Reliability issues with RTPI

thomasj

Anyone finding reliability issues with RTPI particularly late at night? Just to give you an example have a look at screenshots of the last 39. They were a few minutes apart, the difference is huge.

It seems to be an on/off thing. Some nights there would be no problem. Other nights there would be a steep jump all of a sudden

Anyone come across this before?

StreetLight

I thought this sort of thing would have been sorted out long ago as teething problems. Sadly, from looking at the amount of similar complaints on the Dublin Bus Twitter page, it seems as prevalent as ever.

Mrs OBumble

Don't know about Dublin, but in Galway the only route where I believe the RTPI is vaguely resembling real-time is 403. I also use 405 and 409 a lot, and am convinced that the figures only ever show timetabled times (if that), and bear little resemblance to reality. Still.

lil5

There's so many caveats about the RTPI (depending on the route) that in some cases it's just totally meaningless and most definitly not good enough for planning around it.

The 7 route within 10 minutes of either termini is hit and miss regarding the departure times or the timetabled v the actual time (which should really be indicated on the displays and the app by now).

If it's not just showing up as 'Due' it could appear or disappaear at any stage.

The hourly Cherrywood bus on the RTPI is frequently so far from reality that it's almost funny to watch it flying past with the RTPI showing it to arrive in 10 minutes.


The 4 route is generally better, but there's also still quite a few phantom buses or disappearing acts.

LXFlyer

RTPI will be off kilter today.

A full service is operating but with virtually no traffic - hence buses are travelling much faster than the RTPI schedules would expect.

The driver of my early bus this morning didn't rush but was still 25 minutes ahead of where he should be by the city centre.

This will also mean buses arriving early for driver hand overs - so people should be patient!

gctest50

lxflyer wrote: »

RTPI will be off kilter today.

A full service is operating but with virtually no traffic - hence buses are travelling much faster than the RTPI schedules would expect.............!

It's a bit of a sad effort though, vehicle tracking has been around for years, the buses have wifi/internet access on them
and something is made aware of the route number to run the display on the front etc

With all that you could have nearly live tracking on a map of where the bus actually is too

This could run alongside what ever is there already

LXFlyer

gctest50 wrote: »

It's a bit of a sad effort though, vehicle tracking has been around for years, the buses have wifi/internet access on them
and something is made aware of the route number to run the display on the front etc

With all that you could have nearly live tracking on a map of where the bus actually is too

This could run alongside what ever is there already

Well unless a special RTPI schedule were implemented for the day, it's impossible.

I'm merely warning anyone, particularly those mid-route, that buses will show up faster than the RTPI will suggest due to a complete lack of traffic and indeed passengers.

gctest50

Something like this :

LXFlyer

gctest50 wrote: »

Something like this :

I didn't mean that would be impossible - far from it!!

I meant the predictive times this morning!!

smackyB

It's annoying that they won't make the location available on their realtime API as this is usually a much better indicator rather than a vague minutes countdown. I've heard that the GPS data for tracking each bus is available however this data will never be publicised due to an agreement with Dublin Bus unions, any truth to this?

gctest50

If they hold it up too long someone will write a little phone app

Since phones have GPS , it could suss which bus you are on and update stuff

Should work ok-ish if enough people had the app

devnull

gctest50 wrote: »

It's a bit of a sad effort though,

Unfortunately RTPI will never be exact science because of the nature of the beast, RTPI is based on historical data and real time data, the Monday to Friday timetable is being operated so RTPI times are based on historical data for a Monday to Friday timetable.

Just saying that a bus is in x and to get to y it'll take z minutes is a fallacy, z is dependent on many things.

devnull

smackyB wrote: »

It's annoying that they won't make the location available on their realtime API as this is usually a much better indicator rather than a vague minutes countdown. I've heard that the GPS data for tracking each bus is available however this data will never be publicised due to an agreement with Dublin Bus unions, any truth to this?

Location isn't always the best indicator because of the fact that traffic plays a huge part in how long it will take to get from one place to another, today it may take 20 minutes to go to one place or another whilst on your average Friday night it could take double that.

RTPI predictions are generally based on the arrival times of that particular timetabled departure on the same schedule in the past along with current position and whilst it's not perfect or an exact science, it's more accurate than just disregarding previous schedules and basing it on distance and not taking into account the traffic.

Since unfortunately the RTPI system does not possess a crystal ball, past performance is the best indicator of future arrival times, since most traffic and pinch points and delays tend to be the same on most days, but of course there are exceptional events.

LXFlyer

Exactly - I was merely warning people, based on my own experience earlier that traffic was non-existent.

Cilar

I've been seeing the same issues on other routes, the countdown suddently dropping from 20 minutes to 8 minutes. Seen this on a daily basis on the 70 bus in Clonee around 19:30 towards Dunboyne. I stopped taking the bus since then as it made me miss buses several times and there are only bus every 30 minutes at that time. I sent an email to the NTA who transmitted to Dublin Bus. According to them "everything is working fine with their BAC App" (whatever that is) and they won't bother investigate any further. So the conclusion I took: RTPI is totally unreliable, we won't do anything, move along.


Dublin Bus:
"Both buses were tracking correctly providing updates. Looking through the predictions sent there no major jumps for either trip. Depending on when a refresh was done on the BAC App it can give an impression of a jump in predicted arrival times. "

NTA:
"Dublin Bus checked their data and didn’t find a problem (see above)"

magicbastarder

gctest50 wrote: »

With all that you could have nearly live tracking on a map of where the bus actually is too

one simple thing for people living near the terminus (and for whom the time displayed would be based on estimated departure) would be a flag which indicates whether the bus has actually left the terminus.
also, might be useful not necessarily to indicate the position of the bus (which could be problematic to display), but maybe just the number of stops left for the bus before it reaches the one you're at.

dudara

Three bus routes pass near my house, and the timing on each route is absolutely unreliable. It can go from 22 mins to 5 mins which isn't enough time to make the bus. It's extremely annoying.

LXFlyer

This kind of app would be very useful here:

http://traintimes.org.uk/map/london-buses/#73

magicbastarder

dudara wrote: »

Three bus routes pass near my house, and the timing on each route is absolutely unreliable. It can go from 22 mins to 5 mins which isn't enough time to make the bus. It's extremely annoying.

how close are you to the terminus? we're about two miles away and the timing is generally reliable.

markpb

devnull wrote:

Unfortunately RTPI will never be exact science because of the nature of the beast, RTPI is based on historical data and real time data, the Monday to Friday timetable is being operated so RTPI times are based on historical data for a Monday to Friday timetable.

It would seem that RPTI in Dublin is based only on some internal timetable and some historical data for that route. However there's another source which is vital: all the other vehicles currently or recently on the same roads. This would allow them to more accurately predict how each bus will progress.

If all the buses inbound on Dorset st are way ahead of schedule / faster than usual, there's every reason to believe the next few buses on that road will make the same progress so the timetable and historical data should be ignored.

devnull

markpb wrote: »

It would seem that RPTI in Dublin is based only on some internal timetable and some historical data for that route. However there's another source which is vital: all the other vehicles currently or recently on the same roads. This would allow them to more accurately predict how each bus will progress.

No, actually that wouldn't work, that kind of approach has been tried by operators in other cities and was discarded in favour a model that Dublin Bus currently uses now which was found to be more accurate because it takes into account what happened exactly the same time on the same day in the past number of weeks and aggregates it.

The reason being? On some routes a bus leaving say 4.55pm may take 30 minutes and the bus leaving 10 mins later may take double that Reason why? the later bus has more passengers because lots of people knock of work at 5, which makes the roads far busier than they were 10 minutes earlier.

A better guide of how the 5.05pm bus from a terminus takes on a Tuesday, is how long it took over a large number of Tuesdays at the same time, rather than what happened 10 minute earlier before everyone knocked off work.

LXFlyer

markpb wrote: »

It would seem that RPTI in Dublin is based only on some internal timetable and some historical data for that route. However there's another source which is vital: all the other vehicles currently or recently on the same roads. This would allow them to more accurately predict how each bus will progress.

If all the buses inbound on Dorset st are way ahead of schedule / faster than usual, there's every reason to believe the next few buses on that road will make the same progress so the timetable and historical data should be ignored.

The RTPI reads from a schedule that has been prepared by schedulers on the basis of historical information. There are a number of schedules that are used and these are changed on a regular basis depending upon the expected traffic conditions over a period.


An example of such a schedule is attached, which is the schedule for the 15b - first is Monday-Friday, then Saturday and then Sunday.


Each departure will have a full estimated timetable for each bus stop. The RTPI system will find where the bus is using its GPS and then apply the running times between its position and each stop along the route from that schedule. The same system is used in London to predict the times.


The system is only as good as the data put into it. If a bus gets delayed unexpectedly by heavy traffic or an accident, it will just keep showing the same expected times.

cython

devnull wrote: »

No, actually that wouldn't work, that kind of approach has been tried by operators in other cities and was discarded in favour a model that Dublin Bus currently uses now which was found to be more accurate because it takes into account what happened exactly the same time on the same day in the past number of weeks and aggregates it.

The reason being? On some routes a bus leaving say 4.55pm may take 30 minutes and the bus leaving 10 mins later may take double that Reason why? the later bus has more passengers because lots of people knock of work at 5, which makes the roads far busier than they were 10 minutes earlier.

A better guide of how the 5.05pm bus from a terminus takes on a Tuesday, is how long it took over a large number of Tuesdays at the same time, rather than what happened 10 minute earlier before everyone knocked off work.

Ok, how about a hybrid/elaboration of the above suggestion, whereby the "base" reference is the historical data and schedule, but some form of acceleration or progress factor could be applied based on how much ahead of their own historical data the previous services in the day were? Plenty of simulation and control systems in other fields already adopt a similar approach whereby they calculate their prediction based on various data of different ages and weightings, so there could be a tangible benefit to RTPI doing something similar. For example, the first few services today would have been out of whack still, but subsequent service predictions would have twigged that something was different, and that things were running quicker than normal. I can see drawbacks still when a major delay arises and/or things resume running to normal schedules, that these go out the window, but this is not that different from the status quo, especially in the instance of disruptions.

To say there is no way to improve the model (and I am not suggesting that anyone has said this, but there does seem to be a resistance to the idea of it all the same) is, IMHO, just wrong.

howiya

lxflyer wrote: »

The RTPI reads from a schedule that has been prepared by schedulers on the basis of historical information. There are a number of schedules that are used and these are changed on a regular basis depending upon the expected traffic conditions over a period.


An example of such a schedule is attached, which is the schedule for the 15b - first is Monday-Friday, then Saturday and then Sunday.


Each departure will have a full estimated timetable for each bus stop. The RTPI system will find where the bus is using its GPS and then apply the running times between its position and each stop along the route from that schedule. The same system is used in London to predict the times.


The system is only as good as the data put into it. If a bus gets delayed unexpectedly by heavy traffic or an accident, it will just keep showing the same expected times.

Can the Dublin Bus not use the historical data from last year's Christmas period?

As you say the system is only as good as the data put into it. So are we to assume then that Dublin Bus just don't want to put the data into it?

Anyone who orders a taxi can see where the taxi is in relation to their location, the technology exists. We should be able to to see how far away the bus is.

gctest50

howiya wrote: »

.........

Anyone who orders a taxi can see where the taxi is in relation to their location, the technology exists. We should be able to to see how far away the bus is.

this is what we need here :

lxflyer wrote: »

This kind of app would be very useful here:

http://traintimes.org.uk/map/london-buses/#73

thomasj

I was on a bus heading home today and commented to the driver how easier it was to get to blanch without centre traffic.

He was showing me a new box they have above their head in The cab that when the bus comes to a stop at a location it shows how far of scheduled time the driver is.

As testament to a quiet Christmas Eve been treated as a busy weekday , the driver was 45 minutes ahead of schedule!

LXFlyer

howiya wrote: »

Can the Dublin Bus not use the historical data from last year's Christmas period?

As you say the system is only as good as the data put into it. So are we to assume then that Dublin Bus just don't want to put the data into it?

Anyone who orders a taxi can see where the taxi is in relation to their location, the technology exists. We should be able to to see how far away the bus is.

Last year there was a Saturday service on Christmas Eve!!


The agreement with the unions apparently is for four days of Saturday service over Christmas - the four days this year are the 28th-31st!

thomasj

devnull wrote:

Just saying that a bus is in x and to get to y it'll take z minutes is a fallacy, z is dependent on many things.

It seems to be a waste to me.

Christmas eve people are generally winding down. Time with family etc.

New years eve would be more busier and with new years festivities etc on I would have thought would have been more suited to a weekday timetable than Christmas eve.

LXFlyer

thomasj wrote: »

It seems to be a waste to me.

Christmas eve people are generally winding down. Time with family etc.

New years eve would be more busier and with new years festivities etc on I would have thought would have been more suited to a weekday timetable than Christmas eve.

The issue again is down to the fact that there is apparently an agreement only for four days of Saturday service.


A Saturday service would have been far more appropriate yesterday - early morning buses were flying around with no passengers.


New Years Eve is similar - really a Saturday service is more than adequate.

thomasj

howiya wrote:

Anyone who orders a taxi can see where the taxi is in relation to their location, the technology exists. We should be able to to see how far away the bus is.

Yep but the technology that gives the positioning of taxis, relate to the GPS positions of taxis. The apps that have this technology hailo and a few taxi companies use the gps from the mobile phones of taxi drivers.

How is this going to work with Dublin bus? Do you assign a mobile phone to every bus?

thomasj

Yep I agree but if we were to base this on the 4 days union agreed, I would say that you would get more value for money for having a weekday service on new years eve than Christmas eve because of what's on in the city new years eve. Everything closes up early on Christmas eve.

dfx-

The RTPI is reliable outside the first few stops of the terminus.

I was surprised to see all the Xpressos yesterday evening, Phibsboro buses on the 25s and Harristown buses on the 38s and presumably on the 39s yesterday evening. Hadn't checked the site and presumed it was a Saturday service.

thomasj

dfx- wrote:

The RTPI is reliable outside the first few stops of the terminus.

In my case (see op) the stop where I was getting the bus is closer to the destination than it is to the departure terminus

LXFlyer

thomasj wrote: »

In my case (see op) the stop where I was getting the bus is closer to the destination than it is to the departure terminus

That sounds like a faulty GPS on the bus to be honest

howiya

lxflyer wrote: »

Last year there was a Saturday service on Christmas Eve!!


The agreement with the unions apparently is for four days of Saturday service over Christmas - the four days this year are the 28th-31st!

Either way they should have historical data for the quieter Christmas period and use it to make RTPI more reliable

howiya

thomasj wrote: »

Yep but the technology that gives the positioning of taxis, relate to the GPS positions of taxis. The apps that have this technology hailo and a few taxi companies use the gps from the mobile phones of taxi drivers.

How is this going to work with Dublin bus? Do you assign a mobile phone to every bus?

Buses have GPS

LXFlyer

howiya wrote: »

Either way they should have historical data for the quieter Christmas period and use it to make RTPI more reliable

OK - to do that for Christmas Eve would have required the schedulers to prepare a unique schedule for every departure in the city for one day only.

These schedules don't appear automatically - someone actually has to do the leg work of developing a full timetable for every single departure. That's an enormous amount of work.

The problem won't arise next week as Saturday schedules apply.

thomasj

I'm surprised monday 28th both Irish rail and Dublin bus are running Saturday services.

Isn't monday st Stephens day in lieu?

LXFlyer

thomasj wrote: »

I'm surprised monday 28th both Irish rail and Dublin bus are running Saturday services.

Isn't monday Stephens day in lieu?

Saturday is considered a normal working day at Dublin Bus and Irish Rail. The staff got their bank holiday on that day.


It's only where either Christmas Day or St Stephen's Day falls on the Sunday that there will be an extra day of Sunday services in lieu.

thomasj

lxflyer wrote: »

Saturday is considered a normal working day at Dublin Bus and Irish Rail. The staff got their bank holiday on that day.


It's only where either Christmas Day or St Stephen's Day falls on the Sunday that there will be an extra day of Sunday services in lieu.

Ah okay that's where I got confused !

Bank of Ireland stated that both Monday and Tuesday will be bank holidays.

cython

thomasj wrote: »

Ah okay that's where I got confused !

Bank of Ireland stated that both Monday and Tuesday will be bank holidays.

That's BOI being BOI. Monday is a bank holiday, but Tuesday is a day where the banks happen to have decided to remain closed, and certainly not a public holiday in the sense of a conventional bank holiday.

loyatemu

cython wrote: »

That's BOI being BOI. Monday is a bank holiday, but Tuesday is a day where the banks happen to have decided to remain closed, and certainly not a public holiday in the sense of a conventional bank holiday.

it's an extra day negotiated by the bank unions after some strike in the 1980s. So it is technically a "bank holiday" (as they're all closed) but not a public holiday. Usually bank holiday and public holiday are interchangeable.

TheChizler

wheresmybus.ie used to display a live map of all the bus positions, very handy for telling whether the bus had a passed or not when the real time info was saying it was due. It's disappeared since.

howiya

lxflyer wrote: »

OK - to do that for Christmas Eve would have required the schedulers to prepare a unique schedule for every departure in the city for one day only.

These schedules don't appear automatically - someone actually has to do the leg work of developing a full timetable for every single departure. That's an enormous amount of work.

The problem won't arise next week as Saturday schedules apply.

I'm aware there is work involved before a service/good appears before customers/potential customers. That is not unique to Dublin Bus.

But sure its grand, leave people at bus stops while at the same time you're telling people there's a bus due...

Polar101

I live 5 minutes from terminus, so I'm aware that the RTPI updates aren't really realtime until about 2 minutes before the bus is at my local stop. What is still confusing that some buses are completely missing until they are close.

An example from yesterday - this particular route has buses every 15 minutes in the Saturday timetable. About 7 minutes before a bus was due to depart, RTPI for my stop displayed the next buses due in 22 minutes, 37 minutes and 52 minutes. So the next bus was not in the system but the following 3 were. And the bus operated normally. Why was it not in the system?

thomasj

I had that with the 39A this morning. The next 39as were showing as due then 31 minutes. Its meant to be a 15 minute frequency.

When I start walking it showed up 2 minutes later online on RTPI

What seems to happen is it's in schedule mode and then it seems to disappear and then there's a delay before it shows back up in live mode.

Maybe when the driver sets the service up it goes off schedule mode but doesn't show up in live mode until the bus leaves the stop.

I don't know.

Polar101

thomasj wrote: »

I had that with the 39A this morning. The next 39as were showing as due then 31 minutes. Its meant to be a 15 minute frequency.

Yes this was 39A as well, actually.

howiya

RTPI completely unreliable again today. I should really learn to drive!

LXFlyer

howiya wrote: »

RTPI completely unreliable again today. I should really learn to drive!

It should be working ok as there is a Saturday service today and journey times should be in line with the predictive times.

The only issue would be the extra buses that are operating which will not appear on RTPI.

howiya

lxflyer wrote: »

It should be working ok as there is a Saturday service today and journey times should be in line with the predictive times.

The only issue would be the extra buses that are operating which will not appear on RTPI.

My issue was a bus not turning up despite it appearing on RTPI before its scheduled departure

LXFlyer

howiya wrote: »

My issue was a bus not turning up despite it appearing on RTPI before its scheduled departure

Maybe the bus had a mechanical issue?