pyRDDLGym is a python framework for auto-generation of OpenAI Gym environments from RDDL description. The discrete time step evolution of variables in RDDL is described by conditional probability functions, which fits naturally into the Gym step scheme. Moreover, RDDL is a lifted description, thus modification and scaling up of environments to support multiple entities and different configuration become a trivial action instead of tedious task prone to bugs. We hope that pyRDDLGym will serve as a new wind in the reinforcement learning community by enabling easy rapid development of benchmarks with the unique expressive power of RDDL. By an explicit access to the model in the RDDL description, pyRDDLGym may also facilitate research on hybrid approaches for learning from interaction while leveraging model knowledge.

RDDL repository is home for all things RDDL. This repository is a collection of all verified RDDL files. Specifically it is an archive for the problems used in probabilistic and learning track of the International Planning Competitions.

This repository contains also visualizers for the pyRDDLGym framework, for appropriate domains.