Cooperation of autonomous agents based on supervisory control of DES

The alternative approach to the synthesis of agent cooperation in multi agent systems (MAS) is proposed and presented. It is based on knowledge in both the area of DES control and the area of agent-based intelligent control. The agents are modelled by place/transition Petri nets (P/T PN). DES control knowledge is based on the P/T PN invariants and the Parikh's vector. The autonomous agents are forced to cooperate in order to achieve a goal prescribed for the whole group of agents. The theoretical introduction (mathematical description) of the approach is introduced and its applicability is illustrated on a case study and two simple examples. The approach seems to be suitable for material agents (inside of transport systems, manufacturing systems, etc.), social agents and so forth.