Custo AR: Smart Custom Shelving in AR
Key Methods
Product Research, User Research, Competitor Analysis, Design Research, Rapid Prototyping, MVP
My Role
Product Research, Generative Design & AR Prototyping
Team
Everyone (product research), 1 UIUX designer,
1 design prototyper, 1 AR prototyper
Tools
Photoshop, Figma, Unity, C#, XCode, ARKit, Swift
Duration
8 months (capstone), 2018
Client & Stakeholders
Custo CEO & CTO, 2 Supervising Professors
Overview
The Product
CUSTO is an online custom shelf ecommerce website, where users can customize their own shelf designs and purchase them.
The Problem
Currently, users customise their own shelf designs in the desktop browser platform. They face a few problems, namely difficulty to visualise in physical space, manual effort to make measurements, and inconvenience with a desktop browser platform. Additionally, they are limited by the constraints of furniture eCommerce.
The Goal
Client identified AR technology as an opportunity to solve these problems. Our team was responsible in researching and assessing the viability and effectiveness of this approach, and created a working MVP to demonstrate the potential.
Plan
Product Research
An intensive assessment on concept viability and effectiveness is done through a literature review which includes:
User Research, Platform Analysis, Competitor Analysis, Market Research, and AR Feasibility Study.
Identifying Furniture Consumer Needs
Core user needs of furniture consumers are identified and assessed across the different types of furniture retailers.
Assessing Furniture Retailer Constraints
Competitors across both Brick & Mortar and eCommerce were assessed in terms of user needs and constraints.
Both types of retailers are not able to fully address the user need of finding a furniture of good fit.
Assessing eCommerce Potential
Consumers purchasing behaviours increasingly influenced by online presence at an exponential rate.
However, furniture eCommerce usage remains limited due to the inability to try for a good fit before purchase.
Assessing Augmented Reality Potential
AR has been proven to significantly increase consumer engagement.
Furniture has been identified in surveys as the highest product of interest for shopping with AR.
Assessing Augmented Reality Feasibility
Technical feasibility of existing AR platforms were assessed in terms of accuracy & ease of use.
10 competitor AR platforms were assessed on quality of user experience and features.
iOS ARKit was identified as a suitable platform for MVP development.
Scoping the MVP
Through AR visualisation and smart generative designs, product limitations, user needs, and retailer constraints can be tackled:
Real Time AR Measurements - eliminates manual effort required in making measurements
Real Time AR Visualisation - enables testing for good fit in real space in real time
Mobile Flexibility - enables visualising of custom designs anytime anywhere
Assisted Measurements - provide automatic adjustments of measurements using CV
Assisted Designs - provide personalised design suggestions through generative design algorithms
MVP Design & Rapid Prototyping
The MVP work is split into 3 streams - overall UX, algorithm-driven shelf designs, and AR prototyping.
My role was in the algorithm-assisted shelf designs, and AR prototyping.
Algorithm-driven Shelf Designs
First, research on shelf user needs & wants and existing layout designs were outlined to illustrate the potential of a state-of-the-art generative design.
This includes user inputs such as room type (living room, bedroom), user preferred features (seat, table), user preferred items to store, and more.
For the MVP, various shelf subdivision algorithms including slicing, population and Mondrian were explored, rapid prototyped and implemented.
AR Prototyping
AR platform feasibility testings across ARCore, ARKit, Vuforia, Unity were performed, with XCode, ARKit selected for its stability.
AR measurement process was iterated, prototyped, and simplified to 3 steps: measure base, raise height, select furnitures to exclude.
Showcase Exhibition
Akin to a physical showroom, 3 different areas with different space configurations were setup for users to against various real-life scenarios.
3 Different Space Configurations
Each space is designed to portray various real-life scenarios users will use the app in.
Wall with window - AR shelf designs automatically generate below the window.
Wall with television - AR shelf designs automatically generate around the television.
Empty wall - Allow user to test the limits of the designs given an empty setup.
For more details on Custo AR, please reach out to me at mail@alanng.me