Coco/R was used to define the language and generate the scanner and parser. The parser constructs the tree that holds the system (see the coco/r definition file for details).

Fuzzy rules can contain complex antecedents, that is the systems constructed here are not necessarily Mamdani. The notion of antecedent predicates (antecedent snippets) has been used. This concept will be very useful in the next version of the script, where fuzzy behaviours will be tackled.

Each node in the tree can deliver its equivalent in c# code so that an optimized system can be compiled and then used at run-time.

A number of example files are provided with the project. Best example is System1.txt, where a robot controller for a machine having three ultrasonic sensors is defined. This controller has been actually implemented in practice, but that’s another story :-)

Last edited Apr 15, 2009 at 10:24 AM by cgafa, version 2