Sensing · policy and alerts
Alerts that inform without shaming
A sensor can flag vapor. The school still decides what happens next, and that is the part I designed for. Notifications, logs, and keeping an alert from turning into a public scene.
Power budgets, false positives from cleaning sprays, and simple service access shaped the story as much as the circuit did.
I split the doc into sensing science, system behavior, and human outcomes, so a school can change its policy without anyone rewriting firmware.
- Role
- Industrial design intent, UX for alerts
- Domain
- Environmental sensing
- Type
- Research and campus initiative
- Stage
- Proof of concept
If you open the technical PDF, come back here for the ethics and the workflow. That is where the real design lives.
Sensing fundamentals
Signal versus nuisance
The first thing I separated was what the sensor measures from what people assume it measures. That gap is exactly where false confidence grows.
I modeled HVAC, humidity, and cleaning sprays as interference from the start, not as rare quirks to patch later.
1. Baseline capture
Record ambient traces across a full week to learn what normal variance actually looks like.
2. Labelled events
Run controlled tests with facilities, so a positive reading means something you can stand behind.
3. Threshold policy
Pair the engineering thresholds with a human review before anything gets sent automatically.
Device and placement
Visibility, service access, airflow
Placement had to account for ceiling height, dead air, and how you actually reach the unit. A box you cannot service quietly turns into a false-negative machine.
I kept the enclosure neutral. No punitive icons. It should read as infrastructure, not as theater.
Alerting workflows
From ping to documented follow-up
I staged notifications from a local staff cue up to admin dashboards and an audit trail with retention rules.
I kept the copy out of accusatory territory. The system reports an environmental event. People decide the response.
Privacy and governance
Collecting the minimum useful data
Retention and export rules followed data-minimization thinking. If policy cannot explain why the data stays, the feature should not ship.
I brought facilities, student affairs, and legal into the workshops when they were available. Alignment early beats an ethics retrofit later.
Field learnings
What I would change next revision
Firmware should expose sensitivity per room profile. No two buildings behave the same way.
Maintenance belongs in the rollout plan, not tucked into an appendix nobody reads.
What did I learn?
- Hardware stories need honest uncertainty, not clean guarantees.
- The alert copy is part of the sensor.
- Governance workshops are design workshops.
Stakeholder notes
We cared less about flashy dashboards and more about knowing the alert was real before talking to a student.
- Lead
- Laolu James