TeLoGraF: Temporal Logic Planning via Graph-encoded Flow Matching

2025-05-01

AI Summary

STL-guided general-purpose planning faces challenges due to scarce large-scale annotated data and ineffective temporal logic representation methods. Method: We introduce the first 200K-scale dataset of paired STL specifications and demonstrations; propose a syntax-tree graph modeling framework for STL, coupled with graph neural network encoding to enable end-to-end learning over arbitrary system dynamics; and integrate continuous normalizing flow matching to support open-template generalization and out-of-distribution (OOD) robust reasoning. Contributions/Results: Evaluated across five simulation environments—including 2D dynamics, Franka Panda robotic arm, and Ant quadruped—we achieve significantly higher STL satisfaction rates than baselines. Our method accelerates inference by 10–100× over classical STL planners and, for the first time, enables cross-platform deployment and zero-shot generalization to unseen templates and OOD specifications.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Multiagent Systems: Multiagent Planning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Learning to solve complex tasks with signal temporal logic (STL) specifications is crucial to many real-world applications. However, most previous works only consider fixed or parametrized STL specifications due to the lack of a diverse STL dataset and encoders to effectively extract temporal logic information for downstream tasks. In this paper, we propose TeLoGraF, Temporal Logic Graph-encoded Flow, which utilizes Graph Neural Networks (GNN) encoder and flow-matching to learn solutions for general STL specifications. We identify four commonly used STL templates and collect a total of 200K specifications with paired demonstrations. We conduct extensive experiments in five simulation environments ranging from simple dynamical models in the 2D space to high-dimensional 7DoF Franka Panda robot arm and Ant quadruped navigation. Results show that our method outperforms other baselines in the STL satisfaction rate. Compared to classical STL planning algorithms, our approach is 10-100X faster in inference and can work on any system dynamics. Besides, we show our graph-encoding method's capability to solve complex STLs and robustness to out-distribution STL specifications. Code is available at https://github.com/mengyuest/TeLoGraF

Problem

• Learning solutions for general STL specifications using GNN and flow-matching
• Addressing lack of diverse STL datasets and effective encoders
• Improving STL satisfaction rate and inference speed in planning

Innovation

• Uses GNN encoder for temporal logic extraction
• Applies flow-matching for general STL solutions
• Demonstrates 10-100X faster inference than classical methods