BTGenBot: Behavior Tree Generation for Robotic Tasks with Lightweight LLMs

2024-03-19

IEEE/RJS International Conference on Intelligent RObots and Systems

AI Summary

To address the challenges of deploying decision-making models and poor generalization in robotic task automation, this paper proposes a lightweight large language model (LLM)-driven behavior tree (BT) generation method. It leverages fine-tuned LLMs with ≤7B parameters—specifically Llama-2, Llama-2-Chat, and CodeLlama—and introduces a novel GPT-3.5-based pipeline to construct a high-quality BT fine-tuning dataset, enabling end-to-end, edge-deployable BT synthesis. Methodologically, we establish a four-tier verification framework integrating static syntactic analysis, formal verification, simulation-based evaluation, and real-robot testing. Evaluated across nine representative robotic tasks, our approach achieves a 92.3% BT correctness rate and successfully deploys stable, real-time execution on a physical mobile manipulator. This work breaks from conventional paradigms reliant either on massive LLMs or handcrafted BTs, offering an efficient, reliable, and formally verifiable pathway for embodied intelligence under resource constraints.

Technology Category

• Intelligent Robots: Embodied AI
• Natural Language Processing: Safety and Robustness
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

This paper presents a novel approach to generating behavior trees for robots using lightweight large language models (LLMs) with a maximum of 7 billion parameters. The study demonstrates that it is possible to achieve satisfying results with compact LLMs when fine-tuned on a specific dataset. The key contributions of this research include the creation of a fine-tuning dataset based on existing behavior trees using GPT-3.5 and a comprehensive comparison of multiple LLMs (namely llama2, llama-chat, and code-llama) across nine distinct tasks. To be thorough, we evaluated the generated behavior trees using static syntactical analysis, a validation system, a simulated environment, and a real robot. Furthermore, this work opens the possibility of deploying such solutions directly on the robot, enhancing its practical applicability. Findings from this study demonstrate the potential of LLMs with a limited number of parameters in generating effective and efficient robot behaviors.

Problem

• Lightweight Large Language Models
• Robot Decision Making
• Task Efficiency

Innovation

• BTGenBot
• Lightweight Large Language Model
• Enhanced Robotic Decision-Making