Alberto Pozanco

Turn your planner into a bug-finder: this work shows how to minimally modify planning tasks to expose and eliminate flaws by making the task provably unsolvable.

LLMs can evolve surprisingly effective, interpretable Python planners that rival state-of-the-art classical planners, at a fraction of the computational cost.

LLMs can slash grounding time in classical planning by orders of magnitude, simply by sniffing out irrelevant objects and actions *before* the grounding process even starts.