Integrated Optimization of a Smart Hanger for Garment Inspection Using Multi-Objective Genetic Algorithm

Abstract

This paper presents the formulation and application of Multi-Objective Genetic Algorithm (MOGA) for the design and optimization of a smart hanger, which simulates the manual handling of garments to facilitate the inspection process. Due to the symmetrical nature of clothes, the hanger is expected to expand only half of the clothes. Thus, a four-link robot consisting of three revolute joints and one prismatic joint is employed to model the hanger. With a Proportional and Derivative (PD) controller, the hanger can regulate the link positions by adjusting the applied torques and force. The objective functions are to (1) maximize the link portions within the "effective regions" at final state, (2) minimize the force or torque required to finish the stretching process, and (3) minimize the settling time of the stretching process. In addition to the control gains, the lengths of the four links as well as the desired movements of the joints are the design variables in this optimization problem. The required transient behavior of the system is defined by the constraints on the settling time and the maximum overshoot. Besides, to prevent the clothes from being destroyed by the links when stretching, geometrical constraints are imposed to the motions of the links. MOGA is applied to tackle this integrated optimization problem. The optimization results are presented in the form of Pareto solutions. After analysis, optimal parameters are selected, and numerical simulations are conducted. Results show the feasibility of the hanger.