A genetic planner for assembly automation

Genetic algorithms have been widely applied to solve combinatorial optimization problems such as the traveling salesman problem (TSP), job shop scheduling and other sequencing problems. Although the assembly sequence planning (ASP) problem also belongs to this category of problems, the GA application seems impossible to this problem because of its highly constrained nature. Most assembly planners introduced in research performed extensive search for the exact solution, had a part-based or subassembly-based decision process, and/or were restricted to linear sequence solutions, where only one part is connected at a time. As a result, they suffered a high sensitivity to large increases in product part, which is the case of almost all industrial products, thus having very narrow application range. We introduce a novel approach for solving the ASP problem by applying a modified GA version. The main concept is to cluster the search space into families of similar sequences, where each family contains only one feasible sequence-representative-satisfying the problem constraints. The GA is applied on a population of representatives, and its stochastic output which might contain family members which are not representatives is transformed back into a population of corresponding representatives, by a new transformation function. The new algorithm generates a best solution without searching the complete space. It can generate linear and nonlinear solutions, and its decision process is performed on a sequence population-basis rather than a part-basis