MATSim DRT RL

Repository to combine DRT rebalancing and reinforcement learning.

All important source code is contained in src/main. The java code implements a RemoteRebalancingStrategy that accepts rebalancing instructions from a client. The reinforcment learning part is implemented in python and contained in the src/main/python.

Installation

Install python requirements in:

pip install -r src/main/python/requirements.txt

Running scenarios

Run the java class org.matsim.run.RunDrtRLScenario to start the MATSim run. Start the python process in parallel, it will send rebalancing instructions to MATSim.