PhD thesis,
DESIGN AND IMPLEMENTATION OF AN OBJECT-ORIENTED GEOMETRIC CONSTRAINT SOLVER
UNIVERSITY OF FLORIDA, (2001)
Abstract
In this thesis, I address the main reasons for the underutilization of
geometric constraints in today's CAD/CAM design and assembly
systems. I discuss advantages, functionality and implementation
challenges of the FRONTIER, a constraint solving system built by
Dr. Meera Sitharam's group. My contribution to the FRONTIER project
includes some of the design and most of the implementation of 3 key
modules of FRONTIER called Universal Transfer Unit, DR-Planner and
Simple Solver.
FRONTIER fully enables (a) the use of complex, cyclic, spatial
constraint structures and (b) feature-based design.
Although Issue (a) requires close-to-optimal decomposition methods
more sophisticated than merely detecting triangles and other rigid
patterns, Issue (b) requires such optimal decompositions to
additionally be consistent with an underlying feature hierarchy and
admit entirely independent and local manipulation of features and
subassemblies.
FRONTIER enables complex variational geometric constraint
structures to be mixed in freely with other intuitive representation
languages for design and assembly, such as explicit drafting on
screen, B-rep, CSG, history based specifications, etc;
FRONTIER deals efficiently with under and overconstrained systems,
nongeometric, engineering (equation) constraints, and persistent
naming of shape elements that are not explicitly specified, but may or
may not be generated as a result of constraint resolution; it also
aids maintenance of multiple constraint views and constraint
reconciliation; and admits parametric constraint solving as a special
case.
Each of FRONTIER's components is individually and easily portable into
existing CAD/CAM systems.
