Improved Generalized Planning with LLMs through Strategy Refinement and Reflection
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914410734288896 |
|---|---|
| author | Stein, Katharina Hodel, Nils Fišer, Daniel Hoffmann, Jörg Katz, Michael Koller, Alexander |
| author_facet | Stein, Katharina Hodel, Nils Fišer, Daniel Hoffmann, Jörg Katz, Michael Koller, Alexander |
| contents | LLMs have recently been used to generate Python programs representing generalized plans in PDDL planning, i.e., plans that generalize across the tasks of a given PDDL domain. Previous work proposed a framework consisting of three steps: the LLM first generates a summary and then a strategy for the domain, both in natural language, and then implements that strategy as a Python program, that gets debugged on example planning tasks. In that work, only one strategy is generated and passed directly to the program generation. If the strategy is incorrect, its implementation will therefore result in an incorrect generalized plan. Here, we introduce an approach that generates the strategy in the form of pseudocode and enables automatic debugging of the pseudocode, hence allowing us to identify and fix errors prior to the generation of the generalized plan itself. Additionally, we extend the Python debugging phase with a reflection step prompting the LLM to pinpoint the reason for the observed plan failure. Finally, we take inspiration from LLM code generation to produce several program variants and pick the best one. Running experiments on 17 benchmark domains with two reasoning and two non-reasoning LLMs, we show that these extensions substantially improve the quality of the generalized plans. Our best performing configuration achieves an average coverage of 82% across the domains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_13876 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Improved Generalized Planning with LLMs through Strategy Refinement and Reflection Stein, Katharina Hodel, Nils Fišer, Daniel Hoffmann, Jörg Katz, Michael Koller, Alexander Artificial Intelligence Computation and Language LLMs have recently been used to generate Python programs representing generalized plans in PDDL planning, i.e., plans that generalize across the tasks of a given PDDL domain. Previous work proposed a framework consisting of three steps: the LLM first generates a summary and then a strategy for the domain, both in natural language, and then implements that strategy as a Python program, that gets debugged on example planning tasks. In that work, only one strategy is generated and passed directly to the program generation. If the strategy is incorrect, its implementation will therefore result in an incorrect generalized plan. Here, we introduce an approach that generates the strategy in the form of pseudocode and enables automatic debugging of the pseudocode, hence allowing us to identify and fix errors prior to the generation of the generalized plan itself. Additionally, we extend the Python debugging phase with a reflection step prompting the LLM to pinpoint the reason for the observed plan failure. Finally, we take inspiration from LLM code generation to produce several program variants and pick the best one. Running experiments on 17 benchmark domains with two reasoning and two non-reasoning LLMs, we show that these extensions substantially improve the quality of the generalized plans. Our best performing configuration achieves an average coverage of 82% across the domains. |
| title | Improved Generalized Planning with LLMs through Strategy Refinement and Reflection |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2508.13876 |