AI solutions for traffic coordination

[florian_z]

Hey guys,

I know that this is kind of off-topic but I'm convinced that you can help me on it.
I'm currently working on a research project for university and we're trying to find solutions to improve future traffic coordination in big cities (up to the year 2030).
I believe that proper design and programming of AI programs will be of particular importance for solving this issue.

So, to come to the point: Do you have any ideas (algorithms, AI design, what else could you think about?) for improving future traffic coordination? If not yourself, do you know anybody who is a specialist and could help me on that?

I’m very much looking forward to your answers and thank you for your help in advance!

Florian

[mac]

Suggestion: set up an AI contest in which you provide a game engine that models the world (roads, junctions, peak hours... and have participants to develop bots to manage the entities generated by the game engine...

...uh... it reminds me of something but I can't recall what!

PS: just be clear in the contest rules that the university reserves itself the right to look at bets performing bots algorithms to develop non-commercial software out of it...

[dimkadimon]

Here is a good solution using Reinforcement Learning:

Silvia Richter, Douglas Aberdeen, Jin Yu: Natural Actor-Critic for Road Traffic Optimisation. NIPS 2006: 1169-1176
books.nips.cc/papers/files/nips19/NIPS2006_0844.pdf