Hi Pavel

[stephenlp]

Hi Pavel, This isn't a real pull request. I just couldn't find any other way to reach out to you:)

I've just found your code here (RL-on-SUMO) and think it's interesting. However, looks like I can't even run the example there (Israel_zipping). Would you be kind to give me a quick help here?

Nice to know you and your work here. and Happy new year!

Regards, Stephen (stephenlowis@ymail.com)

[NikuKikai]

@stephenlp Sorry for late response, but it's been too long so I'm afraid I cannot help you with details. This repo is a simple hands-on sample so I think you could reimplement it which might be more clear. Several steps to do:

• use SUMO to create a project
• use 'traci' to communicate with SUMO from python
• code a simple RL algorithm using latest torch or jax instead TF1.x, that's painful lol. OR just use openAI's baselines.
• wrap SUMO simulation environment as a class in a openAI gym style, which can be smoothly connected to your RL codes.