Objective: This study examined the impact of increasing automation replanning rates on operator performance and workload when supervising a decentralized network of heterogeneous unmanned vehicles. Background: Futuristic unmanned vehicles systems will invert the operator-to-vehicle ratio so that one operator can control multiple dissimilar vehicles, connected through a decentralized network. Significant human-automation collaboration will be needed due to automation brittleness, but such collaboration could cause high workload. Method: Three increasing levels of replanning were tested on an existing, multiple unmanned vehicle simulation environment that leverages decentralized algorithms for vehicle routing and task allocation, in conjunction with human supervision. Results: Rapid replanning can cause high operator workload, ultimately resulting in poorer overall system performance. Poor performance was associated with a lack of operator consensus for when to accept the automation"s suggested prompts for new plan consideration, as well as negative attitudes towards unmanned aerial vehicles in general. Participants with video game experience tended to collaborate more with the automation, which resulted in better performance. Conclusion: In decentralized unmanned vehicle networks, operators who ignore the automation"s requests for new plan consideration and impose rapid replans both increase their own workload and reduce the ability of the vehicle network to operate at its maximum capacity. Application: These findings have implications for personnel selection and training for futuristic systems involving human collaboration with decentralized algorithms embedded in networks of autonomous systems.

This paper presents the outdoor flight test results of a decentralized multi-UAV system supervised by a human operator. The system balances the roles of the human operator and the UAV autonomous behaviors with the objective of maximizing the execution performance. The operator manages the mission by inputting and modifying tasks instead of controlling individual UAVs. The Consensus-Based Bundle Algorithm (CBBA) is used as a real-time, scalable, dynamic multi-agent multi-task planning algorithm to allocate tasks approved by the operator to UAVs. A team of three quadrotors and one fixed wing UAV collaborated in an operationally relevant scenario supporting a cargo UAV resupply mission. Thirteen of fourteen multi-UAV outdoor flight test trials successfully accomplished the mission objectives. The framework was shown to be robust to system failures and degradations commonly encountered during field testing primarily because of health monitoring and management tools that were incorporated in the design. Instances of task allocation and path planning churning were observed which are linked to uncertainties of operating outdoors. Lessons learned during flight test operations are highlighted as they are relevant to other similar types of systems and missions.

These results have important implications for personnel selection and training for futuristic multi-UV systems under human supervision. Although gamers may bring valuable skills, they may also be potentially prone to automation bias. Priming during training and regular priming throughout missions may be one potential method for overcoming this propensity to overtrust automation.

For complex systems that embed automation, but also rely on human interaction for guidance and contingency management, holistic models are needed that provide for an understanding of the individual human and computer elements, and address the critical interactions of such complex systems. Discrete event simulation (DES) models and system dynamics (SD) models are two different approaches that can be used to address these requirements. Both modelling approaches can support the designers of future autonomous vehicle (AV) systems by simulating the impact of alternate designs on vehicle, operator, and system performance. However, the DES modelling approach is likely best suited for using probabilistic distributions to accurately model an operator who is a serial processor of discrete tasks, as well as an environment with randomly occurring events. The SD modelling approach is better suited for modelling continuous performance feedback that is temporally dependent and is affected by qualitative variables such as trust.