Agent-Based Cooperative
Full Title
Agent-Based Cooperative Autonomy for Multi-UAV SystemsDescription
The Agent-Based Cooperative Autonomy for Multi-UAV Systems project has started at LASIGE, exploring how Large Language Models (LLMs) and intelligent agents can support autonomous coordination and decision-making in teams of unmanned aerial vehicles.
Multi-UAV systems have the potential to perform complex missions more efficiently than individual platforms, but their operation typically relies on predefined behaviours, centralized coordination or extensive human supervision. The project investigates an alternative approach in which UAVs are represented by autonomous software agents capable of reasoning about mission objectives, exchanging information and coordinating their actions dynamically.
At the core of the proposed architecture is the use of LLMs as a high-level reasoning layer. These models will be investigated as a mechanism for interpreting mission-level information, supporting task allocation and planning, and enabling agents to adapt their decisions as the operational environment changes. The resulting decisions will be integrated with the UAVs’ conventional mission execution and control systems.
The project will develop and evaluate a modular agent-based architecture for cooperative UAV autonomy, initially through simulation and subsequently in representative multi-UAV scenarios. Particular attention will be given to the reliability, robustness and scalability of LLM-supported coordination.
By combining multi-agent systems, Large Language Models and autonomous aerial robotics, the project aims to contribute towards more flexible and intelligent multi-UAV systems capable of performing increasingly complex cooperative missions.