To achieve this goal, we explored two potential solutions: machine learning (ML) and computer vision (CV). After evaluating implementation complexity, scalability, and cost-effectiveness, we opted to use CV as the foundation to drive future ML advancements.
Technical Roadmap Decision: We conducted sessions testing different types of potential sensors. Considering our team’s expertise, development timeline, and budget constraints, we decided on a solution that combines a Sharp sensor with a motorized camera.
Then, we had a series of brainstorming and prototyping sessions to validate our roadmap with a funtionable MVP prototype.
Information Architecture & Lo-fi Interface:


Key Usability User Tasks:
Task 1: Add a new device to the map.
Task 2: View real-time pest distribution and humidity data.
Task 3: Check alerts when multiple pea weevils are detected and identify the alerting device.
Task 4: Respond to alerts based on the situation: If it's a false alert, adjust the pest count and mark it as resolved. If it's a real infestation, order chemicals and mark it as resolved.
Task 5: Check a device’s operational status.
Usability Testing - Results
We conducted testing sessions with Andrew's team. Here're the results
Metrics |
Task 1 |
Task 2 |
Task 3 |
Task 4 |
Task 5 |
Task Completion Rate (%) |
92.6 |
73.6 |
100 |
87.6 |
100 |
Avg. Time on Task (sec) |
24.87 |
12.70 |
3.39 |
112.37 |
12.50 |
Error Rate (%) |
8.16 |
7.76 |
0.66 |
12.70 |
1.45 |
Cognitive Load Score (1-10) |
3.92 |
3.52 |
1.62 |
4.16 |
1.62 |
Usability Testing - Insights
✅ What Works Well:
Checking alerts (Task 3) & device status (Task 5) are smooth – high completion, low errors, minimal cognitive load.
Users quickly identify alerts (3.39 sec) – UI is clear and efficient.
❌ What Needs Improvement:
Viewing real-time data (Task 2) confuses users – lower completion (73.6%), likely due to unclear visualization.
Responding to alerts (Task 4) is slow (112 sec) & error-prone (12.7%) – complex decisions, unclear workflow.
🎯 Action Plan:
Simplify data display for Task 2 – clearer graphs, better layout on homepage.
Improve Task 4 workflow – reduce decision steps, add confirmations, highlight urgency levels.
Usability Testing - Homepage sprint, testing and decision
For the user experience (UX) design, based on the target user demographics—such as age and skill level—as well as team size and project timeline, we developed two prototypes: Plan A and Plan B. We conducted independent usability tests with the Microsoft team members, and the results indicated that Plan B had higher user satisfaction, achieving an 90% approval rate.
Design Decision: Field research showed farmers mainly need pest alerts and action steps, so we moved detailed data to a separate page, leaving only key info and calls-to-action on the Dashboard.
Pros:
1. Reduces cognitive load
2. Streamlines workflow (no need to sifting through data)
3. Improves usability for all users
Cons:
1. An extra click to get in-depth data