Analysis of Geometric Obstacles Avoidance in Assembling Complex Products as a Decision-making Problem

Abstract

Computer aided assembly planning (CAAP) of complex products is an important and urgent problem of state-of-the-art information technologies. A configuration of the technical system imposes fundamental restrictions on the design solutions of the assembly process. The CAAP studies offer various methods for modelling geometric constraints. The most accurate results are obtained from the studies of geometric obstacles, which prohibit the part movement to the appropriate position in the product, by the collision analysis methods. An assembly of complex technical systems by this approach requires very high computational costs, since the analysis should be performed for each part and in several directions.The article describes a method for minimizing the number of direct checks for geometric obstacle avoidance. Introduces a concept of the geometric situation to formalize such fragments of a structure, which require checking for geometric obstacle avoidance. Formulates two statements about geometric heredity during the assembly. Poses the task of minimizing the number of direct checks as a non-antagonistic two-person game on two-colour painting of an ordered set. Presents the main decision criteria under uncertainty. To determine the best criterion, a computational experiment was carried out on painting the ordered sets with radically different structural properties. All the connected ordered sets are divided into 13 subclasses, each of which includes structurally similar instances. To implement the experiment, a special program has been developed that creates a random ordered set in the selected subclass, implements a game session on its coloration, and also collects and processes statistical data on a group of the homogeneous experiments.The computational experiment has shown that in all subclasses of the partial orders, the Hurwitz criterion with a confidence coefficient of 2/3 and that of Bayes-Laplace demonstrate the best results. The Wald and Savage criteria have demonstrated the worst results. In the experiment, a difference between the best and worst results reached 70%. With increasing height (maximum number of levels) of an ordered set, this difference tends to grow rapidly. In the subclass of pseudo-chains, all criteria showed approximately equal results.The game model of geometric obstacles avoidance allows formalizing data on geometric heredity and obtaining data on the composition and the minimum number of configurations, the test of which objectifies all existing-in-the-product geometric constraints on the movements of parts during assembly.