Fresh.ai
Designed a unique IoT system to combat the issue of household food waste by better understanding consumer behaviour and presenting strategies to minimize food waste.
For their fourth year thesis project, students were challenged to create innovative solutions to problems they are passionate about.
My journey into conducting a project on household food waste was sparked by my previous professional experience in the fine-dining industry as French trained chef. I often found myself forgetting the contents of my crowded refrigerator and losing track of expiration dates, which would lead to unnecessary waste. This prompted me to explore if similar issues were common in other consumer households.
Skills: User Research, Product Design, Machine Learning, AI
Tools: Figma, Adobe CC, Arduino, Pixy2 Libraries
Duration: Winter 2024 Semester (8 months)
Award: Best Technical Experimentation for Major Project (Year End Show 2024)
RESEARCH GOALS
Key Insights
A common issue is the invisibility of items, which often leads to them being overlooked and eventually wasted.
Participants indicated that being able to see their food items serves as an effective reminder to use them.
There is a significant lack of understanding about food labels and expiration dates, contributing to misinformation and waste.
The findings highlight a need for better organizational tools in households to reduce food waste and suggest a potential area for technological innovation to address these challenges.
IDEATION
Initial Ideation and Challenges
During the initial ideation phase, I explored the idea of redesigning a refrigerator to simply improve user access and visibility, aiming to reduce forgetfulness and food waste.
However, I recognized that such a futuristic concept might be impractical for a typical household due to space constraints and my lack of refrigerator design knowledge.
This insight highlighted the importance of designing without a solution in mind and led me to gather more user data design for user needs.
ADDITIONAL RESEARCH
Data Collection and User Cognition
To better understand user interaction with food-related information, I conducted a card-sorting exercise with six classmates, structured into two phases:
Phase 1: Participants sorted cards without any context to gauge instinctual categorization.
Phase 2: Participants sorted cards labeled with descriptors such as technology, colors, and emotions related to their food-related feelings.
This exercise was crucial in developing an understanding of how users categorize and process information and in designing an effective information architecture for managing food items.
SOLUTION
IoT Concept
I designed an aftermarket IoT device equipped with sensors and an AI-enabled smart camera to convert standard refrigerators into "smart fridges."
This device would be paired with a user-friendly app that would automatically track food spoilage and intuitively manage inventory, which would offer a practical solution for food management.
Additionally, this solution would be a cost-effective and convenient alternative to purchasing a new smart refrigerator, simplifying user interaction by autonomously scanning and recording new groceries.
Features, Screens & UI
Simplicity in Design: Users indicated a preference for a straightforward UI that effectively manages inventory without unnecessary complexity.
Customizable Display: Users can control how much information is displayed, tailoring the interface to their specific needs.
Visual Appeal and Functionality: The UI is designed to be visually attractive while focusing primarily on inventory management.
Color-Coded Alerts: The app uses color coding to convey important spoilage-related information quickly, using minimal text or numbers to avoid information overload.
BEHIND THE SCENES / PHYSICAL PROTOTYPE
Color Recognition Training
Training and teaching the camera how to understand and recognize the color yellow, allowing it to recognize a banana
To enhance the accuracy of color recognition, the surrounding colors in the camera’s field of view are darkened to help isolate the banana as a target.
Demonstration of Functionality
In this behind the scenes shot, the camera can be seen effectively identifying the color red, thus showcasing its ability to distinguish an apple inside the demo fridge.
This prototype was built using Arduino, Pixy2 Vision Sensor, and ESP32.
Pardon the messy desk! 😅
CONCLUSION
The final representation below showcases the camera system in action within the demo fridge. As fruits are introduced, it scans, recognizes, and tracks their quantity, displaying the results in real time.
This project earned the “Best Technical Experimentation for Major Project” award at the George Brown College's Year End Show. It was very well-received by peers and the community.