SENSORY

The concept began with a simple distinction: occupancy data does not have to be surveillance data. Thermal sensing can describe where warmth is changing without creating a recognizable portrait of the person creating it.

Research from Aalto University showed that people consistently associate different emotions with different patterns of bodily sensation. I treated that work as an inspiration, not a diagnostic claim. It opened a larger design question: what could a home understand from heat, and where should it deliberately stop?

That boundary shaped the architecture. Thermal input could prompt a check-in, surface an appliance anomaly, or reveal a fall-like change in posture. It could not label a resident's emotion as fact or replace a medical professional.

The hard part was not inventing more features. It was proving that each capability belonged to the same system. I mapped the home as a continuous thermal environment, then organized actions around four human needs: understand, assist, protect, and respond.

The platform covered room occupancy, appliance heat, safety alerts, health-oriented check-ins, accessibility assistance, security anomalies, and environmental controls. Each feature used the same floor-plan model so users did not have to relearn the house every time they changed tasks.

I explored a folding display, a compact round control, a solar-powered panel, and a wall-mounted screen paired with detachable sensors. The final direction separated sensing from control: small modules could be placed where coverage mattered, while the larger display stayed at a reachable, predictable location.

A folding hinge added failure points. A small round screen hid too much information. Solar charging made a safety system weather-dependent. The wall-mounted panel offered the clearest tradeoff between reach, legibility, repairability, and room-scale context.

The UI uses the floor plan as its anchor. Rooms, people, appliances, and alerts all resolve to a location instead of becoming separate app destinations. That reduces navigation and makes unusual changes easier to interpret.

Accessibility is a primary mode, not a settings appendix. Large touch targets, direct emergency access, navigation assistance, and seated-use reach were considered in the first layout. Thermal views stay visible as evidence, while plain-language labels explain what the system believes is happening.

The original concept separated health, safety, accessibility, and environment into different feature lines. Today I would use a shared contextual model, but keep the dashboard as a transparency layer: what was sensed, what changed, what is uncertain, and what action is proposed.

I would also move more inference on-device, reduce retained data, and give every resident a visible privacy state. Better models make the concept more feasible, but they do not change the design principle: the most vulnerable resident is the primary user, not an edge case.