This paper presents a novel solution to the problem of autonomous task allocation for a self-organizing satellite constellation in Earth orbit. The method allows satellites to cluster themselves above targets on the Earth’s surface. This is achieved using Coupled Selection Equations (CSE) - a dynamical systems approach to combinatorial optimization whose solution tends asymptotically towards a Boolean matrix describing the pairings of satellites and targets which solves the relevant assignment problems. Satellite maneuvers are actuated by an Artificial Potential Field method which incorporates the CSE output. Three demonstrations of the method’s efficacy are given - first with equal numbers of satellites and targets, then with a satellite surplus, including agent failures, and finally with a fractionated constellation. Finally, a large constellation of 100 satellites is simulated to demonstrate the utility of the method in future swarm mission scenarios. The method provides efficient solutions with quick convergence, is robust to satellite failures, and hence appears suitable for distributed, on-board autonomy.
Garrie S. Mushet,