Tmpe: If cims can't 'park and ride' due to lack of parking, they need to adapt better

[s2500111]

Description

I first watched this weird behavior after I added some bus stops to the main arterial leading into the city. Instead of following the road, cars are now driving into the neighborhood next to it and are searching for a parking spot. If they find one, they'll walk to the next bus stop, but more likely they are driving around round after round, because everyone else parked there and used public transport for getting into the city. While this behavior is probably a real world city planners dream, it's not very helpful in the game. I'll try to build a park and ride facility, but it shouldn't be like that.

To reproduce this error, it is probably necessary, to have a city without a highway network, so that the problem concentrates around one or two exit ramps. It will probably spread out if there are multiple options and become indistinguishable from normal behavior.

Reproduction Steps

1. Add public transport near an important highway exit.
2. Watch.

Mod List

Only listing mods that could possibly affect this. For a full list look at the output log file.

• TM:PE 10.20
  -- improved AI enabled
  -- Realistic parking enabled
• Improved Public Transport 2 [r4.3.1]
• WG Citizen Lifecycle Rebalance v2.6 (I tried to disable it, didn't help, I don't think it messes with the pathfinding for the citizens but only alters the destinations of their journeys)

Output Log

https://drive.google.com/file/d/1gfELnYu1S1UrvXaLFOSe2aJz69EKuBqF/view?usp=sharing

Savegame (Optional)

https://steamcommunity.com/sharedfiles/filedetails/?id=1755969736

Screenshots (Optional)

[aubergine10]

What would you expect them to do?

[s2500111]

I'm expecting them to behave a little bit more like real people and drive to their destination instead of queuing up for parking in the first neighborhood they are finding.

I like the Idea that people are using park and ride facilities, but only if this is useful and if they aren't bunching up. So it should be only a small percentage.

Another option would be, that they are trying it, but go further, if they don't find a spot in the first attempt, so that they aren't circling around this spot.

[aubergine10]

So if Parking AI is enabled, and a cim can't find parking to transfer to Public Transport (PT), they should instead drive to where that PT was going to take them to and resume their journey from there?

For example, if the planned route was:

drive to bus stop ---> get bus to next district ---> walk to shop

And there was no parking near bus stop, then their route would become:

drive to bus stop --> no parking, so drive to next district --> walk to shops

Issues might arise if they can't use car to get to next district (eg. only way is a dedicated bus lane road), and also what happens if there's no parking in next district (do they give up and go home or 'pocket car' or...)?

Overall, I like the idea that a car driver would skip a PT portion of a route if they can't 'park and ride'. Just need to work out to do in scenarios above.

[s2500111]

This, or a lower amount pf sims that try to park and ride. In reality, it is hard to convince people to use these P+R options, so it shouldn't be that extreme in game.

[aubergine10]

Yup, I think this would be good enhancement.

So, to summarise all this:

• A percentage of cims try park and ride
  
  
  
  • That percentage could be based around how much parking (total, irrespective of available) within range of where they start their journey
  
  
  • So, less total parking = fewer cims try park and ride at that location
  
  
• If no parking when they get there, they either:
  
  
  
  • Drive to destination (preferred, that's what a driver would do)
    
    
    
    
    
    • Or whatever the next stop would be on the PT route?
    
    
    
  
  
  • Walk (if no driving route)
    
    
    
    
    
    • EDIT: Actually, where would they park in order to get out of the car and walk?
    
    
    
  
  
• Could possibly find some way to say "mark this PT stop as having parking issues" so over time cims would know not to try park and ride there?
• If all else fails (no parking, driving or walking possible) they give up and go home
  
  
  
  • Ideally some way to flag up the problem to user - either via info view, chirp, or whatever
  
  
  • Cim happiness should reduce as direct penalty to user for not providing enough parking
  
  

[s2500111]

Update. The AI is doing incredible stupid things on this. After they weren't able to find a spot, they start to drive around this place in a big circle (over the big roundabout intersection, to the next, heading into the city and driving out again), only to search a parking spot on the same roads where they came from again ...

I think the easiest, temporary solution would be a flag to disable the park and ride behavior altogether.

[s2500111]

Something has changed. They now don't even try to drive to a parking spot nearby, but are straight up driving to the bus stop and are waiting there in the lane, while thinking about where to go. That's especially stupid ...

I've made a video of it.

This only happens with the improved AI, but if I turn off the realistic parking AI, they start to just despawn their cars at this point.

So two problems:

• Massive overuse of the P+R mechanic.

• Cars stopping while searching for a parking spot.

[aubergine10]

We're working on a load of bug fixes to the parking AI at the moment, so hopefully the 10.21 release will solve this issue.

Would you be willing to help us test the updates to parking AI? If so:

• Unsubscribe any workshop versions of TM:PE first
• Download and extract TrafficManager.zip
• Depending on what you use to unzip the file, there might be multiple nested TrafficManager folders; you want the one that contains a bunch of .dll files.
• Put the TrafficManager folder that contains .dll files in to the following location:
  C:\Users\USERNAME\AppData\Local\Colossal Order\Cities_Skylines\Addons\Mods replacing USERNAME bit with whatever your Windows username is.

That will add a beta version of TM:PE v10.21 which contains most of the parking AI fixes. You'll need to enable it in Main Menu > Content Manager > Mods.

[s2500111]

I only geht Version 10.20 from doing so. Am I missing something?

[aubergine10]

Ah, the version number in that zip isn't bumped yet (we usually only do that during release process). So yea, it will still say 10.20, but it's got the updated parking AI code from pull request #359 in it.

Does it improve the parking behaviour any?

[s2500111]

As far as I can tell, they seem to distribute better over the available parking spots and stopped piling up in the first neighborhood.

But the issue with cars stopping right before the bus stop isn't solved.

[aubergine10]

Thanks for testing and the feedback. It's good to hear that it's improved some aspects of the parking.

For the park and ride aspect (eg. piling up at bus stop) we need to do some more extensive changes so it will probably be a while yet before that's fully sorted.

[VictorPhilipp]

I guess that before path-finding starts the road segment with the bus stop is treated as having road-side parking spaces. But as they approach the segment they realize that parking is not possible there. That would be even an issue of the base game itself. I will check my theory.

[aubergine10]

That might also be why cims aren't parking near my lakeside pier (as per earlier comments in #359). I have a bus stop there, but by time cims try and park they'll likely find better alternatives to the opposite side of the road.

[VictorPhilipp]

Yes, this vehicle (and also many others) wants to park here.

(Test was performed with current LABS version and without PR #359)

[VictorPhilipp]

This is only solvable by adding an indication to each segment whether

• road-side parking is possible (= has parking spaces and no bus stop) and/or
• whether a parking building with free parking spaces is present within a short range.

[VictorPhilipp]

Tagging #176