Language Models For Generalised PDDL Planning: Synthesising Sound and Programmatic Policies
This addresses planning in AI for domains like robotics or logistics by providing a novel method to generate sound policies without external verification, though it is incremental in applying LMs to a specific formalism.
The authors tackled the problem of using language models for PDDL planning by synthesizing provably sound Python programs as policies, achieving results where their LMPlan planner solved more PDDL problems than existing planners and LM approaches under fixed constraints, handling problems with hundreds of objects.
We study the usage of language models (LMs) for planning over world models specified in the Planning Domain Definition Language (PDDL). We prompt LMs to generate Python programs that serve as generalised policies for solving PDDL problems from a given domain. Notably, our approach synthesises policies that are provably sound relative to the PDDL domain without reliance on external verifiers. We conduct experiments on competition benchmarks which show that our policies can solve more PDDL problems than PDDL planners and recent LM approaches within a fixed time and memory constraint. Our approach manifests in the LMPlan planner which can solve planning problems with several hundreds of relevant objects. Surprisingly, we observe that LMs used in our framework sometimes plan more effectively over PDDL problems written in meaningless symbols in place of natural language; e.g. rewriting (at dog kitchen) as (p2 o1 o3). This finding challenges hypotheses that LMs reason over word semantics and memorise solutions from its training corpus, and is worth further exploration.