Glasgow AI bus trial

The Scotsman

Glasgow AI bus trial   Brian Davidson, engineering officer, Glasgow City Council talks to The Scotsman's Transport Correspondent Alastair Dalton,Transport Correspondent

Transcript

00:00Hi there, Brian Davidson, Engineering and Officer at Trafcom, Glasgow State Council.

00:04You've been awarded nearly half a million pounds from the Scottish Government to develop an AI system

00:09to improve traffic flow on a key route into Glasgow.

00:12Could you tell me about that and the advantages it will bring?

00:15Okay, so the first thing is we want to improve the journaling time for buses.

00:19We want the buses to be more reliable, people to have more confidence that the buses only be there

00:23when it's supposed to be there, and also that the travel time for the bus will be reduced.

00:27So that's our key aims, is to improve the bus flow.

00:31But in doing so, because the corridor we've picked is shared with general traffic,

00:36it will actually improve the travel time for all vehicles.

00:39So it will improve the travel time for buses, but it will also improve the vehicle for HGVs

00:43and improve the travel time for cars and all traffic as well, and hopefully cyclists as well if they're on the route.

00:49And in simple terms, how will AI improve the situation on top of the technology you have already?

00:54So at the moment, we have an adaptive control system called Scoot,

00:58and it's really good at managing traffic flow, but it tends to work on a junction-by-junction basis.

01:04Now we have an input into Scoot called bus priority, so when a bus is running late,

01:08it can request priority to traffic lights.

01:10The problem we've got is, we don't know the bus is there and late until it's about 10 seconds from the junction.

01:16So we can usually help at 10 seconds away, but if we knew a couple of minutes away that the bus was coming,

01:22and it was late, and it was going to need help, then we could make plans in advance to try and ease the flow for that bus.

01:30We know where it is, we're tracking all the buses in real time, and we know that traffic signals will do it.

01:34But to be able to predict where the bus is going to be in 2 minutes' time rather than 10 seconds' time gives us lots more flexibility.

01:42But in order to do that, there's a lot of complicated calculations, because the bus might stop in a bus stop,

01:47but it might not, it might be going slower, the traffic might be heavier than usual.

01:51All these things will affect it, and the further back you're trying to detect the bus and model it,

01:56then the more complex that question becomes.

01:58And what sort of improvement in journey time would you expect the system to bring?

02:03So we were hoping to get 10 or 20% improvement, but we did a trial last year using just six junctions,

02:09using the same technology, and when we did the trial, we found that on occasions it could improve journey time by 50%.

02:16And we were a bit reticent about that, we didn't think that was, we thought the modelling was wrong.

02:21So what we did was, we implemented the plans that the AI had suggested for a real pond on the street,

02:27and we went out in a van, and we measured it, and it actually did improve travel time,

02:31and it improved all the general flow.

02:33So that gave us confidence to say, this technology should work on a larger scale.

02:37So now, the plan is to do it on 23 junctions, from the city boundary,

02:42at Cormarant Mary Lee, all the way up Cormarant Road, Shollardys Road, Egleton Street,

02:45right up to the Clyde.

Category

Transcript

Be the first to comment

Recommended