The Starling Crossing

A new, high tech approach to the pedestrian crossing.

I love the way engineers and designers are always looking at ways to use technology to improve aspects of our day to day lives that we take for granted, particularly in the urban environment.

In recent years, I have noticed several design consultancies, universities and city authorities looking for ways to improve the humble pedestrian crossing, including:

I think this new design concept, The Starling Crossing, commissioned by Directline Insurance and created by design consultancy, Umbrellium, is something really special. It’s a really interesting approach that creates a crossing in real time as where and when it is needed.

Woman on a smart crossing with an umbrella
The LEDS have to be visible in daylight, and slip-free in the rain

As a proof of concept using existing technology, they recently built this full-scale prototype in South London. The temporary surface had to support the weight of vehicles (obs!), remain slip-free in pouring rain and, importantly, it had to display markings bright enough to be seen during daytime.

So what does it do and how does it work?

Well, it uses a neural network of cameras that tracks moving objects, identifies them as vehicles/cyclists/pedestrians, fixes their locations, works out their trajectories and velocities, and here’s the clever bit… predicts their track, so that it can configure the surface lighting to manage their interaction.

That means if someone is distracted by their phone and wanders too close to the road or a kid chases a football into the road, it can instantly paint a  buffer zone around them to warn drivers. If it’s raining and the road is slick, it’ll paint a bigger buffer zone.

Man chases ball into smart street
The crossing can pre-empt pedestrians’ movement and where their eyeline is most likely to be, which help to ensure it grabs attention to limit danger.

And, of course as a neural network, it learns. Interestingly it uses the principles of stigmergy* so that it can recognise pedestrian patterns and ‘desire lines’ over long term.

It already compensates for quiet and busy periods in the day – for example, when it’s quiet the crossing may only appear when someone approaches,and in a location that it has learned over time is the safest. When it’s busy it can create a wider than usual crossing for longer to handle more pedestrians – but the ‘desire line’ pattern recognition will enable it to work out that, for example, if most people exiting a tube station end up walking diagonally across the road towards a park entrance, it can reconfigure the crossing as a diagonal in the right place.

Now you have got to admit, that’s pretty clever!

* Stigmergy is where the name comes from; STigmergic Adaptive Responsive LearnING Crossing)

(Photos: courtesy Umbrellium)


Join the Conversation →

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.