Using eye-tracking into decision makers evaluation in evolutionary interactive UA-FLP algorithms
- PDF / 1,297,853 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 11 Downloads / 184 Views
(0123456789().,-volV)(0123456789(). ,- volV)
ORIGINAL ARTICLE
Using eye-tracking into decision makers evaluation in evolutionary interactive UA-FLP algorithms Lorenzo Salas-Morera1
•
Laura Garcı´a-Herna´ndez1 • Adoracio´n Antolı´-Cabrera1 • Carlos Carmona-Mun˜oz1
Received: 2 October 2019 / Accepted: 7 February 2020 Ó Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Unequal area facility layout problem is an important issue in the design of industrial plants, as well as other fields such as hospitals or schools, among others. While participating in an interactive designing process, the human user is required to evaluate a high number of proposed solutions, which produces them fatigue both mental and physical. In this paper, the use of eye-tracking to estimate user’s evaluations from gaze behavior is investigated. The results show that, after a process of training and data taking, it is possible to obtain a good enough estimation of the user’s evaluations which is independent of the problem and of the users as well. These promising results advice to use eye-tracking as a substitute for the mouse during users’ evaluations. Keywords User ergonomics User fatigue Engineering design UA-FLP Artificial neural networks
1 Introduction The plant layout design is a critical issue in industrial manufacturing as well as other fields, like schools, hospitals, and offices, among others. It deals with the arrangement of spaces, machinery, and facilities in order to satisfy certain objectives, as minimizing material handling cost; facilitating supervision and control; integrating man, machines, and support services; workers’ safety, adaptability to changing conditions, or waste minimization [43, 48]. The general problem of facility layout planning (FLP) and its variant unequal area facility layout problem (UAFLP) have been addressed by means of several methods. Initially, the problem was addressed by exact methods, which are useful for finding solutions to problems of reduced size since the problem has been classified as NPHard [41]. For example, quadratic assignment problem (QAP) [22, 26, 31], branch and bound [6], integer programming [37], and mixed integer programming [50]. & Lorenzo Salas-Morera [email protected] 1
Universidad de Co´rdoba, Campus Universitario de ´ rea de Proyectos de Ingenierı´a, 14071 Co´rdoba, Rabanales, A Spain
Due to the high computational cost of finding optimal solutions when problems are complex, heuristic methods that find good enough sub-optimal solutions in a reasonable time have been proposed. Firstly, two kinds of algorithms appeared: improvement algorithms and construction algorithms. The first ones start with a solution and try to improve it by interchanging facilities locations, as for example CRAFT [4], but they have the problem that the solutions depend critically on the initial one [7]. The second group obtains the solutions locating successively the facilities in the remaining space until completing a design. Solutions obtained by these algorithms may be far from
Data Loading...
