Resource Optimisation for Distributed Teams of Manned Aircraft and Drones

With the significant increase in onboard computing capabilities, modern aerial robotic systems, can execute a broad range of perception and control algorithms simultaneously. Moreover, more and more, they are also deployed as heterogeneous collaborative teams, where manned and unmanned assets need to collaborate in a manned-unmanned teaming concept. This introduces the challenge of determining the optimal distribution of cognitive processes across aerial platforms, edge computing nodes, and cloud-based services. In this paper, we propose a novel load distribution methodology tailored to the aerial domain. The approach adopts a decentralized framework for allocating perception and control processes by evaluating communication parameters (e.g., bandwidth, latency, and cost), the computational capabilities of the drones and supporting infrastructure (including CPU, GPU, memory, and storage performance), and the real-time delivery requirements of high-quality output data. The proposed methodology is validated in a simulated environment, demonstrating promising performance and scalability in handling dynamic operational conditions.