A system that implements algorithmic data referencing into existing databases within the food supply chain to help detect, alert, and eventually prevent potential food safety hazards.

2 Weeks
Sarah Strickler
Sara Tieu
Claire Kantner
Ian Yu
Interface Design


This October, with four other students in UW Design, I participated in the 2018 Cooper Hewitt Design Challenge. Although the time was short that we only had two weeks, we still managed to enter the finalist of the competition. The prompt of the challenge is:

How might automation change the mobility of people, goods,
and services?

Define the Scope

After a group discussion, we decided to focus our design direction on the mobility of goods, specifically the transportation of food. We realized that foodborne illness is a major rising concern in the modern global food supply chain. The complexity of modern supply chains inhibits food safety regulatory agencies from locating the origin of contamination, resolving immediate issues, and preventing future foodborne hazards.  

Due to the time constraint, we further narrowed down our scope to focus on only the transportation of frozen food.

Photo of group discussion

Design Methods

Secondary Research

To start, we did the secondary research on topics below:    

1. Factors of food contamination
2. Foodborne illness and case studies
3. Food supply chain
4. Current food regulation agencies
5. Existing technologies
6. Blockchain

Screenshots of parts of the research documents


Through the research, we found that:    

1. In the U.S, it is estimated that approximately 12% of food waste occurs during distribution, mainly because of the inappropriate refrigeration.

2. 59% of selected food facilities did not comply with FDA's recordkeeping requirements.​

3. ​56% of food facilities have gone five or more years without an FDA inspection.​​

4. It is estimated that the meals in the United States travel about 1,500 miles to get from farm to plate.

5. It is very difficult to identify the specific ingredient that is contaminated at which stage of the supply chain after the outbreak of foodborne illness because the food could go bad in any section. ​​​

We discussed the research findings and insights


1. Respondents in the supply chain should be immediately notified of potential food hazards to prevent safety violations and mitigate wastes.

2. Temperature and location tracking of food needs to be accurate, constant, and real-time.

3. ​The FDA and USDA need patterns indicating food safety violation and hazards to schedule targeted inspections.

4. Data should be transparent and constantly shared between supply chain sectors and regulatory agencies.


To further understand how the food supply chain and the industry actually work, I interviewed Max, the sous chef of UW Housing & Food Services. He told me that dealing with frozen food,

"The only thing we need to know is if the frozen food has been frozen the whole time."

He showed me the sensor that was already implanted on the frozen container; if the temperature rises, the blue chemical inside the sensor will melt and create a blue line showing the food has gone above the freezing point any amount of time. Then he would just throw the food away.

Photos of the food loading dock & Max


1. Prevention is the main goal.

2. The current sensor only works as an indicator, showing the food has not been frozen for a while so that the next person in the supply chain can notice the problem and throw away the product.
3. For our design, we want to go a step forward by alerting the respondents in the supply chain when there are potential food hazards detected, to prevent the food from being out of the safe temperature zone in the first place.

Design Process

From the research and insights, we identified the problem and came up with a "How Might We statement":

How might we create a system that ensures food safety requirements are met during transportation points in the
supply chain?

Our response is:

A system that implements algorithmic data referencing into existing databases within the food supply chain, to generate predictive alerts and suggestions in order to detect and prevent potential food safety hazards.

We acknowledged that there were many sectors along the food supply chain. Due to our time constraint, we decided to focus on monitoring food based on temperature and location to prevent food safety hazards, and mitigate waste in the distributor and retailer sectors

System Model

We first built a system model which showed how the system would work with the human factor.

Initial sketch of the system model
Refined system model

In the model, we identified two types of users: the Respondents and the Regulatory Agencies.  
Respondents: each sector of the food supply chain (the supplier, processor, distributor, retailer, and customer)    

Regulatory Agencies: food regulatory agencies, mainly the FDA and USDA who are responsible for conducting the inspection.  

System Model

Refined system model

The conceptual model shows that the system generates predictive alerts and suggestions through a paired interface to help supply chain sectors recognize food hazards, and take necessary steps to prevent consumers from receiving compromised food.  

Regulatory agencies will be alerted of food hazard patterns and given access to data that will help them curate future inspections. Our solution assumes in the future, data within the supply chain will be comprehensive and real-time (i.e. blockchain implementation)

Discussing and revising system models


The storyboard shows one of the scenarios when our system successfully detects and alerts USDA and the next sector in the supply chain the abnormality of temperature during frozen food transportation. Eventually, the system prevents food waste.

Sketch of the storyboard

Final Design

Our design requires less human intervention along the supply chain making a larger impact.

State View - Condition Stable

When all conditions are stable. The map is zoomed out so that the respondent can see an overall status of more trucks.

City View - Condition Stable

The respondent can zoom in to see more accurate
real-time locations of the truck.

City View - Abnormality detected

When there is abnormality detected, the system will send an alert to the respondent. The left graph shows when one abnormality detected. The right graph shows when multiple abnormalities detected.

When multiple alerts are happening at a time, the system will prioritize the more emergent incident. However, we should still give users the autonomy to choose which incident to solve first.

Detailed - Abnormality detected

The respondent can click to view more information about the abnormality and see the suggested action provided by the system. The respondent still has the autonomy and control of the reaction.

Alert Resolved

The system back to the stable state. The respondent can view the status of the incident, checking if every problem has been resolved or not.


Lack of Knowledge in the Related Field
We noticed that while designing the interface, sometimes we designed based on assumptions. For example, we assumed that these were the tools and functions the respondents need. We realized that we made assumptions because most of the design decisions were based on the secondary research. We did not have a comprehensive understanding of the industry.

Future improvement
Our design will be stronger if we can actually conduct interviews with people working in the USDA and the food supply chain so that we can have a better understanding of their needs.

Interface for Food Regulatory Agencies
Due to the time constraint, we did not design the interface for the food regulatory agencies.

Future improvement
I would like to continue the project by designing the interface
for regulatory agencies. However, before that, I will first do more field research.

Data Collection
In our model, we collect data such as the temperature and humidity of each box through sensors attaching on the container.

Future improvement
We need to consider if the design is feasible, desirable, and viable; specifically, whether the producer is willing to spend the money on the sensor.

Other Projects

Back to the Top