In: Robotics and Flexible Manufacturing Systems, pages 417-426. Elsevier Science Publ., 1992.
Abstract: A great attention has been recently paid in the AI community to the problemof providing a semantics for model-based diagnosis. In particular, the attention focused on logical formalisms and onthe declarative characterization of the setof solutions for a diagnostic problem.In this paper we propose an innovative approach in which the diagnostic process is captured within a framework based on the formalism of Petri nets. An interesting aspect consists in the fact that the model is obtained using a particular type of deterministic Petri net. Moreover, we discuss how the formalization of the diagnostic process can be obtained in terms of reachability in the Petri net model and can be implemented by means of backward simulation and exploiting the explicit parallelism in the model. This allows us to consider both declarative and procedural aspects in the formalization of model-based diagnosis