Collaboration coordination via cellular automata

Engineering activities are seldom solitary due to their high degree of cross discipline dependencies. Of special consideration is collaboration in the early exploratory phases of design and development. The work presented here considers collaboration as a special case of the multi-agent resource allocation problem in light of the fact that potential collaborators play the roles of both resource and consumer. The benefits of collaboration emerge from the dynamic interactions of the participants and as such are not limited solely to human agents. The paper examines the use of totalistic cellular automata with reinforcement learning local update rules for the purposes of distributed and fair collaboration coordination in a manner which maximises the overall utility of the organisation or agent community as measured by the degree to which the set of collaboration allocations approximate Paraeto optimality. This research represents the first time that reinforcement learning local update rule driven cellular automata have been used for the purpose of collaboration coordination within multi-agent systems.