Description

1. Introduction 

Multi-robot systems (MRSs) have been widely investigated in the recent years due their appealing characteristics in terms of flexibility, redundancy, fault tolerance, and the possibility they offer for using distributed sensing and actuation. Nowadays, a large literature on MRSs exists and covers aspects like system architecture, task allocation, team heterogeneity, and communication (e.g. see Parker (2008) and references therein). The most recent approaches for MRSs’ control mainly focus on distributed or decentralized techniques for networked robots; these are MRSs where each robot has limited sensing and communication ranges and can only use data from its onboard sensors or information received from its direct neighbors. In this case, the main control problem concerns how to move the single robots using only local information and in order to achieve global tasks that may depend on the overall system configuration. Decentralized algorithms for networked robots cover control and communication problems such as, for example, controlling the MRS centroid, variance, and orientation (Belta and Kumar, 2004), cooperative reconfiguration in response to a sensed, distributed environment (Ögren et al., 2004), flocking in presence of switching communication topologies (Tanner et al., 2007), formation control via communication and coordination (Fax and Murray, 2004), and rendezvous and formation control while keeping connectivity (Ji and Egerstedt, 2007). A wide overview on networked robots can be found in (Kumar et al., 2008) or in the recent books (Bullo et al., 2009; Gazi and Passino, 2010). Dealing with networked (static or mobile) systems poses the problem of how to reach an agreement regarding a variable, either exogenous or depending on the state of single agents; such problem, known as the consensus problem, has been recently investigated by a wide number of researchers. Recent studies are summarized in Mesbahi and Egerstedt (2010) and Ren and Beard (2008), while in Olfati-Saber et al. (2007) and Ren et al. (2007) the consensus algorithms are investigated with an emphasis on robustness, time-delays, and performance bounds. The work in Jadbabaie et al. (2003) deals with the stability analysis of several decentralized strategies to achieve emergent behaviors as moving in the same direction despite the absence of centralized coordination. A non-linear stationary consensus protocol for fixed topologies is presented in Bauso et al. (2006) and further extended in Cortés (2008) for a more general class of consensus functions