Distributed Adaptive Macroscopic Ensemble Task Allocation of Heterogeneous Robot Teams in Dynamic Environments

Robot teams that build highly predictive models of environments like the ocean require effective methods to collect sensor data. A robot team can collect this data, but the challenge is continuously assigning robots to informative sampling locations. Existing methods design specialized solutions for individual robots that lack the necessary flexibility. Instead, recent extensions allow macroscopic ensemble allocation to adapt with environmental changes, but currently relies on impractical centralized assumptions. To address the need for centralization, we introduce a heterogeneous formulation with unmanned aerial vehicles and a communication task where a small portion of the surface team ferries information between robots performing spatially distributed sampling tasks. We show a distributed adaptive macroscopic allocation solution with similar performance to the centralized strategy.