Metamodels for Planar 3R Workspace Optimization

Robotic workspace optimization is a central element of robot system design. To formulate the optimization problem, the complex relationships between design variables, tuning parameters, and performance indices need to be accurately and efficiently represented. The nature of the relationships suggests that metamodels, or models of the models, should be used to derive suitable objective functions. A comparison of two metamodeling techniques for robotic workspace optimization problems for several trial cases suggests that non-uniform rational B-spline models, derived from computer graphics and computer-aided design techniques, are as suitable as response surface models to solve planar 3R workspace optimization problems. Promising nonlinear modeling results with B-spline models suggest future work is justified and performance gains can be realized.