International Journal of Computer and Information Technology (ISSN: 2279 – 0764) Volume 03 – Issue 02, March 2014 www.ijcit.com 328 A Solution to Multi Criteria Robot Selection Problems Using Grey Relational Analysis Babatunji Omoniwa Department of Computer Engineering COMSATS Institute of Information Technology Islamabad, Pakistan E-mail: tunjiomoniwa {at} yahoo.com Abstract— Grey relational analysis (GRA) is a vital instrument suitable for optimal selection which can be used for modelling, forecasting and decision making. Grey theory gives reliable solution of systems in which the model is poor with little or incomplete information. This paper aims to use GRA for solving Multi Criteria Robot Selection Problems (MCRSPs). In this paper, GRA steps are implemented using a fast computational tool on two practical cases and results were compared with previous methodologies to confirm the validity of the GRA approach. Results show that the distinguishing coefficient has minimal impact on the GRA solution, thereby making this approach appropriate for accurate modelling of MCRSPs. Keywords- MCRSPs; grey theory; grey relational analysis; distinguishing coefficient I. INTRODUCTION Over the years, rapid growth in computers and applied sciences has led to an explosion of scientific innovations amongst which robots, computer aided machines and other automated systems belong. As technology becomes more sophisticated, the role of man in the manufacturing process is gradually replaced with industrial robots. Robots can function in environments dangerous to humans such as radioactive, heated, toxic and noisy settings. Due to the high cost of acquiring and implementing these industrial robots, manufacturers often seek an optimal solution. It is observed that there are so many mutually conflicting performance criteria, like dynamic accuracy, repeatability, speed, load capacity, program flexibility, handling coefficient, memory capacity, manipulator reach, supplier’s service quality etc. that influence the robot selection decision [1]. Dynamic accuracy and repeatability is the ability of a robot to follow a desired trajectory with little or no variance. Speed is how fast a robot can position itself. Carrying capacity refers to how much weight a robot can lift. Memory capacity is the capacity to store the steps of a predefined program in memory by a robot. Manipulator reach is defined as the boundary in which the manipulator of a robot can reach. Considering these criteria listed, some are advantageous while others are not. The load capacity, memory capacity, flexibility in robot program and manipulator reach are advantageous criteria where higher values are desired, while repeatability and cost are non- advantageous in nature, that is, lower values are desired. As market for robots is on the rise, it becomes a difficult task to make a selection decision on the appropriate robot to be deployed for optimal results. A rigorous performance check is essential, in which the effect of various selection criteria is examined. Several approaches including multi criteria decision making (MCDM) approaches and optimization techniques have previously been proposed by the earlier researchers for robot selection. The present paper proposes using grey relational analysis (GRA) for solving MCRSPs accurately and faster. In [4], an analytic network process and mixed integer goal programming (MIGP) model was used to select robot for a computer integrated manufacturing system. The model considered multi criteria, interdependence property and optimization for selecting robots. A unique multiplicative model and algorithmic approach was proposed in [6] and [5], [7] respectively. The use of Choquet integral based decision making method was employed in [13], previously published data set was used to in the study and results were compared to previous approaches. The data envelopment analysis (DEA) approach used in [1] was computationally complex. Electre II method is a time consuming approach used to discard alternatives that are unacceptable and uses a different MCDM approach to make selection. Grey relational analysis was presented in [8] using interval fuzzy numbers, where interval valued indices are used to apply multiplicative operations in place of interval numbers and [3], [11] aimed at developing a fuzzy MCDM model to solve complicated systems with multiple objectives. [9], [10] and [12] used grey relational analysis in solving selection problems using various applications. It is observed that previous researchers used different approaches to solve the robot selection problem. The VIKOR method in [15] made ranking selection of conflicting criteria based on closeness to ideal solution. The GRA methods adopted in [9], [10] and [12] was effectively applied to solve variety of problems, but failed to consider MCRSPs. The GRA is mathematically comprehensible and less rigorous than other methodologies. Thus, this paper presents a faster and efficient solution to MCRSPs using the GRA and examines the effect of using various distinguishing coefficient on GRA results.