AutoSafeLTL introduces a novel framework that leverages large language models to automatically convert natural language requirements into safety-compliant formal specifications for cyber-physical systems.

• Achieved 0% violation rate of safety constraints in formal specifications
• Implemented a self-supervised framework that evaluates and refines LTL formulas
• Demonstrated how LLMs can effectively bridge the gap between natural language and formal logic representation
• Enhanced safety verification in critical systems through automated specification generation

This research significantly advances security in cyber-physical systems by automating the traditionally error-prone process of creating formal specifications, ensuring systems operate within safety boundaries while reducing human expertise requirements.

Automatic Generation of Safety-compliant Linear Temporal Logic via Large Language Model: A Self-supervised Framework