Representation of Action Spaces in Multiple Levels of Detail

Abstract

Operators of complex dynamic systems like industrial process control or air traffic control should be able to predict future system states and also the actions needed to reach their goals. Models of human planning processes suggest that humans use plans with various levels of details to guide their actions. An assistance systems that would be able to propose plans in multiple levels of detail could lead to a better understanding of plans and thus to a better performance of the operator, because of the possibility to represent plans graphically similar to the operators mental model. In principal the visualization of abstract plan is more robust to changes of details, additionally less changes of visualized plans are necessary. This contribution describes a method to compute multiple representations of action spaces with different levels of detail, which could be used in an assistant system. The method assumes that the process can be described by discrete events and thus the action space of the process contains discrete states. The method is based on the state space generated by a Coloured Petri Net model. First the state space of this model will be computed, in the next step intermediate goal states will be identified, which will constitute the states of a more abstract state space. The arcs connecting these states will be generated according to the reachability of the states in the detailed state space. This procedure can be repeated and finally a representation of an action space in multiple levels of detail results.