domain?) or forum listing all the AI's (...a lot of spare time as I said), and their known stability with the current version, plus maybe one or two columns noting known issues (like creating new towns as i mentioned above),and known things that will cause it to crash, and how compatible it is with GRFs and other such, if at all. I feel it's also needed that there is a more simpler chart in-game when you click on AI.

Other thoughts, but definitely not doable by me. Is it possible to make AIs checkpoint themselves every so often. Then if a script errors, it can be, either automatically or otherwise, reset back to the state it was in at that checkpoint, tiles, units and all. Although I suspect in this state, it wouldn't work because the script will, unless they've got a bit of random to it, just make the same decisions on the same situations and crash again? I dunno, just a thought to try making a script error not as fatal.

Lastly of course, is it worth me doing any of this, keeping some abandoned AI's updated to some degree, do people actually use the AI? I feel the way it is now, AIs are a bit of a scary prospect for new players as you don't know precisely what you're getting into.

On a different note, is it possible to make links on the script output clickable? Some AI's have links like this to report bugs, but making them non-clickable doesn't encourage reports and therefore repairs.
2023-03-15 01_53_43-OpenTTD 13.0.png
2023-03-15 01_53_43-OpenTTD 13.0.png (21.99 KiB) Viewed 7634 times
Last edited by Redirect Left on 31 May 2023 12:45, edited 5 times in total.
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
User avatar
jfs
Tycoon
Tycoon
Posts: 1771
Joined: 08 Jan 2003 23:09
Location: Denmark

Re: AI Thoughts, listing & Upkeeping of them

Post by jfs »

There has been a bit of talk on the developer chat (IRC and Discord) recently about making the AI/GS developer console more useful in various ways.

Reasons that AIs fail are probably often subtle changes to game behaviour, that cause assumptions the developer made (consciously or unconsciously) to no longer be true.
On the other hand, the "too much CPU usage" is entirely a configuration thing, and I think it should be possible to disable that. Like nearly everything else in OpenTTD, AI and GS execution is single-threaded, so any number of CPU cores won't help performance at all.
User avatar
odisseus
Director
Director
Posts: 580
Joined: 01 Nov 2017 21:19

Re: AI Thoughts, listing & Upkeeping of them

Post by odisseus »

I have a feeling that most of the old AIs are not worth maintaining. They had been created by amateurs who wanted to learn something new and try out a few things, and were duly abandoned when the author lost interest. Such AIs should be hidden from the add-on list, so that the user can download them only if he already has a saved game that requires them.

That said, some AIs are worth keeping even though they crash in some situations. I think there is a common cause of these crashes, namely incorrect assumptions and insufficient testing. For example, one author of an otherwise good AI had assumed that the list of towns will never change. Others had been assuming that a certain function will never return null, or that CPU time will not be limited. (By the way, OpenTTD is unable to utilise multiple CPU cores.) To safeguard against such crashes, one has to adopt a defensive style of programming and a thorough testing routine, but this would turn AI development into an ordeal rather than a fun challenge. Furthermore, some errors are just impossible to predict at the time of development, as the game API will inevitably change over the years.

To sum up, some of the broken AIs have more potential than others, and a ranking that arranges them by stability and strength would indeed be quite useful. If you indeed have the time and desire, by all means go ahead and do the great work.
AAAHogEx
Engineer
Engineer
Posts: 119
Joined: 30 Jan 2022 15:37
Location: Japan

Re: AI Thoughts, listing & Upkeeping of them

Post by AAAHogEx »

I too think that ease of reporting is necessary. Right now openttd has so many settings and newgrfs that it is difficult to do a test that covers everything. If there was a one-button upload of logs and save data when the AI crashes, users would be more likely to report it, and it would be easier to improve quality than it is now. Also, with such a feature, we might be able to see which AIs are prone to crashes. ?(
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

jfs wrote: 15 Mar 2023 08:46On the other hand, the "too much CPU usage" is entirely a configuration thing
I'm not sure how it is calculated either. I found an AI that reliably crashed on demand at a specific time, with that error. Knowing this, I stared at a very busy game (150 years+ of just AIs fighting it out), there was zero slowdown to the game immediately before the crash was displayed with the CPU usage, even in this busy situation. I had presumed that it was a warning that this AI was using way too much CPU for too long, to the point it was slowing down either the game, or the entire system, in my case this was 100% not what happened. But it may be due to my setup, with my i9 pushing 5.2GHz on a specific core if it thinks it needs to, so those using much weaker systems i guess its feasible it does make an issue for them? I am guessing it is meant to protect against AIs that got stuck in a loop, but it's also viable that they're just struggling to finish waypointing and pathfinding because the map is bigger, or full of massive mountains, or much busier than it was as the start of the game, and most cities are fully kitted out with stations, I'm using entirely vanilla settings, because no new users want to fart about with settings for basic stuff, it isn't a good look, so I'm hesitant to change the configurations.

BaNaNaS is an open system, which for the most part is good. However, I do feel some AI's really need a bit of a barrier to entry, given how much they can add, and how much they can... break, thankfully they only break themselves, but some of them are borderline unplayable currently, either crashing or stalling immediately, or almost certainly crashing within 5 years. I will give the benefit of the doubt and say they were perfectly fine at one point, however i do feel there should be a checklist of requirements and points to meet before an AI is available to download, and some guarantee it can handle any reasonable (lets say 'non cheat, and non in-game GRF changes' for example) circumstances. There also needs to be a way for users to flag an AI (or anything else on BaNaNaS i guess?) as broken, unusable, or any other reason you'd want to report something. I found no button to do this within the content system in-game. I can see some AIs being bad enough if a new player tries to use the wrong sequence of them and constantly hitting walls, they may just give up on the entire thing, which I never want to risk happening. Perhaps there should be a way of automating the removal of AI (and maybe GRF's as a whole?) if the user doesn't log into the BaNaNaS platform within a period of time. If a user doesn't log in near this time, send a warning email so they have a chance to say "yup, I'm still paying attention" a few days or weeks beforehand to be fair.
AAAHogEx wrote: 15 Mar 2023 12:13 I too think that ease of reporting is necessary. Right now openttd has so many settings and newgrfs that it is difficult to do a test that covers everything. If there was a one-button upload of logs and save data when the AI crashes, users would be more likely to report it, and it would be easier to improve quality than it is now. Also, with such a feature, we might be able to see which AIs are prone to crashes. ?(
This is definitely true, it's why i suggested link detection & clickability in the output. Granted, most AIs don't throw a link to report to, but it'd be super helpful to just have that for the ones that do, your idea of having an entire log & save data is also fantastic, although I'm not sure how you (or an AI creator) would define where precisely that log & save game needs throwing to.
I've also just noticed that the "Detailed NewGRF Info" box when you click on a GRF also lacks the ability to click on links, and some GRFs don't enable the "visit website" button, only list a website in the description, so that's a bummer too there.
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
AAAHogEx
Engineer
Engineer
Posts: 119
Joined: 30 Jan 2022 15:37
Location: Japan

Re: AI Thoughts, listings & Upkeeping of them

Post by AAAHogEx »

Redirect Left wrote: 15 Mar 2023 16:12 This is definitely true, it's why i suggested link detection & clickability in the output. Granted, most AIs don't throw a link to report to, but it'd be super helpful to just have that for the ones that do, your idea of having an entire log & save data is also fantastic, although I'm not sure how you (or an AI creator) would define where precisely that log & save game needs throwing to.
I've also just noticed that the "Detailed NewGRF Info" box when you click on a GRF also lacks the ability to click on links, and some GRFs don't enable the "visit website" button, only list a website in the description, so that's a bummer too there.
Ideally, reports would be uploaded to a server such as bananas and associated with a corresponding version. The forum is a bit of a hurdle for users, as it requires the creation of an account to post. However, just making the link clickable from the crash screen would be a good idea, as it would likely increase the number of people reporting crashes.
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

AAAHogEx wrote: 15 Mar 2023 17:12 Ideally, reports would be uploaded to a server such as bananas and associated with a corresponding version.
That seems a good plan. Shame it'd be effort to do BaNaNaS side reporting and accessible reports for not just the admins, but also the author of each newgrf to see any reports of their own entry, whether its a glitch in a scenario, NewGRF, or misbehaving AI behaviour.
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
HGus
Engineer
Engineer
Posts: 123
Joined: 12 May 2013 22:28
Location: Argentina

Re: AI Thoughts, listings & Upkeeping of them

Post by HGus »

Redirect Left wrote: 15 Mar 2023 02:08 [...]

So, why did I make a thread? Good question! As weeks go by, the AI situation seems to be... slowly getting worse. Because of long-term illness I now find myself at home, a lot. So I'm considering taking on an AI as a project, taking a working but perhaps slightly flawed AI that has been long abandoned (as far as we can tell) by its original author, but is still actively used by people. What is the protocol for taking on an AI like that and just patching it, keeping it ticking over, fixing bugs. Is it a case of just cloning things (where licences allow, of course) and taking it on, or do I need to fork it as a specifically separate thing, with a new name in BaNaNaS which also means people who want to keep updates need to know that has happened, and abandon any current games with a ticking AI timebomb (potentially), and start a new one, with other stuff.
[...]
Fork it as a separate thing. That was the approach taken by the developer of NoNoCAB that continued over NoCAB fixing its bugs and expanding and updating the code for the new OpenTTD features.

As far as you credit the original work everything is fine. All AI code is currently under GPL.

No AI can be perfect. I usually leave them alone until they crash (and reset) or just go bankrupt. Some can be useful like the ones that connect cities by road or the one that build trees.
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

So here is my findings so far, it's actually better than i expected, but as luck would have it, the dodgy AI are (almost entirely) abandoned ones, based upon activity on TT-F or Github (if it has one associated with it).
'Stable' just means it didn't crash, under a 250 year period (years were 1950 to 2200, this seemed a reasonable play length). Anything in red failed stability tests (crashed in at least one of the 250 year tests), and anything in green passed it, although some of them passed with issues. I've not tested all AI yet, and I won't be testing anything that just sits idle, because if that crashes too, then I'm not sure anything is worth it anymore.
- Older ones tend to fail CPU usage (maybe it was introduced sometime after AI's were, so newer bots are more careful with CPU time than older ones to account for it?)
- There's a clear lack of foresight in a lot of bots. With some seeing a vehicle turning a profit and cloning it, seemingly without any sanity checks for length of track/route, and the length of each vehicle on it, to know if that, based on its building method (ie: junction type, and distance between signals) will cause a stall, and them all to turn into negative, often when this happened the depots it built were at unhelpful spots, so it was trying to sell the vehicles, but never could as they never got to the depot.
- Both old & new have questionable routes at frequent intervals, 1 2 & 3. Multiple bots also build stations without ever using them, such as this, with free extra weird route in the middle, ticked & crossed with in use and not in use stations, some stations weren't even given tracks to it.
- Apparently, OpenTTD updates can break bots (this is based on TransAI, and its simple fix to a sudden bug when no AI update was done at same time), I cannot find any info from the developers of OpenTTD, on openttd.org or tt-forums.net that suggest any update about (from what I can tell is) the time it broke that your AIs may be about to break, so maybe there should be better transparency between devs and their content creators? It is also possible I overlooked something because i did that at 3am, or this information was emailed to people wth AI in BaNaNaS, in which case i'd not find it anyway.

Anything marked as 'tested OK' at the top neither crashed, nor did anything I noticed that was beyond logic. I'm currently running more AI simulations on spare computers, doing random mix & match of AI in each one, so i dunno how long they'll have the tag of glory. Maps were randomly generated each time, over 1024x1024.

One thing I also noted was RAM usage. Most use less than about 5 to 10MB. However some spike at over 50MB. Is there such a thing as a problematic usage of RAM, or are devs not bothered and AIs are allowed to use whatever RAM they can get their hands on?
Overall it seems, honestly. No one really cares about AI, which is sad, as some of the AI I've tested are genuinely really good at what they do, even if they are a bit flawed (looking at you ChooChoo, with your magnificent networks), with some clearly having lots of love and attention given originally.
2023-03-16 05_27_11-OTTD AIs.odt - OpenOffice Writer.png
2023-03-16 05_27_11-OTTD AIs.odt - OpenOffice Writer.png (49.25 KiB) Viewed 7477 times
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
xarick
Transport Coordinator
Transport Coordinator
Posts: 341
Joined: 26 Feb 2015 00:52

Re: AI Thoughts, listings & Upkeeping of them

Post by xarick »

I have a solution for "excessive CPU usage in valuator function". I call it "Slow Valuate".
https://github.com/SamuXarick/OpenTTD/c ... 4338a80f2e

At the end of each compat.nut file, I created a function that acts as a replacement for the Valuate function. It overrides the openttd's own Valuate with this one.

The difference is, instead of doing the valuation of lists in a single step, often with perceived stutters, and ensure the number of operations falls within a maximum limit, it does the valuation in multiple steps, without any limit of operations. By multiple steps I mean it allows the script to be suspended and resumed while processing through the valuation. The downside is that, while in the original Valuate, the game state is the same for the entirety of valuation, on the Slow Valuate, the game state changes every time the script is suspended and resumed, which may not be desirable.

The end result is a script that won't ever crash to "excessive CPU usage in valuator function", but will feel a bit more smooth cpu-usage wise, and slower-thinking.
Formerly known as Samu
AAAHogEx
Engineer
Engineer
Posts: 119
Joined: 30 Jan 2022 15:37
Location: Japan

Re: AI Thoughts, listings & Upkeeping of them

Post by AAAHogEx »

xarick wrote: 16 Mar 2023 22:41 I have a solution for "excessive CPU usage in valuator function". I call it "Slow Valuate".
https://github.com/SamuXarick/OpenTTD/c ... 4338a80f2e
I think this is a fair solution. Right now, using Valuate is more advantageous for AI with heavy use because of the speed of AI, but there is an increased risk of stopping due to map size, number of vehicles, etc.

I haven't read the VM code properly, so I could be wrong, but I feel that if the VMs are running in separate threads, it would be possible to suspend in Valuator and use multi-core, which would solve a lot of problems. However, We might need to be careful about API calls inter threads, synchronization on save, etc.
HGus
Engineer
Engineer
Posts: 123
Joined: 12 May 2013 22:28
Location: Argentina

Re: AI Thoughts, listings & Upkeeping of them

Post by HGus »

Good work!
Please post a link to the .odt when you finish it.

I like ShipAI (even it tends to build excesive buoys/depots), NoNoCAB (very tough) and SimpleAI (stronger version of original AI).
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

HGus wrote: 17 Mar 2023 08:29 I like ShipAI (even it tends to build excesive buoys/depots), NoNoCAB (very tough) and SimpleAI (stronger version of original AI).
All of these are on the currently tested list. You also may have solved a problem for me too. The game has a lot of buoys, most of them unused, and the game doesn't tell you who built a buoy originally, or the last person to use it, even if you land area info tool. Further problem there is i have more than one AI building ships and using buoys in the current testing of 14 AIs at a time, if I can't work which one is responsible I might have to use only 1 of them in the next test batch until i find the culprit. I am guessing one is building them, then not deleting them when they retire the boat that used it, or some other AI was using it so it couldn't destroy it, but that AI is also not using it now, and also didn't delete it as the last one to use the buoy.
I appreciate a good buoy, but you don't need them in spitting distance of each other :mrgreen:
2023-03-17 16_29_58-OpenTTD 13.0.png
2023-03-17 16_29_58-OpenTTD 13.0.png (81.71 KiB) Viewed 7362 times
Now I've got lots of AI with RV's in the current batch, which I also threw the updated versions of AAAHOGX & ChooChoo into, I've also started running into AI that give absolutely no care if their vehicles block the road and other competitors, so that's nice of them. So far the main problematic ones for this is PAXLINK & AAAHOGX (red in this image), maybe that is what the 'HOG' stands for, it hogs the road :lol:. Having said that though, a lot of AI blindly duplicate vehicles purely due to profit and no other thought, as witnessed also in this image by Purple, who is KrakenAI, this isn't a unique issue though. AdmiralAI & TrAIns do the same, but perhaps even worse as they do it with trains which can entirely jam up their network and no one moves ever.
2023-03-17 16_42_02-OpenTTD 13.0.png
2023-03-17 16_42_02-OpenTTD 13.0.png (189.15 KiB) Viewed 7362 times
HGus wrote: 17 Mar 2023 08:29 Please post a link to the .odt when you finish it.
That's definitely my plan, as well as make a more basic table either on a webpage, the wiki (if permitted, i dunno what the rules are on that) and/or on TT-F here.
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
HGus
Engineer
Engineer
Posts: 123
Joined: 12 May 2013 22:28
Location: Argentina

Re: AI Thoughts, listings & Upkeeping of them

Post by HGus »

Redirect Left wrote: 17 Mar 2023 16:50
That's definitely my plan, as well as make a more basic table either on a webpage, the wiki (if permitted, i dunno what the rules are on that) and/or on TT-F here.
The wiki is always open to improvements. There are already a few AI tables done in the past where you can add your results or create a new one at the bottom.

https://wiki.openttd.org/en/Community/A ... 20of%20AIs
Jaume
Engineer
Engineer
Posts: 57
Joined: 24 Aug 2007 15:43

Re: AI Thoughts, listings & Upkeeping of them

Post by Jaume »

I think that BaNaNaS is a big cemetery. Most of AIs found in BaNaNaS are discontinued, and some ones aren't upgraded (for example SuperSimpleAI, because I lost my BaNaNaS account and I can't upgrade this AI. I can create a new account, and "fork" my SuperSimpleAI to SuperSimpleAI-NG, but then SuperSimpleAI becomes trash).

I think that other AI repository is needed, with two different directories. One free, which somebody can put their untested AIs, and other that only with AIs that were very tested by a quality team.
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

Jaume wrote: 18 Mar 2023 10:06 I think that BaNaNaS is a big cemetery.
In my opinion, I think you're right. BaNaNaS was probably a good idea at the time, but, and this is no insult to the developers, i do feel a lack of forethought was given towards the long term issues BaNaNaS, or any other in-game repository of GRFs/scripts would cause, such as where we are now. I feel someone really needs to go through it, and identify what is worth keeping on there, what is abandoned, and what is just straight up non-working, or actively bad enough (or in AI terms, crash enough) it might turn away users who tried them.
Jaume wrote: 18 Mar 2023 10:06 I think that other AI repository is needed, with two different directories. One free, which somebody can put their untested AIs, and other that only with AIs that were very tested by a quality team.
That would be pretty neat, I agree with that, or in the ingame directory of available stuff being turned into a grid or table listing stuff
HGus wrote: 18 Mar 2023 01:02 The wiki is always open to improvements.
Yes, I've seen that page. It's a bit of a mess, and the tables are no where near as detailed as I was thinking of. It's a shame I cannot do HTML or CSS tables here on the forum. I'm currently writing up a HTML page of it all.

Here's my current working document with the passes & fails. The ones at the top have so far never crashed, or done anything offensive or highly illogical. The ones with problems I listed as either stable or not, in case you want to use them including possible problems, red is inactive (in terms of developer / updates), if i have anything in strikethrough it's (seemingly) fixed since i reported it. I made the document in OpenOffice, so i thiiiink it'll work in basically any major office suite.
OTTD AIs.odt
(25.93 KiB) Downloaded 154 times
Any problematic AI's that had active developers, I've reported the problems I've seen to date in their relevant threads. The rest i suspect is just... as Jaume said, in the graveyard.

I'm currently trying to identify which AI is responsible for massively excessive buoys (ShipAI and one more i think), and there's at least one that is blindly terraforming and blocking off boats, so that's both of them going to be added to the list of problematic ones when i find them. Always fun to put random AI together for each test, I find new things each time around!
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
AAAHogEx
Engineer
Engineer
Posts: 119
Joined: 30 Jan 2022 15:37
Location: Japan

Re: AI Thoughts, listings & Upkeeping of them

Post by AAAHogEx »

AAAHogEx
2.Uses RV stops that sometimes delay competitor vehicles
I do not see what the problem is. Since the roads are shared, I don't think it is possible to create stops that don't block other companies at all, and the AAAHogEx stops are not intended to block other companies' RVs, but to grow the town and transport passengers to the station.
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

AAAHogEx wrote: 19 Mar 2023 02:49
AAAHogEx
2.Uses RV stops that sometimes delay competitor vehicles
I do not see what the problem is. Since the roads are shared, I don't think it is possible to create stops that don't block other companies at all, and the AAAHogEx stops are not intended to block other companies' RVs, but to grow the town and transport passengers to the station.
A lot of the AI's I've tested so far build off-road stops, rather than an on-road halt. So when the RV is at the stop, it is off the road and not an obstruction. This only helps if the AI doesn't endlessly reproduce vehicles and jam up the station too much so RVs overflow on the road again, but usually it mitigates the problem reasonably. AAAHOGEX also builds plentiful stations in the same town/city, presumably in an effort to grow them, and when they're all on-road stops it really does slow things down if anyone else wants to use RV's in the town. For these tests, I've basically written down anything that is possibly disruptive to others, if not themselves, in at least one of the tests.

In terms of profit, AAHOGEX is top of all tests that included that AI, so that's something neat.

In case my vague wording is hard to figure out (honestly I struggled to word them),
On Road (this is also AAAHOGEX):
2023-03-19 03_13_38-OpenTTD 13.0.png
2023-03-19 03_13_38-OpenTTD 13.0.png (65.94 KiB) Viewed 7256 times
Off-Road (not built by AAHOGEX in this example)
2023-03-19 03_01_21-OpenTTD 13.0.png
2023-03-19 03_01_21-OpenTTD 13.0.png (146.48 KiB) Viewed 7264 times
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
AAAHogEx
Engineer
Engineer
Posts: 119
Joined: 30 Jan 2022 15:37
Location: Japan

Re: AI Thoughts, listings & Upkeeping of them

Post by AAAHogEx »

Redirect Left wrote: 19 Mar 2023 03:07 A lot of the AI's I've tested so far build off-road stops, rather than an on-road halt. So when the RV is at the stop, it is off the road and not an obstruction. This only helps if the AI doesn't endlessly reproduce vehicles and jam up the station too much so RVs overflow on the road again, but usually it mitigates the problem reasonably. AAAHOGEX also builds plentiful stations in the same town/city, presumably in an effort to grow them, and when they're all on-road stops it really does slow things down if anyone else wants to use RV's in the town. For these tests, I've basically written down anything that is possibly disruptive to others, if not themselves, in at least one of the tests.

In terms of profit, AAHOGEX is top of all tests that included that AI, so that's something neat.
Thanks for the quick reply!
Isn't on-road the most common way to build a bus stop in a city, since there is little space for an off-road bus stop in a city? As far as I know, most AIs that build bus stops in the city do so.
Also, since roads are a shared resource, it is almost impossible to use them without inconveniencing other companies. I also think the criteria for how much is allowed and how much is not is vague.
User avatar
Redirect Left
Tycoon
Tycoon
Posts: 7279
Joined: 22 Jan 2005 19:31
Location: Wakefield, West Yorkshire

Re: AI Thoughts, listings & Upkeeping of them

Post by Redirect Left »

AAAHogEx wrote: 19 Mar 2023 03:26I also think the criteria for how much is allowed and how much is not is vague.
In the instance I wrote it down, it was in a large city that was absolutely rammed with vehicles from different AI. So granted that one was a chaotic scenario as lots of AI built there (perhaps at the start it was the highest population town so everyone ran for it). The only stop that was causing slow downs was from AAAHOGEX, with the other stops slightly out of the way of the flow of traffic (either by design, or i think more likely coincidence). So its possible i was a bit harsh there based on one thing.
One thing that's nice though, whilst it does spam stops around towns, it does actually use them all, and they're orientated the correct way for the road grid, which some AI struggle with, a lot build excess stations and don't use them.

An attempt was made, I guess? This is NotPerfectAI.
2023-03-19 03_45_27-OpenTTD 13.0.png
2023-03-19 03_45_27-OpenTTD 13.0.png (63.51 KiB) Viewed 7244 times
Image
Need some good tested AI? - Unofficial AI Tester, list of good stuff & thread is here.
Post Reply

Return to “OpenTTD AIs and Game Scripts”

Who is online

Users browsing this forum: No registered users and 2 guests