RisConFix: LLM-based Automated Repair of Risk-Prone Drone Configurations

TL;DR

This work tackles robustness degradation in drone flight control caused by risk-prone configurations by introducing RisConFix, an online, iterative system that uses an LLM to diagnose configuration–state relationships and generate corrective parameter updates in real time. The method combines a Runtime Anomaly Monitor with an LLM-based Repair loop, guided by official parameter ranges, and operates until stability is restored or a repair limit is reached. Evaluation on ArduPilot with 1,421 misconfigurations across two LLMs shows high effectiveness (up to 97% repair success) and reasonable efficiency (average repairs around 2). The results demonstrate real-time, automated repair of configuration-induced instabilities and suggest potential applicability to broader CPS domains.

Abstract

Flight control software is typically designed with numerous configurable parameters governing multiple functionalities, enabling flexible adaptation to mission diversity and environmental uncertainty. Although developers and manufacturers usually provide recommendations for these parameters to ensure safe and stable operations, certain combinations of parameters with recommended values may still lead to unstable flight behaviors, thereby degrading the drone's robustness. To this end, we propose a Large Language Model (LLM) based approach for real-time repair of risk-prone configurations (named RisConFix) that degrade drone robustness. RisConFix continuously monitors the drone's operational state and automatically triggers a repair mechanism once abnormal flight behaviors are detected. The repair mechanism leverages an LLM to analyze relationships between configuration parameters and flight states, and then generates corrective parameter updates to restore flight stability. To ensure the validity of the updated configuration, RisConFix operates as an iterative process; it continuously monitors the drone's flight state and, if an anomaly persists after applying an update, automatically triggers the next repair cycle. We evaluated RisConFix through a case study of ArduPilot (with 1,421 groups of misconfigurations). Experimental results show that RisConFix achieved a best repair success rate of 97% and an optimal average number of repairs of 1.17, demonstrating its capability to effectively and efficiently repair risk-prone configurations in real time.

Figures (4)

• Figure 1: Overview of RisConFix.
• Figure 2: Shortened example of an LLM prompt. Red text indicates prompt variables, the remaining text forms the static prompt template, and blue text marks the key elements.
• Figure 3: Example model response.
• Figure 4: Comparison of flight trajectories before and after applying the proposed method. The upper subfigure illustrates the flight trajectory with the original configuration, while the lower one presents the trajectory after applying the configuration update shown in Figure \ref{['fig:response example']}.