It is common, that in big US cities 1 person wastes over 1oo hours per year, and tons of fuel. Hopefully, reserachers from Aston University can make long jams at trafic signals a thing of a past, using new AI techniques.
Adjusting the lights by scanning live video footage, AI, using deep reinforcement learning techniqes, learns on its previous outcomes and constantly improves and progresses. The AI system already surpassed all other, more traditional approaches in testing. But how do we know that?
The Aston University researches built a 3D traffic simulator program, where they control the AI’s award system. Everytime the car passes the traffic, an AI gets additional 1+ point. If the car has to stop at the jam, the AI gets -1 point.
The AI’s goal is to reach as many awards as possible. Therefore, learning on it’s own, the traffic AI progresses and learns autonomously on its previous patterns, aiming for another positive point.
The researchers hope on testing this system on real roads, even this year! Can’t wait to see that!
Sources:
Interesting, I also realli like the picture
That sounds really interesting! I wonder if they’re going to test it in some big cities like Los Angeles. Traffic over there is terrible for many reasons. I wonder if the changes will be able to be observed by a regular person.
AI definately needs to be implemented into our communication schedule