Evolving User Needs Identification through AI Augmented Approaches
Name
schelhaas-booker-sm-meche-2024-thesis.pdf
Description
Thesis PDF
Size
1.62 MB
Format
Adobe PDF
Checksum (MD5)
9b0cc3d2de7ca794ff7c00c66ef9eb7d
Author(s)
Schelhaas, Booker B.
Advisor(s)
Yang, Maria C.
Date Issued
May 2024
Publisher
Massachusetts Institute of Technology
Abstract
In a human-centered design approach to the product design cycle, conducting a user needs analysis is critical to the long-term success of the project. Designers routinely are tasked with engaging in stakeholder studies in an effort to identify their needs that then drive the design. Sometimes users are aware of their needs, but often they are not conscious of some important yet hidden needs, called latent needs, which are particularly difficult to identify. The identification process can be laborious and resource intensive, including interviews and in-depth observations by experts to extract workarounds and pain points that suggest the highest potential for product success. This thesis aims to first explore the current status quo for user need extraction through observation and interviews, and then presents a preliminary and novel AI based method for augmenting designers’ abilities to aid in the process. The first chapter presented demonstrates what can be done with traditional methods. We conducted videos and observations of older adults to understand their ability to stand and opinions on devices to aid them. After conducting many interviews and observations, we identified that the use of stand assist devices is in itself a latent need, as there exists a perception gap in the older adults between their perceived ability to stand and their actual ability to stand, as diagnosed by a trained physical therapist. The following chapter in response to some of the difficulties observed in the first study presents a novel AI tool to augment designers’ abilities to identify user needs from observational videos. Our tool utilizes pose estimation to calculate ergonomic risk of users as they engage in a task, as well as object segmentation to identify objects that could be affecting the user’s behavior. 2 These are then compiled into a computer interface for designers to use when watching an observational video of a user. Methods, experimental design, and future work are discussed for the study which is pending to be completed.
MIT Department
Massachusetts Institute of Technology. Department of Mechanical Engineering
Terms of Use
Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
Copyright retained by author(s)
Persistent DSpace Link
