As You Wish: Mission Planning with Formal Verification using LLMs in Precision Agriculture
Marcos Abel Zuzu\'arregui, Stefano Carpin
Why It Matters
What makes this one worth your time
This work addresses the challenges of operating robotic systems in agriculture by improving the usability and reliability of mission planning through natural language, which could facilitate broader adoption of autonomous technologies in the field.
A novel mission planner that integrates LLMs and formal verification for precision agriculture.
Summary
The paper presents an enhanced mission planning system for precision agriculture that utilizes large language models (LLMs) and incorporates feedback loops with linear temporal logic (LTL) to improve the accuracy of mission specifications derived from natural language inputs.
Key contributions
- Development of a mission planner that synthesizes plans from natural language descriptions.
- Introduction of feedback loops in the planning architecture to enhance specification accuracy using LTL.
- Empirical evaluation of the system's strengths and limitations in generating LTL formulas.
Notable insights
- The integration of LTL for verification within an LLM-driven planning architecture is a clever approach to mitigate ambiguities in natural language.
- Using two different commercial LLMs for specification and verification tasks may help reduce bias and improve robustness.
Possible limitations
- Potential challenges in the LLM's ability to consistently generate high-quality LTL formulas may not be fully addressed.
- Not stated in the abstract.
Abstract
arXiv:2606.18519v2 Announce Type: replace-cross Abstract: Though robotic systems are now being commercialized and deployed in various industries, many of these systems are highly specialized and often require an advanced skill set to operate and ensure they perform as instructed. To mitigate this problem, we recently introduced a mission planner leveraging LLMs to synthesize mission plans in precision agriculture based on mission descriptions provided in natural language. While the system demonstrates impressive performance, it also suffers from the inherent ambiguities of natural language. In this paper, we extend our system to address this issue by introducing multiple feedback loops in the planning architecture that leverage linear temporal logic (LTL) to ensure the mission planning system meets the specifications formulated by the user while still using natural language. To mitigate potential bias, this is achieved by using two different commercial LLMs in charge of the specification and verification subtasks. Through extensive experiments, we highlight the strengths and limitations of integrating mission verification into a fully autonomous pipeline, particularly regarding an LLM's ability to generate valuable LTL formulas, and show how our proposed implementation addresses and solves these challenges.