Hacker News Reader

Comment by Ajedi32 | Hacker News Reader

I'm guessing they wouldn't need to fiddle with an app; the software would just automatically signal their approach based on GPS/navigation info.

I agree theoretically a camera + machine-learning based system could work just as well; I've always wondered why that wasn't standard on traffic lights these days given that the hardware would probably cost a few hundred dollars (if procurement for said hardware were done competently, which I suppose as you said, it probably wouldn't be).

View on HN

Are we are the stage where it is normal to be giving governments always on background access to our gps location voluntarily?

I mean, if your concern is government, your already shit out of luck between your cellphone pinging cell towers, license plate readers, CCTV cameras, Ring cameras, Wi-Fi pinging, etc.

(I would agree that I wouldn't necessarily want fine-grained gps location data being sent to the lowest bidding for-profit traffic light monitoring company.)

I can think of a few ways to do this in a privacy-preserving way. The traffic light doesn't need to know your location or who you are, just whether there is a car approaching. But yeah good chance whoever designed this didn't put that much thought into it.

> Are we are the stage where it is normal to be giving governments always on background access to our gps location voluntarily?

Yes. See Palantir for examples.

why would you need machine learning? just detect a minimum of one bright light source, assume it's a headlight, and switch the lights. What the worse a false positive could do? this doesn't need AI

Comparable systems based on inductive loops work during the day. A design based on detecting headlights wouldn't. Software/compute is cheap and as you said reliability is not a safety concern.

We’ve had camera-triggered intersections for years now in Estonia. They seem to work equally well in daylight as well as darkness. I don’t think you need any ML for it either, you basically only need to detect motion within a predefined region which can be done with traditional CV algos quite reliably.

Reply on news.ycombinator.com