How Much Do You Trust Your Stoplight?

Readers who worry too much about The Matrix or Terminator may want to skip this post.

Scientists are now envisioning a new approach to managing traffic flow by turning traffic lights into self-organizing nodes in a network. Move over SkyNet; say hello to RoadNet.

The plan is to build sensors into each traffic light so that it can automatically adjust its timing to best fit traffic conditions right where it is. What’s new about that – we’ve had locally ‘aware’ lights for a while now, right? Well, yes, that’s true and that’s why the other crucial factor in this approach is giving the light the ability to dynamically coordinate its actions with other lights in the network. There is no central supercomputer trying to over-think things here. The whole approach is based on distributing the system’s intelligence as locally as possible while ensuring enough connection for the system to keep an “eye” on the big picture. Autonomy balanced with connection. Sounds familiar.

Yes, stuff like this does sound a bit scary but I like two things about this idea. One, it’s grounded in networked, decentralized intelligence and control, which means it’s more likely to work. And two, this information systems approach might just reduce traffic delays by 10-30% – and do that without hugely expensive investments in concrete, asphalt and steel infrastructure.

Yes, it’s possible we could simply squander this added capacity much the way thoughtlessly adding an extra lane to a bridge eventually just fills it up with more vehicles. With a little creativity and some good transit planning, however, a system like this might well be ‘tweaked’ to give the edge to high occupancy vehicles. Yeah, you have to do it without mucking too much with local autonomy or adding too much complexity. That’s where you have to apply a little creative thinking, and if you read the actual paper (which is pretty dense), you’ll see that that’s exactly what the system’s designers have in mind.

Get that right and you cut time and carbon.