In: LNAI 1638: Symbolic and Quantitative Approaches to Reasoning and Uncertainty, pages 352-pp. European Conference, ECSQARU'99, London, UK, July 1999. Proceedings / A. Hunter, S. Parsons (Eds.) --- Springer Verlag, 1999.
Abstract: A method is proposed for determining the state of a dynamical system modeled by a Petri net, using observations of its inputs. The initial state of the system may be totally or partially unknown, and sensor reports may be uncertain. In previous work, a belief Petri net model using the formalism of evidence theory was defined, and the resolution of the system was done heuristically by adapting the classical evolution equations of Petri nets. In this paper, a more principled approach based on the Transferable Belief Model is adopted, leading to simpler computations. An example taken from an intelligent vehicle application illustrates the method throughout the paper.