- Project Green Light uses travel data from Maps users to model driver behavior at specific intersections.
- Google uses its analysis to make recommendations about adjusting signal timing to optimize traffic flow.
- The goal is to expand to hundreds of cities within the next few years.
No one’s going to blame you for feeling a little burnt out on AI. Artificial intelligence seems like it’s been at the dead center of the tech zeitgeist for so long that it’s enough to have you feeling nostalgic for NFTs. And while there are absolutely situations where it’s being used poorly, or being used in ways that feel threatening, there are others where AI trying to smarten up an ancient, dumb system sounds like the very best thing that could happen for everyone involved. That is just what Google’s been up to with Project Green Light, an AI-driven effort to improve traffic efficiency, and this week the company offers a little insight into how well that’s been going.
The problem Green Light wants to help with is simple to understand: optimally controlling traffic signal scheduling is difficult and expensive. It’s hard because cities lack a complete data set on all the traffic flowing through a given intersection, and efforts to acquire data, like with car-sensing inductive loops buried in the pavement or rubber-tube traffic counters draped across the roadway, are limited in what they can gather. But Google has Android and all its detailed location data to mine, giving it a massive dataset of driver habits. Project Green Light is basically Google realizing “if we crunch these numbers, we can probably make some really accurate recommendations on adjusting traffic light timing to cities.” And that’s just what’s been happening.