Co-design Workshops

A series of workshops

Co-design Workshops

Brief

A series of workshops co-designed with specific users, quickly producing prototypes by 3D printing, deep learning model, UI design and other methods

Duration

Sept 2023 - Apr 2024

Collaboration

--

Find Report ↗

My Role and Core Product

User Experience Researcher & Designer

Methodology and Process

  • 3D Printing, Arduino - prototyping
  • Semi-structured interview
  • Python - coding
  • Figma, Adobe Pr - UI/video delivery

Session 1 - 3

Session 4

This project presents a assistive technology prototype designed to enhance the navigation experience for Visually Impaired(VI) individuals. Integrated with computer vision and ultrasonic sensor, the system is able to detect and recognize a variety of objects in both indoor and outdoor environments.

Framework

the overall architecture of this system

Customised Computer Vision Model

Most existing convolutional neural network (CNN) models for road surface semantic segmentation are tailored for the perspective of cars, unsuitable for visually impaired individuals’ walking view.

To address this, a custom dataset was created specifically for visually impaired wayfinding, focusing on roadblocks, pedestrians, and other elements. The dataset was used to train a customised model with YOLOv8.

create custom dataset fot visually impaired people and train the dataset via YOLOv8

Integrated with ultrasonic sensor

At the hardware level, the system includes a micro-controller with an embedded camera (ESP32 with OV2640), an ultrasonic sensor (HC-SR04), Bluetooth bone conduction headphones, and an external battery.

The prototype is low-cost and easy to wear, avoiding the need for potentially expensive or specialised components.

show the work flow of the system and arduino prototype

UI and User Test

Users can customise their experience based on individual preferences.

To enhance usability for the visually impaired, the interface adopts strategies such as increased font size, prominent icons, and the use of pictures and buttons. All text components within the app are accompanied by voice prompts, ensuring that the content remains accessible.

simplified UI and results of user tests