Intelligent Agent for Modeling and Processing Decisional Workflows in Logistics

Abstract

The authors present the design and some implementation trials of Atlas, a new reasoning and decision making assistant used for processing complex and heterogeneous procedural workflows. Benefiting from a multicore implementation, Atlas includes different solving engines that are selected according to the intrinsic complexity of the problem being processed. The operational knowledge of Atlas is accessed through 2 different views. In an analytical view, the knowledge is modeled on elementary if-then rules, which are processed by a resolution engine written in the Soar architecture. A synthetic view offers a pictorial representation of all the knowledge, and in particular, shows the inter-dependence of the rules and their procedural references. In addition to allowing an efficient processing, the system checks the coherence of the knowledge and produces a justification of the decision with respect to relevant operational procedures.