Comparison and Application of Metaheuristic Population-Based Optimization Algorithms in Manufacturing Automation

Abstract

The paper presents a comparison and application of metaheuristic population-based optimization algorithms to a flexible manufacturing automation scenario in a metacasting foundry. It presents a novel application and comparison of Bee Colony Algorithm (BCA) with variations of Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) for object recognition problem in a robot material handling system. To enable robust pick and place activity of metalcasted parts by a six axis industrial robot manipulator, it is important that the correct orientation of the parts is input to the manipulator, via the digital image captured by the vision system. This information is then used for orienting the robot gripper to grip the part from a moving conveyor belt. The objective is to find the reference templates on the manufactured parts from the target landscape picture which may contain noise. The Normalized crosscorrelation (NCC) function is used as an objection function in the optimization procedure. The ultimate goal is to test improved algorithms that could prove useful in practical manufacturing automation scenarios.