Ambient Intelligence: Context-Aware Device Handoff

Occupant & Sensing

Walking speed 1.2 m/s

Sensor noise σ 0.05

Activation threshold 0.55

Dwell-time weight 0.40

Room State

Active device

— none —

Confidence

Handoffs

How it works

Ambient computing means the environment itself infers what the occupant needs and quietly activates the right device — no explicit command, no screen to tap. This simulator models that context inference as multi-signal sensor fusion, computed independently for every device in the room:

proximity  = clamp(1 - dist/maxDist, 0, 1) + noise(σ)
dwell      += (dt/τ) · (proximity − dwell)        // exponential low-pass filter
approach   = clamp( v̂ · d̂ , 0, 1)                 // is the occupant walking toward it?

raw   = 0.5·proximity + w_dwell·dwell + 0.3·approach
score = 1 / (1 + e^(−5·(raw − 0.5)))               // logistic squashing → 0..1

The device with the highest score becomes active only once its score clears the activation threshold — and to switch away from the currently active device, a challenger must beat it by a safety margin. That hysteresis is what keeps a noisy room from flickering between two devices every frame; raise sensor noise to see the raw proximity readings jitter while the smoothed dwell signal keeps the handoff decision stable.

Proximity — instantaneous distance-based signal, the noisiest input.
Dwell — a running average of proximity (time constant τ ≈ 1.2 s), modelling how ambient systems build confidence the longer you linger near something.
Approach — the dot product of walking direction and the direction to the device, so a device you are walking toward is favoured even before you arrive.

Faint tiles on the floor accumulate and slowly decay with occupancy — a visual stand-in for the kind of passive presence history (PIR sensors, ultra-wideband, Wi-Fi CSI) real ambient-computing platforms use to build spatial context over time. Click anywhere on the floor to steer the occupant there directly.

Click the floor to walk there

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.