News:

The Forum Rules and Guidelines
Our forum has Rules and Guidelines. Please, be kind and read them ;).

Passengers walking between intermediate stops

Started by jamespetts, August 13, 2011, 11:51:28 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

jamespetts

In the forthcoming 10.0 release (see discussion here), one of the new features being tested is a setting enabling passengers to walk between intermediate stops on their journey. So, for example, a journey could be A ----> B --walk--> C ----> D. In existing versions, there would have to be a transport connexion between B and C for passengers to be able to travel from A to D. Passengers will be able to walk between stops when one of the stops is within the coverage radius of the other. This should help to avoid having to make stops public to allow an interchange between different players' networks, which should be of great assistance in multi-player games.

However, as Knightly has pointed out, there are also some issues with this feature that are unavoidable without making the routing system much more complicated, which is not feasible because of its likely impact on performance. As a result of these issues, I have made this feature optional. It is disabled by default, but can be enabled by setting allow_routing_on_foot = 1 in simuconf.tab, or checking the allow_routing_on_foot box in the settings dialogue (press "i" in Pak128.Britain-Ex to get to the settings dialogue). The issues derive from the fact that the routing system is such that there is only one best route between any two pairs of stops. So, in the example:

A ----> B --walk--> C ----> D

the route between B and C will be a walk for all passengers going between B and C. This will be so even if there is a non-walking route between B and C (although only if the non-walking route is slower than the walking route, assuming that people walk at 5km/h). So, if there is another stop, E, within walking distance of B, then passengers going from E to D will:

E --walk--> B --walk--> C ----> D.

Notice that passengers there walk consecutively between two stops. This will happen even if stop C is not within stop E's coverage area, meaning that passengers will, in cases of consecutive stops like this, walk further than is intended. As stated above, this cannot be avoided whilst maintaining reasonable performance. Further, if passengers are simply going from E to C, they will walk all the way:

E --walk--> B --walk--> C

even if C is not within the designated walking distance (set in simuconf.tab) of E. Moreover, this effect means that C, in effect, shares its coverage radius with B and E, which might not be ideal if they are operated by rival players.

However, this feature is still potentially useful, and it is not easy to judge without testing whether instances of the anomalies caused by this feature will be more (or more significant than) instances of the anomalies caused by the absence of this feature. (It might well be that this is best enabled in multi-player games and disabled in single-player games, for instance, but that may depend on the results of testing). I should be very interested in feedback on this topic.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Xelofino

I'm looking forward to test it soon  ;D
I think for Multiplayer it's a huge advantage because there is no more use of the OpenHand (what's it in English?) for building stations and the whole complicated connecting stuff of two players tracks. I will surely test it, even in a multiplayer game, if demanded.

Xelofino

P.S: Mai english saunds teribel  ;)

jamespetts

Quote from: Xelofino on August 13, 2011, 06:16:51 PM...
I think for Multiplayer it's a huge advantage because there is no more use of the OpenHand (what's it in English?)...

Public player?

Glad that you are looking forward to using it!
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

ӔO

I think this would work great for rail stations, but would be a problem for bus stops.
My Sketchup open project sources
various projects rolled up: http://dl.dropbox.com/u/17111233/Roll_up.rar

Colour safe chart:

wlindley

So in the final case of the first post here, if we are in a multiplayer game, and you have stations E and C separated by four squares, and I build station B which overlaps the coverage areas of both E and C -- then that will "steal" all your passengers between E and C?  Even if I don't connect stop B to anything, or have any routes stop there?

jamespetts

W. Lindley,

if you don't have any routes stopping at B, then you can't transport the passengers from B to anywhere, other than E or C, and you won't get any money for doing so, of course, because passengers don't pay you to walk. Further, passengers will only actually walk via B rather than take the 'bus or tram (etc.) if walking is faster, which would mean that the public transport service would have to be poor for the walking route to succeed.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

wlindley

OK, that makes sense.  I was concerned that a malicious player could sabotage a competitor by merely building "nuisance" stations.

Carl

It's difficult to test this usefully at the moment since the routing problems in the latest version of the sources make it difficult to observe the effects of walking passengers in isolation.

Some thoughts in principle, though: if the "anomalies" arise only in cases where it is faster to walk between a series of stops than to take the player's transport, then in principle this doesn't sound too bad. It creates an incentive for players to create faster connections between this series of stops.

rsdworker

that's sounds good - cause people could walk from B to C for example  i build a subway system with 7 stops - one station with two platforms but they are separate so people from A to B travel on train to interchange then exit station then walk to other station and enter and travel to the final destinain

chs

This implementation sounds very good to me. From my experience, I usually walk 2 bus stops (or more), if I have to change bus lines - unless I the waiting time happens to be zero (when I pass the stop just when th bus is coming). Since rarely happens, I do the same as is implemented.
So without having tested it, it sounds great.
Next step: test it!

wlindley

At least with the 11 September version, passengers do not seem to walk from one company's stations to an adjacent company's station. 

Starting with a new map, I started two companies in the same large city, "A" on the west side and "B" on the east.  Their transit networks connect at stations that are in immediately adjacent squares.  No passengers ever are shown as traveling from A's stations to B's or vice versa.  Surely that should be the case, if passengers know they can walk to the other station? 

jamespetts

Don't forget that this was disabled by default, and needs to be enabled by a simuconf.tab setting (or use the "Experimental" tab in the "settings" menu).
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

sdog

i suppose it's turned on for pak128.britain-exp 8.1?

the walk distance seems a bit high. Pax walk in my cities to every stop. I'm not sure if this could get a bit problematic for route searches if the game gets a bit more complex.

ӔO

Quote from: sdog on October 03, 2011, 02:34:18 AM
i suppose it's turned on for pak128.britain-exp 8.1?

the walk distance seems a bit high. Pax walk in my cities to every stop. I'm not sure if this could get a bit problematic for route searches if the game gets a bit more complex.

If the walking distance is determined by the station coverage, then maybe the station coverage is too high?
From my research, apparently, the ideal bus stop interval is 400m apart, which means most people are only willing to walk up to 200m to get to their destination when using public transit.

since pak britain uses 250m per tile and station coverage=3, that would mean each station has a coverage of at least 750m in each direction.
My Sketchup open project sources
various projects rolled up: http://dl.dropbox.com/u/17111233/Roll_up.rar

Colour safe chart:

jamespetts

It is indeed turned on by default in Pak128.Britain-Ex 0.8.1; the distance is set, as AEO states, on the basis of coverage radii (or else, there would be inconsistency in walking distances). The station coverage is set taking into account not just 'bus stops, but also railway stations, harbours, airports, etc.: that 400m is the ideal distance between 'bus stops does not mean that 200m is the furthest that anyone is prepared to walk to get to any public transport, much less that it has always been so (my grandparents' generation, for instance, would routinely make 20 mile round trips on foot to do the shopping). A 750m radius seems to be a reasonable compromise, bearing in mind that, in a large city, one will want to have more than one stop in many coverage zones to ease overcrowding and allow more routes.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

sdog

I  think there might also be a difference between north america and europe, stops seem to be spaced closer here in Toronto in general.

It must have been a fluke then, sadly i don't have a screenshot, i had a screen indicating passengers were walking to all stops in a town that was certainly larger than station coverage.

fbfree

A really good in-depth discussion of the stop spacing issue can be found here:

http://www.humantransit.org/2011/04/basics-walking-distance-to-transit.html

750m is quite a long distance to walk to get to a local bus stop, but it's not unreasonable when walking to an express route or central station stop.

jamespetts

There are some known (and, without extensive and fundamental changes to the code that would significantly degrade performance, irresolvable) issues with walking, including the fact that passengers can and will walk between an unlimited number of consecutive halts, as there is no way for the routing system to tell apart a walking connexion between stops from any other sort of connexion. Is it possible that this is the effect that you saw?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

sdog

There were two dense local networks in a town. With a strongly frequented dock in the south owned by one company, and a long-distance train station in the north. Distance between the two roughly 20 tiles. On the dock readout a direct connection to the train-station was mentioned, with type "by foot".

I can't quite interpret what i saw. Just keep this as kind of anecdotal indication that there perhaps might be something going different than expected, but until more and also more reliable reports come in not too much attention is perhaps needed.

Last time i looked i could not verify this again. I did not pay enough attention at that point in time as i thought it was expected to work this way.