“DARPA’s OFFensive Swarm-Enabled Tactics (OFFSET) program envisions future small-unit infantry forces using swarms comprising upwards of 250 unmanned aircraft systems (UASs) and/or unmanned ground systems (UGSs) to accomplish diverse missions in complex urban environments. By leveraging and combining emerging technologies in swarm autonomy and human-swarm teaming, the program seeks to enable rapid development and deployment of breakthrough capabilities.” Reference:  https://www.darpa.mil/work-with-us/offensive-swarm-enabled-tactics

DARPA’s vision for OFFSET hypothesizes the ability of a swarm to be controlled at a very high level, generally guided by tactics akin to the tactics deployed by soldiers and marines. A derived requirement emerges wherein the swarm is able to autonomously reassess its progress against the tactical objective and react accordingly. However, existing methods for performing this capability are greatly underexplored in practice. 

Our work with the OFFSET program matured existing academic research of decentralized asynchronous consensus auctioneer strategies into a Resource and Task Allocation capability for swarms. This approach relaxes the requirements for communications topology, situational awareness homogeneity across the swarm, time horizon, and centralized control. In so doing, this RTA solution provides a crucial capability required for any multi-robotic system to achieve tactical swarming outcomes, like those desired by the OFFSET program. 

In June of 2019, Heron Systems participated in the second field experiment (dubbed FX-22) at Ft. Benning, GA on the Combined Arms collective Training Facility (CACTF). The field experiement attempted to demonstrate the operation of up to 50 UxVs (air and ground) as part of an urban assault scenario. At FX-2, Heron Systems successfully integrated our RTA solution into a third party hardware stack (Northrop Grumman, Mission Systems). On successive days of testing, the RTA-enabled swarm demonstrated autonomous tasking, re-tasking after hardware failure, network self-healing, and multiple rounds of resource allocation during a prolonged scenario.


Ongoing Research

Heron Systems defines a collaborative subset of SoS consisting of heterogeneous agents with significantly diversified skills used to address the decomposition of a complex task in an optimal (as defined by an objective function) way a coalition. Coalitions are formulated to provide the SoS a capability to robustly respond to spontaneous opportunities resulting from adaptation of the situational awareness when executing over-the-horizon and reduce overall system coordination/communication burden.

GLUE – GLUE facilitates adapting agent action policies in dynamic response to vulnerabilities of the swarm. Foundational methods (e.g. describing the diversity of a coalition) can be leveraged to generate further abstract methods that are beneficial in assessing the effectiveness of the composition to achieve minimum readiness, redundancy, and diversity.