Anchoring Symbols to Vision Data by Fuzzy Logic
S. Coradeschi and A. Saffiotti
Intelligent agents embedded in physical environments need the ability
to connect, or anchor, the symbols used to perform abstract
reasoning to the physical entities which these symbols refer to.
Anchoring must rely on perceptual data which is inherently affected by
uncertainty. We propose an anchoring technique based on the use of
fuzzy sets to represent uncertainty, and of degree of subset-hood to
compute the partial match between signatures of objects. We show
examples where we use this technique to allow a deliberative system to
reason about the objects (cars) observed by a vision system embarked
in an unmanned helicopter, in the framework of the WITAS project.
S. Coradeschi and A. Saffiotti.
Anchoring Symbols to Vision Data by Fuzzy Logic.
In: A. Hunter and S. Parsons (eds) Symbolic and Quantitative
Approaches to Reasoning and Uncertainty: Proc. of the ECSQARU Conf
LNCS 1638, Springer, DE, 1999.