You are here
Control of Cooperative Engagements with Robust and Distributed Optimization (Concerto(TM))
Title: Analyst
Phone: (805) 968-6787
Email: gcollins@toyon.com
Title: Director of Finance and C
Phone: (805) 968-6787
Email: mlindbery@toyon.com
Contact: Nancy R Lewis
Address:
Phone: (805) 893-4034
Type: Nonprofit College or University
Toyon Research Corporation and the University of California at Santa Barbara (UCSB) propose to develop decision and control algorithms for cooperative agents operating under an intermittent asynchronous communication network. Agents will be expected to perform a variety of cooperative missions such as persistent intelligence, surveillance, and reconnaissance (ISR), target search, acquisition, and tracking, area denial, and battle damage assessment under dynamically-varying bandwidth limitations, intermittent communications, and limited information. The key feature of the proposed cooperative control system is the use of an estimator-based distributed control (EBDC) architecture, allowing each UAV to internally model and anticipate the actions of the others UAVs. Under this EBDC architecture, the control system can gracefully scale from essentially centralized control (complete and perfect inter-vehicle communication) to completely distributed control (no inter-vehicle communication). The estimator-based distributed architecture models a centralized design by forcing each individual agent in the system to compute future control actions for all agents, as a function of the agents' positions and the global information state (GIS). The quality and amount of communication between agents determines the accuracy of each agent's GIS estimate. Each agent's GIS estimate, in turn, determines the accuracy of each agent's control estimates for the other agents. In this way, the EBDC algorithms transition from an essentially centralized solution to a completely decentralized solution based on the communication rate and information available to each agent within the cooperative network. The performance of the system will be guaranteed to degrade gracefully with the level of information available to each agent, because the EBDC algorithms optimize the control actions based on the existing information instead of attempting to optimize the communication network to allow better throughput. In the Phase II effort, Toyon and UCSB will enhance and refine the EBDC algorithms developed under Phase I. Implementation of these algorithms will continue in both MATLAB and Toyon's higher-fidelity SLAMEM(TM) simulation tool. We will test and demonstrate their effectiveness in cooperative search and tracking scenarios under varying levels of information throughput. Bandwidth limitations, time delays with uncertain intervals, and network topology limitations will be incorporated into the simulation. The extent of reliable communications required to ensure robust cooperative control will be evaluated.
* Information listed above is at the time of submission. *