The sensors, which track vehicles via their Bluetooth systems as they travel past sensors at intersections, were installed about six months ago, and have so far revealed a range of insights, including that people are not necessarily taking the routes to get places around the city traffic engineers thought they were, Dunedin City Council transport staff say.
Council transport strategy manager Nick Sargent said the information helped council staff trouble-shoot temporary traffic problems in the city, and the journey and travel time information through the city provided invaluable data for transport planning.
Because it was new, the system was still being test-driven by the council’s transport team and the data set was still being built up.
In the past, the evidence used to criticise Dunedin’s transport network was typically anecdotal, Mr Sargent said.
"Now we’ve actually got the data to have a meaningful conversation.
"We want to have conversations around facts, not around hearsay and conjecture and emotion."
One limitation of the system was that only about 12% of the cars and trucks on Dunedin’s roads had the on-board digital technology required to be detected at an intersection.
However, through its artificial intelligence the system could establish a normal travel time for a section and when congestion or traffic jams occurred, the system would "light up" yellow, orange, or red, to alert council staff to the issue, so council staff could intervene.
The council could, for example, change the phasing of traffic lights to ease pressure on a road when construction reduced it to one-lane rather than two.
In the same way, the council could see when the city suffered from congestion.
"Whereas other cities in the world have long peak hours of congestion — what happens is people experience these adverse travel times for a long time — in Dunedin we have a peak of about 15 or 20 minutes where the network is really stressed," Mr Sargent said.
"So if you hit the network about 5.05pm to about 5.25pm, that’s when you’re experiencing most of the stress.
"But if you hit it outside of that, suddenly it’s gone.
"Because people are creatures of habit, you can actually in this building, at 5pm, press the lift button — everyone goes to the lift, not everyone can get in the lift.
"That’s happening out on the road out there."
As the council prepared for major projects in the city, including the construction of the new Dunedin Hospital in the city centre, the data the sensors provided would take the guesswork out of the planning.
Council transport engineer Blair McKeever said seeing how people actually used the network had already proven beneficial.
The council reported to the NZ Transport Agency travel times for a set of routes each year.
"The old-school way of measuring travel times is someone gets into a car with a stopwatch — drives it [a route], drives back and drives it again," Mr McKeever said.
"We are finding that some of those routes aren’t necessarily the right ones.
"For example, the St Clair one, the route sort of comes along Forbury Rd, Hillside, Princes, to the Octagon — and we’ve found with the Bluetooth system hardly anyone actually drives that route at all."
Some of the work he had done with the system was to categorise the routes vehicles actually took from a given intersection.
"Very roughly, we can work out where people are going," Mr McKeever said.
"People that come in on the motorway, ‘X’ many of them end up in the tertiary area, or have transited the city, gone to north Dunedin ... wherever."
The information the tracking system collected from a vehicle’s Bluetooth system was something similar to a digital serial number.
It was converted into another number for use in the traffic management system and only stored on a short-term basis before it was dropped from the system.
And because the system also only tracked vehicles from the first time they entered an intersection in the network of 50 sensors to the last time they were detected, drivers’ privacy was maintained, he said.
Related Stories
Comments
Some one should ask google where they are getting their traffic data from, a quick "google traffic dunedin" gives a few good results.
There's a fundamental flaw with how they are collecting their data. A flaw that a first year statistics student could identify. Their equipment is detecting bluetooth sensors in passing cars. They note that only 12% of the vehicle fleet is so equipped. There's the first problem. They are basing their assumptions on only 12% of the vehicles on the road. Not a sufficiently large sample to provide meaningful data. Second, bluetooth-equipped vehicles are likely to be more modern vehicles and a large portion of them are likely to be owned and operated by businesses. So their data is based on affluent people (the people who can afford to drive bluetooth-equipped cars) whose driving patterns will probably differ from less well-off people. And the pattern of usage of business vehicles is going to be significantly different to private vehicles. Are they really going to use this data to alter traffic flows?
This is an example of how bad town planners and civil engineers are at research. Absolute rubbish. And it's not just ours - try getting meaningful search results on Barnes dances, for example. All you'll find is a history of flawed and meaningless research.