Barron Associates receives Phase II Award to Develop Graph Neural Networks (GNN) for Collaborative UxS Applications

Barron Associates received a Phase II Applied SBIR award to continue its work entitled, Graph-based Collaborative Autonomy for Intelligent Agents. The primary goal of the project is to develop decentralized swarm behaviors using GNN methods to enable heterogeneous swarms of autonomous agents to adaptively collaborate, communicate, and formulate control actions in contested areas with limited network connectivity and bandwidth. Existing solutions to multi-vehicle, or swarm, collaboration  often rely on centralized solutions which can offer optimal solutions but scale poorly to large numbers of agents, depend on persistent communication, and suffer from prolonged computation times for complex problems. In contrast, the proposed technology offers an efficient and decentralized approach for autonomous and heterogeneous swarms to operate in contested areas with limited network connectivity and bandwidth. The Phase II work leverages and expands foundational technologies developed during the Phase I program and past projects and will culminate with field testing the GNN algorithms with a heterogenous swarm of UGV and UAS multi-rotor vehicles.