NASA Awards Heron Systems Phase 1 Research Grant for Autonomous Collaborative Task Allocation

To maximize the advantages of UAS, technologies enabling the cooperation, coordination, and interoperability through heterogeneous multi-robotic systems (MRS) will be vital in providing the significantly required operational characteristics necessary to adapt to unknown environmental conditions. A prohibiting challenge of diversified applications with MRS architectures is the ability to dynamically assess and reassess available resources, coordinate, and task entities.

Existing state of the art addresses this type of problem by assigning a unique resource to a specific task while optimizing across the set of available choices (i.e. who does what/when/where). In spatiotemporally unstructured environments, the ability to dynamically assess/reassess available resources and assign tasking is much more challenging and heretofore underexplored. What is missing is a mature mechanism for tying the agents together into a cohesive team, the glue.

This research project is to build Generalized Logistics for Unstructured Environments (GLUE), a software framework enabling the explicit coordination of the RTA problem through a novel implementation of decentralized algorithms capable of achieving at a minimum local optimality. Recent advancements in the size, weight, and power (SWaP) capabilities of computing platforms, MEMS sensors, and other robotic peripherals are capable of migrating such technologies from the nation’s research laboratories to the forefront of possibilities in fieldable autonomy.

Capable of dynamically assessing and reassessing the available resources in an MRS, GLUE shall enable optimized task assignments according to the situational awareness of the combined team and status of the mission. GLUE performs the planning and tasking (i.e. Logistics) required by MRS agents while leveraging and preserving the individual agents’ ability to determine their own approach. By structuring the data exchange and enforcing global rules, the coordination of agents in the presence of multiple competing options is rigorously examined and optimized. Crucially, approaching MRS in this manner preserves the benefits offered by heterogeneous agents, opening further opportunities for tuning the construction of MRS teams for optimization within specific missions and/or environments. The result is an algorithmic framework that is principally agnostic to the scenario, tasking, and autonomous platform it is applied to.

Leveraging Heron Systems Inc’s existing multi-vehicle robotic framework MACE to provide the critical infrastructure, the research will be grounded in a quantitative assessment and demonstration of feasibility using a search and rescue scenario as the motivation. Once successful and integrated within MACE, the cumulative solution offers novel capabilities and applications leveraging existing COTS unmanned systems.