Investigating the role of fuzzy sets in a spatial modeling framework

This paper outlines a method of constructing a spatial data collection agent that is capable of adopting a spatial sampling strategy along a transect network in real-time, based on minimal prior knowledge of the measured surface and a concept of spatial heterogeneity. The agent uses fuzzy reasoning to decide upon the appropriate distance to travel along the current transect before making a new measurement. Fuzzy functions for the sets "widely spaced" and "immediate neighbourhood" are constructed "on the fly" based on the deviation of the most recent measurement from the value anticipated from the original crude surface. The two functions are combined to produce a Π curve, and the minimum distance between samples that produces a maximum fuzzy membership, i.e. that balances sample sparseness and sample density, is used to determine the next sample point along the transect. The exact form of the fuzzy functions and their parameterization will be designed using a spatially explicit simulation modeling framework to produce experimental performance measures. It is anticipated that the resulting agent will be sufficiently elegant to be easily incorporated into semi-autonomous field acquisition systems, and ultimately contribute to autonomous acquisition systems