SmartBeam bus shelters are a taste of what physical AI could mean for Aucklanders. These are solar-powered bus shelters equipped with charging ports. Photo / Smart City
“Auckland generates terabytes of data every day. People are moving around. So are vehicles. Sensors are detecting. CCTV is capturing. There are air quality, water quality sensors and footfall sensors. Huge data sets are constantly being generated all over the city by multiple organisation.”
So says Smart City managing director Ben Ransley, explaining that this data needs to be put to work. “We need to be using all of that data, to interpret it, learn from it and, ultimately, act on it and improve the way the city is run and improve people’s quality of life.”
At this stage in Auckland’s evolution, our city is, from an information point of view, data-rich but insight-poor.
“Each area has its own insights, but the parts are not joined up.”
Auckland is not unusual. Cities typically generate data from many separate projects. In each case, the project aims to record data and perform a number of tasks to improve citizens’ quality of life or to make services operate more efficiently. These projects might include reducing congestion, improving safety, managing municipal assets or stretching rubbish collection budgets.
Ransley says data collection started as a way to measure and report on the projects. It asks questions - are the bins full, is the road congested, how many people are waiting at the bus stop?
Initially it was used by city departments to aid human decision-making such as sending a collection truck or switching on traffic signs.
A smart city takes this further. In principle, technology can increasingly take over that decision-making to the point where it doesn’t just tell someone a waste bin is full, but it despatches the collection truck to deal with it in the most efficient way.
Smart city traffic systems don’t just measure cars passing sensors, they adjust the signals so that traffic flows more smoothly.
Ransley says that requires a leap from data collection to automated decision-making. It is where most cities want to go, but few are there yet. He thinks Auckland can make the leap.
He describes the fully automated smart city mechanism as: sense, reason, act.
The idea draws on the way physical AI is used in advanced manufacturing. In sophisticated factories robotic systems don’t simply follow fixed instructions, they perceive what’s happening at each stage, work out what to do next and then act. It may not get everything right every time. When it makes an error, it can learn from the experience.
Over time, this kind of system becomes, as Ransley puts it, “a well-oiled machine that is acting in real time against everything that’s going on”.
The difference between an automated smart city and the technology used today is significant. Traditional IoT (Internet of Things) sensors operate on fixed parameters set at installation: if the bin is above 80% full, send an alert. There is no reasoning, no learning, no adaptation to changing circumstances.
Physical AI replaces that static logic with a system that accumulates context over time and improves its responses accordingly.
Ransley says it would have been impractical to deploy that kind of intelligence at city scale two or three years ago. Today, largely due to the extraordinary leap in AI capability, it is not.
“Physical AI is sensing the environment, reasoning with it, then acting on that to impact the living environment in real time or to cue something to occur. It doesn’t need to have a citizen-facing impact. It could be something internal to the system’s functioning. It’s like autonomous robots running a factory.”
For now, there are few automated workflows in Auckland’s smart city infrastructure. The AI outputs that currently exist act more like consultants, analysts or experts.
SmartBeam bus shelters are a taste of what physical AI could mean for Aucklanders. These are solar-powered bus shelters equipped with charging ports.
Ransley says while the idea is simple, it transforms people’s experience: “You’re out late at night, your phone battery is dying, you want to get home safely. A SmartBeam shelter is a welcome sight. It solves your immediate problems.
“There’s a charging facility. The shelter is lit-up to provide a safe waiting space for anyone who feels vulnerable getting home at night.”
He says Auckland Transport has used SmartBeam for the last six months.
The shelters collect data that goes beyond Auckland Transport’s basic patronage numbers: not everyone waiting for a bus catches one.
The organisation can learn how long people wait at a shelter, when they are busy and when they are quiet. That’s previously unavailable behavioural data about how people actually use public transport infrastructure.
There’s a sustainability angle with people charging phones using solar rather than the grid.
Auckland Transport is currently using the base model of SmartBeam with limited functionality. The features can be extended to include e-bike charging, advertising panels — which can earn revenue to help pay for the infrastructure, speakers, environmental sensors and free Wi-Fi.
Another example of physical AI is at the Rā Hihi flyover, which links Pakuranga Road to Pakuranga Highway in East Auckland. It opened in September. Here, Smart City’s SmartTraf provided real-time data on the flyover’s use from day one.
SmartTraf is a fixed traffic monitoring system using Bluetooth trackers and traffic radar.
Ransley says it delivered striking numbers: it measured around 18,000 city-bound vehicles on the flyover’s first day, rising to around 45,000 by the end of the first week. That rapid adoption curve and its effect on congestion across surrounding roads, was presented by the project manager directly to the Transport Minister. The technology influenced ministerial briefings and acts as an independent check on whether new infrastructure is performing as promised.
He says SmartTraf quickly identified that there were poor travel times at the intersection at the base of the flyover. After his team flagged it, the flyover layout was changed, an after-study confirmed there was improvement.
SmartRadar is a related technology used for transport analytics. It replaces people standing roadside with clipboards or contractors paid to watch camera footage.
“One application is measuring turning movements at roundabouts. This has been difficult in the past. The technology is able to classify vehicles, measure their speed, flow volumes, turning movements and observe trends.
“We’re deploying a number of them, across various sites where they provide previously unavailable insight on exactly what occurs at roundabouts or intersections in a more accurate way than was possible in the past.”
Elsewhere Smart City has provided sensor-equipped compacting bins deployed in the Far North District Council area. The compaction mechanism is key: bins compress waste so they hold roughly seven times the volume of a conventional bin before needing collection. Fewer collection runs means fewer heavy vehicle trips.
Smart bins, bus shelters, and improved flyover data: Individually they are modest improvements. But Ransley’s argument is that joined up and powered by physical AI, they represent the early architecture of a city that doesn’t just collect data about itself, but acts on it.
Smart City is a sponsor of the Herald’s Project Auckland report.