search

LucasAlegre / sumo-rl

Reinforcement Learning environments for Traffic Signal Control with SUMO. Compatible with Gymnasium, PettingZoo, and popular RL libraries.
https://lucasalegre.github.io/sumo-rl
MIT License
746 stars 201 forks source link

Some questions of hierarchical reinforcement learning to sumo_rl. #156

Closed locker2153 closed 1 year ago

locker2153 commented 1 year ago

I tried to configure a top-level agent for multi-agent , but I found that in the sumoenvironment, it seems impossible to change the observation and actions for a top-level agent. In the rllib library, the pettingzoo environment seems to only support all agents using the same observation and action. In order to solve this problem I have used 'supeisuit' trying to modify the top agent of obs as well as the actions. It is a pity that did not work. Multi-agent environment has not been identified as a Parallelenv Do you have any good ideas ? plz

LucasAlegre commented 1 year ago

I suggest you ask the PettingZoo developers for help. You can find them in the corresponding channel in the FaramaFoundation discord: https://discord.com/invite/PfR7a79FpQ

locker2153 commented 1 year ago

I suggest you ask the PettingZoo developers for help. You can find them in the corresponding channel in the FaramaFoundation discord: https://discord.com/invite/PfR7a79FpQ

3Q ，i will try