Hello Michal,
Machine learning is a very wide area. I would like to know what kind of machine learning you are using for the self-learning agent for traffic lights. Maybe you could share some thoughts.
I recently made a control application based on reinforcement learning in Python. It’s just a quick try but it seems to work. The goal was to make the perfect control application, so
it’s far from complete and not trained to get the best results. I only use this for other machine learning studies. The control application works on the vehicle data produced by SUMO.
Best regards,
Martin Barto
Sr. Verkeerskundige
Mobiliteit
![](jpgay9tWPfGGk.jpg)
Vialis is een Koninklijke
VolkerWessels onderneming.
Postbus 184, 3990 DD Houten | tel. +31648510557 mob. +31648510557
.
.
Van: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx>
Namens Michal Lis
Verzonden: vrijdag 2 november 2018 12:34
Aan: sumo-user@xxxxxxxxxxx
Onderwerp: [sumo-user] Environment of machine learning
Hello community. I would like to write a self-learning agent for traffic light environment. It seemed for me that SUMO is the most popular one. However I cannot find any SUMO traffic model that was used to machine learning several times
by different users. It would be very helpful if I could compare my results with someone's else.