Modular Human-in-the-loop Design Framework Based on Human Factors


H. Onan Demirel
Department of Industrial Engineering
Purdue University
December 2015

 Download PDF

Design is an integrative mechanism of scientific methods and artistic approaches, which has been the major driver of human prosperity. Everything is done by, for, or against humans through the inherent activity of design. Thus, considering human needs, abilities and limitations in design process is inevitable. However, this is either neglected or not equally considered when compared to other design contributors. Human-in-the-loop design framework germinates a hybrid design environment towards integrate form (industrial design) and functional (engineering design) requirements of product development - from conception to creation - with human element at the focus.

Human-in-the-loop design framework introduced in this dissertation utilizes Digital Human Modeling (DHM) to incorporate Human Factors Engineering (HFE) design principles early in design process. It embodies scientific methods (e.g., mathematics) and artistic approaches (e.g., visualization) to assess human well-being and overall system performance. This framework focuses not only on ergonomics assessments but also actual design process including, but not limited to, concept development, structural integrity and digital prototyping. It addresses to three major limitations found in HFE literature and practices:

  1. Poor HFE communication between product designers
  2. Poor HFE practice inside product design cycle
  3. Lack of HFE awareness in systems approach

The efficacy of the framework is tested through a design study, where an automobile pillar design with see-through gaps was evaluated for its potential in reducing look-but-failed-to-see obscuration errors. Two human-subject experiments and a simulation experiment were conducted to examine the fidelity of this framework. A blend of statistical (ANOVA/MANOVA), and visual (heat-maps) studies were performed to analyze eye-tracking data. Statistical results obtained from subjects’ feedback (questionnaires, Cooper-Harper test) and structural data (finite element analysis) were combined with eye-tracking data analysis. Results show that human-in-the-loop design framework: 1. demonstrates high test-retest reliability, 2. has potential to overcome HFE design problems associated to conventional design methodologies, and 3. can detect safety and reliability related problems early in design process. Key findings about the design study include: 1. proposed pillar model provides improved visual field to drivers and 2. subjects rated visibility, safety and aesthetics related design attributes of proposed pillar model higher than current pillar models.

Download PDF