Technologies Leading to Unified Multi-Agent Collection and Coordination

Abstract

The problem of managing swarms of UAVs consists of multi-agent collection (i.e., distributed robust data fusion and interpretation) and multi-agent coordination (i.e., distributed robust platform and sensor monitoring and control). These two processes should be feedback-connected in order to improve the over-all quality of data be collected on suitable targets. This paper summarizes work proposed by Lockheed Martin Tactical Systems (LMTS) of Eagan MN and its subcontractor Scientific Systems Co., Inc. (SSCI) of Woburn MA, under contract F49620-01-C-0031 of the AFOSR Cooperative Control Theme 2. LMTS and SSCI have proposed to (1) develop a mathematical programming framework for hybrid systems analysis and synthesis, (2) develop a computational hybrid control paradigm, (3) develop transition-aware anytime algorithms for time-bounded synthesis, and (4) develop suitable modeling and cooperative control of UAV swarms for a SEAD-type mission. Regarding multi-agent collection, LMTS and SSCI will (4) develop new theoretical approaches for integrating multiplatform, multisensor, multitarget sensor management into hybrid systems theory; (5) investigate real-time nonlinear filtering for detecting and tracking low-observable targets; (6) develop new approaches to distributed, robust data fusion; and (7) develop a language for Multi-Agent Coordination broad enough to encompass Bayesian, Dempster-Shafer, and fuzzy-logic inference. The basis of our approach is twofold: (a) a novel hybrid-systems control architecture that integrates the best of the current approaches; and (b) a new foundation for multisensor-multitarget problems called “finite-set statistics.” Our approach integrates theoretically rigorous statistics (hybrid control, point process theory) with potential practicability (computational hybrid control, computational nonlinear filtering).