Decentralized Uav Architectures

There has been significant recent research activity attempting to streamline Unmanned Aerial Vehicle (UAV) operations and reduce staffing in order to invert the current many-to-one ratio of operators to vehicles. Centralized multiple UAV architectures have been proposed where a single operator interacts with and oversees every UAV in the network. However, a centralized network requires significant operator cognitive resources. Decentralized multiple UAV networks are another, more complex possible architecture where an operator interacts with an automated mission and payload manager, which coordinates a set of tasks for a group of highly autonomous vehicles. While a single operator can maintain effective control of a relatively small network of centralized UAVs, decentralized architectures are more scalable, particularly in terms of operator workload, and more robust to single points of failure. However, in terms of operator workload, the ultimate success of either a centralized or decentralized UAV architecture is not how many vehicles are in the network per se, but rather how many tasks the group of vehicles generates for the operator and how much autonomy is on board these vehicles. Task-based control of UAV architectures with higher degrees of autonomy (i.e., decentralized networks) can mitigate cognitive overload and reduce workload. Mutually exclusive boundaries for humans and computers in multiple UAV systems should not be the goal of designers for either centralized or decentralized architectures, but rather more effort needs to be spent in defining mutually supportive roles such that humans and computers complement one another.