Sriram Gopalakrishnan, Ph.D.

Sriram Gopalakrishnan, Ph.D.

AI Research Lead at J.P. Morgan Chase | Sequential-Decision Making Problems, Human-AI Interaction, & Generative AI

pyRDDLGym: Bridging RDDL Planning and Gym RL Environments

Published:

Pillar: Sequential Decision Making
pyRDDLGym RDDL Reinforcement Learning Open Source

Overview

pyRDDLGym bridges the gap between the Automated Planning community (which uses Relational Dynamic Influence Diagram Language - RDDL) and the Machine Learning / Deep RL community (which uses Gymnasium interfaces).

  • Automatically compiles RDDL relational domain definitions into vectorized Python simulation environments.
  • Enables direct benchmarking of symbolic planners against state-of-the-art policy gradient / actor-critic algorithms.
  • Widely adopted across international planning competitions and research labs.