search

flow-project / flow

Computational framework for reinforcement learning in traffic control
MIT License
1.07k stars 376 forks source link

Get started with reinforcement learning and SUMO #925

Open Tabbi7 opened 4 years ago

Tabbi7 commented 4 years ago

Dear all, I have to implement a scenario in which I want to use SUMO simulator to prevent road accidents caused by lane changing, sudden braking, sudden acceleration and so on. I want to use reinforcement learning and I faced the Flow Project.

For each simulation step, I can export from SUMO all the features related to a single simulation saving them in a csv file that provides - for each step - a cell which contains the list of the colliding vehicles, if some vehicles are colliding. So, I would like to prevent and anticipate collisions from previous steps in order to give precious alerts to the drivers. What can I do?

I'm new in the magic world of reinforcement learning and SUMO. Can you please help me?

ncarrara commented 4 years ago

if you are new to RL, I would start with a simple toy game and grid worlds. If you want to dive into the autonomous driving domain directly, I wish you luck !