Petri nets (Murata, 1989) are widely employed to model and control AMSs since they are suitable to detect deadlocks of a system and develop a policy to deal with them (Ezpeleta et al., 1995; Ghaari et al., 2003; Huang et al., 2001; Li and Zhou, 2009; Reveliotis, 2007). Deadlock prevention is one of the most important approaches whose goal is to impose constraints on a system to prevent it from reaching deadlock states (Ezpeleta et al., 1995; Fanti and Zhou, 2004, 2005; Jeng and Xie, 2005; Piroddi et al., 2008, 2009). In this case, the computation is carried out o-line and once the control policy is established and applied, deadlocks can no longer occur.