GBL proposes a Decentralized Autonomous Collaboration Tool (DACT) system using autonomous collaboration algorithms allowing teams of heterogenous weapon systems to coordinate on mission objectives in a dynamic, mission environments GBL will leverage staff experience in developing Artificial Intelligent (AI) / Machine Learning (ML) algorithms and build on technologies developed under SBIR N04-174 for the EA-18G Electronic Combat Decision Support System (ECDSS) capability and SBIR N181-018 for the Rapid Artificially Intelligent Strike Mission Planner (RASP) capability. The DACT approach will use a Distributed Intelligent Agent-based open software architecture and will identify platform agnostic critical core agents, platform specific interface agents that can facilitate integration across heterogeneous systems, and learning agents for offline training. To accomplish diverse missions within an Anti-Access and Area Denial (A2/AD) environment, novel methods will be employed using red and blue proxy agent models in combination with a hybrid Consensus Based Bundling Algorithm (CBBA) to enable decentralize operations of DACT-enabled weapon system assets to optimize the use of limited communication. This will enable inference of the behavior of actors within the mission environment and coordinate tasking/targeting between assets when communications allow. DACT will utilize a modified Particle Swarm Optimization (PSO) algorithm in conjunction with an expert system approach or a Deep Neural Net (DNN) to enable collaborative in-flight autonomous updates of mission plans to counter changes in a dynamic mission environment. DACT will enable multiple weapon systems to autonomously collaborate to determine which platforms are best suited to fulfill mission tasking and account for platform specific capabilities, assigned mission role/profile, and respective positions within the mission environment.
