NeSIG: A Neuro-Symbolic Method for Learning to Generate Planning Problems

TL;DR

NeSIG is a novel neuro-symbolic approach that automatically generates diverse, valid, and challenging planning problems across domains using deep reinforcement learning, reducing human effort and outperforming domain-specific generators.

Contribution

This paper introduces NeSIG, the first domain-independent method for automatic planning problem generation using deep reinforcement learning, capable of producing more difficult and diverse problems.

Findings

NeSIG generates problems 15.5 times more difficult on average.

It produces valid and diverse problems across multiple domains.

The method generalizes to larger problems than those seen during training.

Abstract

In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper we propose NeSIG, to the best of our knowledge the first domain-independent method for automatically generating planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on three classical domains, comparing our approach against handcrafted, domain-specific instance generators and various ablations. Results show NeSIG is able to automatically generate valid and diverse problems of much greater difficulty (15.5 times more on geometric average) than domain-specific generators, while simultaneously reducing human effort when compared to them. Additionally, it can generalize to larger problems than those seen during training.