Explore how an agent interprets a space, attaches comments to a place, and reuses a landmark. A short visual story shows the idea, followed by the prototype recording.
SMAS / SPATIAL SIMULATION & DESIGN FEEDBACK
SMAS / Design through an agent’s eyes
It finds a way. It leaves an insight.
A simulated visitor explores a gallery. What it notices becomes feedback on the design, tied to the place where it happened.
Illustrated proposal · authored example
01
The agent encounters your design.
At a junction, a column hides the gallery sign. The opening ahead looks like the way through.
The agent’s view / First visit
02
It leaves a comment where it happened.
The interpretation stays attached to the junction and the viewpoint. The architect can see which part of the design prompted it.
Comment at this junction
“By the amber window, I couldn’t see the sign. I tried the opening ahead.”
03
The place becomes a landmark.
After finding the gallery, the agent links the distinctive window to a learned direction: turn right here.
Remembered: amber window junctionA place worth rememberingAmber window → turn right for Gallery 03Linked to the original comment and viewpoint.
04
Next time, it knows the turn.
The agent recognises the landmark and uses it to navigate. Its earlier comment still shows the difficulty it encountered.
Recognised on the next visit
“The amber window. This is where I turn right.”
What this gives the architect
Feedback you can place on the plan. A viewpoint, an interpretation and the action that followed, connected to the same location.
Inside the prototype / Product video
See the space. Understand the task.
A short recording from the SMAS prototype. Watch its spatial mapping view, then explore a simplified wayfinding experiment below.
SMAS / Prototype recording00:22 · Spatial mapping
Prototype footage. The interactive station below is a separate, synthetic website demonstration.
Station 03 / A synthetic test environment
Destination / Platform 030 / 120 arrived
Full-run results / computed for the selected scenario
Visitors taking a wrong turn-
Average journey-sec
Average junction queue-sec
Illustration, not a findingThis station and its visitors are invented. Results are calculated from the assumptions below. They are not calibrated journey times or evidence that a signage intervention works.
The difference behind the average
Who has the longer journey?
Clear signs can reduce a wrong turn. A wider junction can reduce a queue. A useful study separates the causes, then checks whether the model captures what happens on site.
Average journey by visitor group
Open the model notebook Assumptions & limits
What is calculated
Each visitor gets a fixed arrival time, walking speed and route-choice draw. The same draws are reused across alternatives.
Baseline wrong-turn probabilities: first visit 46%, regular visitor 8%, luggage 32%. Clear signage multiplies these assumed probabilities by 0.3.
Base speeds: 1.12, 1.36 and 0.91 metres per second, with individual variation.
Junction service interval: 1.1 seconds, or 0.42 seconds when widened. Waiting follows arrival order.
Journey time is route distance / speed, plus the calculated queue and an assumed hesitation.
What this does not establish
These probabilities and capacities are authored for the demonstration. People follow a small route graph; they do not read the sign with a vision model. The model does not simulate collisions, safety, evacuation or accessibility compliance.
For a real pilot, the first test is whether the model ranks a baseline and intervention in the same direction as independent observation or expert assessment. A negative result stays a negative result.
NextA real case, together
What’s about to change?
A building. A service. A policy. Tell us what is changing, who it affects, and which decision is still open.