Combinatorial Planning with Numerical Parameter Optimization for Local Control in Multi-agent Systems

Planning with numeric state variables and goal systems today still poses a challenging task within the field of computational intelligence. In this paper a two-tier planning system is presented that enables the optimization of continuous numeric action parameters in combinatorially enumerated plans. It allows resorting to a “satisficing” strategy by means of partial execution and subsequent repair of infeasible plans in order to deal with certain difficulties concerning reliable and fast detection of action applicability that arise when planning with real-valued action parameters. The functioning of the system is evaluated in a multi-agent simulation of a shop floor control scenario with focus on the effects the possible problem cases and the satisficing approach have on attained plan quality.