Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Diverse Planning with Simulators via Linear Temporal Logic

Mustafa F. Abdelwahed, Alice Toniolo, Joan Espasa, Ian P. Gent

TL;DR

The paper tackles the limitation of single-solution planning in complex, non-symbolic environments by introducing FBI_LTL, a simulator-based diverse planner that encodes semantic diversity with Linear Temporal Logic (LTL). It integrates a modified Iterated Width search with LTL-based novelty constraints to generate semantically distinct plans, demonstrated on PDDLGym, Puzznic, and NASim benchmarks. Key contributions include the FBI_LTL algorithm, an LTL-based behaviour-space model with cost-bound and goal-order features, and empirical evidence that semantically diverse planning improves coverage across diverse domains. This work advances planning in realistic, non-symbolic domains by enabling agents to select among multiple meaningful strategies rather than a single solution.

Abstract

Autonomous agents rely on automated planning algorithms to achieve their objectives. Simulation-based planning offers a significant advantage over declarative models in modelling complex environments. However, relying solely on a planner that produces a single plan may not be practical, as the generated plans may not always satisfy the agent's preferences. To address this limitation, we introduce $\texttt{FBI}_\texttt{LTL}$, a diverse planner explicitly designed for simulation-based planning problems. $\texttt{FBI}_\texttt{LTL}$ utilises Linear Temporal Logic (LTL) to define semantic diversity criteria, enabling agents to specify what constitutes meaningfully different plans. By integrating these LTL-based diversity models directly into the search process, $\texttt{FBI}_\texttt{LTL}$ ensures the generation of semantically diverse plans, addressing a critical limitation of existing diverse planning approaches that may produce syntactically different but semantically identical solutions. Extensive evaluations on various benchmarks consistently demonstrate that $\texttt{FBI}_\texttt{LTL}$ generates more diverse plans compared to a baseline approach. This work establishes the feasibility of semantically-guided diverse planning in simulation-based environments, paving the way for innovative approaches in realistic, non-symbolic domains where traditional model-based approaches fail.

Diverse Planning with Simulators via Linear Temporal Logic

TL;DR

The paper tackles the limitation of single-solution planning in complex, non-symbolic environments by introducing FBI_LTL, a simulator-based diverse planner that encodes semantic diversity with Linear Temporal Logic (LTL). It integrates a modified Iterated Width search with LTL-based novelty constraints to generate semantically distinct plans, demonstrated on PDDLGym, Puzznic, and NASim benchmarks. Key contributions include the FBI_LTL algorithm, an LTL-based behaviour-space model with cost-bound and goal-order features, and empirical evidence that semantically diverse planning improves coverage across diverse domains. This work advances planning in realistic, non-symbolic domains by enabling agents to select among multiple meaningful strategies rather than a single solution.

Abstract

Autonomous agents rely on automated planning algorithms to achieve their objectives. Simulation-based planning offers a significant advantage over declarative models in modelling complex environments. However, relying solely on a planner that produces a single plan may not be practical, as the generated plans may not always satisfy the agent's preferences. To address this limitation, we introduce , a diverse planner explicitly designed for simulation-based planning problems. utilises Linear Temporal Logic (LTL) to define semantic diversity criteria, enabling agents to specify what constitutes meaningfully different plans. By integrating these LTL-based diversity models directly into the search process, ensures the generation of semantically diverse plans, addressing a critical limitation of existing diverse planning approaches that may produce syntactically different but semantically identical solutions. Extensive evaluations on various benchmarks consistently demonstrate that generates more diverse plans compared to a baseline approach. This work establishes the feasibility of semantically-guided diverse planning in simulation-based environments, paving the way for innovative approaches in realistic, non-symbolic domains where traditional model-based approaches fail.
Paper Structure (11 sections, 7 figures, 2 tables, 3 algorithms)

This paper contains 11 sections, 7 figures, 2 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Background
  3. Behaviour Planning with Simulators
  4. Behaviour Space
  5. Forbid Behaviour $\text{Iterative}_\texttt{LTL}$
  6. Experimental Setup, Evaluation & Findings
  7. Experiment Setup
  8. Benchmark Problems
  9. Results and Discussion
  10. Related work
  11. Conclusions & Future work

Figures (7)

  • Figure 1: IW(i) operation. Dotted trees indicate a breadth-first search, and i denotes the novelty pruning threshold. Dotted crosses indicate nodes that failed the novelty check.
  • Figure 2: The extended Iterated Width BFS search filters nodes that fail the novelty test or violate a given LTL formula. The cross represents pruned nodes, while letters represent boolean variables set to true in the current state.
  • Figure 3: The Puzznic game's dynamics are as follows: the cursor moves across the box to the left, and gravity is applied since the box has no objects below it. Once the falling box touches the matched boxes, they are cleared.
  • Figure 4: Violin plots for all planning problems experiments. Planners are ordered as follows: $\texttt{FBI}_\texttt{LTL}$, $\texttt{FBI}_\texttt{LTL}^\texttt{naive}$.
  • Figure 5: Three plans for solving Puzznic's first level. Check supplementary material for the plans' actions.
  • ...and 2 more figures

Theorems & Definitions (1)

  • Definition 1: diversity planning problem