MORLBENCH Framework
Multiobjective Reinforcement Learning Benchmark Suite
Chair of data processing @ Technical University of Munich
Dominik Meyer meyerd@mytum.de, Johannes Feldmaier johannes.feldmaier@tum.de, Simon Wölzmüller ga35voz@mytum.de
This is the MORLBENCH Framework for algorithm and environment performance testing in multiple objective Reinforcement Learning.
written in Python 2.7
-
After cloning the repository you should run
python setup.py develop
in parent directory of the framework. (provided you have python 2.7 on your system installed. Otherwise download it(Python Website)) -
You can use every class in
morlbench/morl_agents.py
andmorlbench/morl_problems.py
where agents stands for algorithms and problems for the scenarios you want to test in. -
For first examples using the framework relate to
morlbench/scratchpad
To run test type python -m unittest2 test.py
in morlbench/test