Embedding Intelligent Planning Capability to DEVS Models by Goal Regression Method
Abstract
This paper presents RG-DEVS (ReGression-Discrete EVent System Specification) formalism for embedding intelligent planning capability to DEVS models. RG-DEVS, an extension of classic DEVS, expands the classes of system models that can be represented in DEVS. The goal regression method of the artificial intelligence (AI) production system is exploited for the dynamic generation of the model’s sequential states and selection of state transition rules during model execution. Thus, generated states and rules form subgoals and an action plan, respectively. The mechanism for detecting missed execution of the plan and building an amended or new plan for performing corrective actions is also provided. An example application to a transportation vehicle is given to show how RG-DEVS can be applied.