Cost Effective Optimal Robotic Assembly Sequence Generation through Artificial Intelligence Techniques & CAD Automation

Abstract

Robots are widely used in manufacturing industries for various operations to reduce the time and increase the quality of the final product. Among all the operations, the assembly is one of the important operations in manufacturing, which dwell solely 15-20% of the overall cost of the product. The use of robots in assembly operations is wide across in manufacturing industries for automation to attain high productivity. In order to achieve quality assembly, an optimal assembly sequence is required. Attaining an optimal sequence is a difficult task as Assembly Sequence Planning (ASP) Problem is a discrete optimization problem. To generate a feasible sequence, assembly attributes (Liaison data/Contact data) are required. Most of the researchers use only contact data/ liaison data to generate feasible sequences, which leads to the infeasible solution during the practical assembly operation. Moreover, many of the researchers use mathematical models to generate assembly attributes from the product drawing or Computer Aided Design (CAD) models, which involve a lot of complexity and need high skills. Apart from extracting the assembly attributes, using them for validating the assembly sequence increases the complexity of the ASP problem. Many methods like mathematical, soft computing based and CAD-based are developed to obtain optimal sequences. Use of mathematical models increases the complexity to generate optimal assembly sequences. Moreover, the applied methods try to generate feasible sequences only. On the other side, soft computing techniques are developed to solve the ASP problem. These Artificial Intelligence (AI) techniques are successful in achieving the optimal assembly sequences for large part assemblies also. However, these AI techniques are somehow struck at local optima during execution and cannot able to generate all the solutions. Meanwhile, CAD-based automatic generation of assembly sequences has been developed to generate optimal assembly sequences. These methods are successful in generating all optimal assembly sequences, but these methods consume large search space even for the small number part assemblies. Keeping the above limitations in view, in the first phase of research work integration of automatically extracted assembly attributes using CAD interface with AI techniques have been developed to generate the optimal assembly sequences. In the second phase of research work, Hybrid Algorithms (HA) algorithms are developed to solve the ASP problem. The developed algorithms are applied to various industrial products to compare with the existed HA algorithms in terms of execution time. In the third phase of research work, CAD-based assembly subset detection method has been developed to generate optimal assembly sequences automatically. This method is applied to various industrial products, and the results are compared with the existing CAD based methods and artificial intelligence methods in terms of execution time and a number of iterations. In order to validate the proposed methodology, different products have been designed in CATIA V5 software. The code for the proposed methodology has been written in computer programs using VB Script to interface with the CATIA environment. The automated process of assembly subset detection method for generating the optimal assembly sequences is validated by considering the outcome results of different assembly products. In the fourth phase of research work, the integration of Design For Assembly (DFA) concept has been included with Artificial Intelligence to generate reduced levels of the assembly sequence with modified topology. The Proposed methodology is implemented on various industrial products to reduce levels of the assembly sequence.