Autonomous behavioral strategy and optimal centralized guidance for on-orbit self assembly

Swarms of satellites, intended as a large number of spacecraft performing a common mission, appear to be particularly apt for an application of the so-called behavioral control strategies in the astrodynamics. The basic concept of the behavioral strategies lies in the imitation of natural swarms of social animals, like flocks of birds, which are characterized by a global group organization acquired in absence of a centralized control. In particular, an on-orbit self assembly manoeuvre is analyzed. Since every satellite takes its own decisions being able to consider only a small number of neighbor satellites and limited intercommunication, full autonomy is achieved. However optimality in terms of control cost and required time for complete self assembly is an issue which is not even taken in consideration. In this work, the behavioral results are compared with a typical centralized control, with an active leader which is able to detect the position and the dynamics of each swarm member and to compute the optimal guidance towards the fittest slot around the leader. In such a way, the optimality of the swarm manoeuvre is reached, at the cost of a great complication in the system architecture of the leader. Special attention will be dedicated to the filtering strategies, in particular to particle filters, apt to solve navigation problems including a large number of close members, when measurements ambiguity can occur. An extensive simulation campaign, for various orbit configurations and swarm dimensions, is performed in order to clearly assess and quantify the advantages and drawbacks of the behavioral and centralized strategies.