this post was submitted on 08 Jun 2024
2117 points (98.9% liked)
Programmer Humor
18961 readers
1074 users here now
Welcome to Programmer Humor!
This is a place where you can post jokes, memes, humor, etc. related to programming!
For sharing awful code theres also Programming Horror.
Rules
- Keep content in english
- No advertisements
- Posts must be related to programming or programmer topics
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Whilst it's a shame this implementation sucks, I wish we would get intelligent traffic light controls that worked. Sitting at a light for 90 seconds in the dead of night without a car in sight is frustrating.
That was a solved problem 20 years ago lol. We made working systems for this in our lab at Uni, it was one of our course group projects. It used combinations of sensors and microcontrollers.
It's not really the kind of problem that requires AI. You can do it with AI and image recognition or live traffic data but that's more fitting for complex tasks like adjusting the entire grid live based on traffic conditions. It's massively overkill for dead time switches.
Even for grid optimization you shouldn't jump into AI head first. It's much better long term to analyze the underlying causes of grid congestion and come up with holistic solutions that address those problems, which often translate into low-tech or zero-tech solutions. I've seen intersections massively improved by a couple of signs, some markings and a handful of plastic poles.
Throwing AI at problems is sort of a "spray and pray" approach that often goes about as badly as you can expect.
I can see the headlines now: "New social media trend where people are asking traffic light Ai to solve the traveling salesman problem is causing massive traffic jams and record electricity costs for the city."
(I know I'm two months late)
To back up what you're saying, I work with ML, and the guy next to me does ML for traffic signal controllers. He basically established the benchmark for traffic signal simulators for reinforcement learning.
Nothing works. All of the cutting edge reinforment algorithms, all the existing publications, some of which train for months, all perform worse than "fixed policy" controllers. The issue isn't the brains of the system, its the fact that stoplights are fricken blind to what is happing.
That was a problem solved seventy years ago. If there's no one around, just go. No one cares.