By Tristan Greene
IBM was granted a patent for an AI-powered traffic management system earlier this month.
What it is: According to the patent, IBM wants to develop a system for controlling traffic lights that relies on cameras. A computer views real-time footage of traffic in each direction and then calculates the optimum traffic pattern.
Why it matters: Have you ever sat at a four-way intersection for three full minutes staring at a red light with a mixture of bewilderment and anger because there isn’t another vehicle in sight? Yeah, that.
We asked IBM Master Inventor Steve Hobson why we couldn’t just write better regular non-intelligent programs to fix the traffic light system, he says:
I’d say most people who “use” signal-controlled road junctions — be they motorists, pedestrians, cyclists, or whatever — would agree that a human observing a junction could help the thing work better.
A human can absorb masses of information that would suggest adjustments to the signal sequence and/or timing. Consider a few:
1. There’s a group of children standing in the pouring rain waiting to cross the road
2. There’s a big queue of traffic on one of the approaches but not others
3. There’s a very slow moving pedestrian (perhaps elderly) part way across the road
4. Someone has fallen part way across the road
5. A backlog of vehicles is building up because they can’t turn across the path of oncoming vehicles
6. There’s a red light stopping a vehicle even though the streets are deserted
Developing protocols for such a complex environment is difficult to say the least and impractical if each junction is to be managed taking into account it’s unique characteristics. But the fact that this sort of information is “easy” for a human to absorb and use suggests AI has the potential to help.
The bottom line: IBM should fast track this patent plan. Because traffic sucks and we hate it.
Tristan Greene is an author at The Next Web.
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius