Speech-guided unmanned aerial vehicle control system based on large language model-structured execution graphs with expert rule constraints
Year: 2026
Type: article
Abstract: As the use of Unmanned Aerial Vehicles (UAVs) expands into industrial and safety-critical applications, there is an increasing need for intuitive speech-driven control systems that reduce reliance on predefined command sets or manual interfaces. Natural language instructions, however, often exhibit ambiguity, logical inconsistency, and a lack of structured representation, which limits their direct applicability to UAV operations. This research develops a speech-guided UAV control system that leverages large language models (LLMs) to construct execution graphs from natural language commands, augmented with expert rule validation for reliable and transparent aerial autonomy. These graphs capture semantically organized control sequences and are refined through expert rule constraints that verify logical consistency, eliminate infeasible actions, and resolve dependency issues. The use of in-context learning enables the LLM to generalize across diverse expressions without requiring additional fine-tuning. The system was evaluated through three experiments targeting language understanding, graph structuring, and rule-based validation. It achieved up to 95.4% accuracy on complex spoken commands, structured 1,000 natural instructions into executable graphs with over 98% accuracy, and corrected all structural issues using expert-defined rules. Real-time performance was maintained with an average latency under two seconds. This work establishes a practical framework for integrating LLMs into voice-controlled UAV systems while ensuring robustness through rule-based validation.
Authors: Seok-Hun Choi, Zeen-Chul Kim, Seok-Jun Buu
Institution: Gyeongsang National University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,200
OpenAlex ID: W7214526159