THE UNDETERMINISTIC MANIPULATION OF SOLID MODELS FOR ROBOT PROGRAM SYNTHESIS

In automatic robot program synthesis the number of variables that should be taken into consideration become prohibitively numerous. Due to the ambiguity and sheer size of items to be considered conventional computation methods cannot satisfactorily solve the problem. A Neural Network model that acquires data from a Solid Modeling data base, combines the completeness of information provided by solid modeling with the uncertainty encountered in the grouping process to perform geometrical classification of objects. The capabilities of Neural Networks to learn non-geometrical patterns in the grasping process, are yet to be achieved. Much progress needs to be made in both the neural model complexity and the computing machinery power before real intelligent program synthesis can be achieved.