Machines see
everything.
They notice nothing.

One beam.
Many questions.

A rat in the dark does not build a map of the room and then walk through it. It reaches, touches, and moves — and the reaching is the sensing. Whiskers probe, detect edges, find openings, estimate proximity, and reveal geometry that vision cannot reach.

Wiskl applies that principle with light. A steered beam is not a camera pointed harder; it is a question asked in a specific direction, and an answer that comes back in microseconds with a range and a surface attached to it.

Guidance creates
the question.

Scanning because a sensor can scan is not perception, it is throughput. The useful order runs the other way: movement needs information, the need selects a direction, and only then does anything point anywhere.

1. ✱Guidance
2. →
3. ◇Attention
4. →
5. ▣Photonics
6. →
7. ◉Measurement
8. →
9. ◈World model

Everything above the beam speaks one language. The steering layer never learns where a request came from — an operator today, an uncertainty model later — so the intelligence can be replaced without touching the optics.

Wiskl should understand
the geometry of its own blindness.

Not-yet-measured and cannot-be-measured-from-here are different problems. One is answered by looking. The other is answered by moving, and knowing which is which is what turns a sensor into a sense.

// The seam the whole platform is built around.
// Today an operator types this. Later an
// uncertainty model emits it. The photonics
// layer cannot tell the difference.

AttentionTarget {
    azimuth   : +18.4°
    elevation : -2.1°
    priority  : 0.92
    dwell     : 100 µs
    beamWidth : 1.6 mm
    reason    : "occluded region on
                 the planned path"
}

// → inverse solve  θ₁ = 116.4°  θ₂ = 242.8°
// → move, measure, correct
// → residual 8.5 µm  (8.5 µrad @ 1 m)

▲Priority is a placeholder scalar today. The intended form is uncertainty × guidance relevance, with later terms for occlusion, hazard, motion, memory and measurement cost — but the optics has to work first.

The bench is a bench.
The organ is the goal.

Lumiskur is the photonic sensing organ inside Wiskl — the subsystem that will eventually own illumination, beam steering, optical measurement and photonic receive as one integrated part rather than a table of components.

• 01 Illumination — generate and shape the probe
• 02 Beam steering — aim it without moving mass
• 03 Optical measurement — recover what came back
• 04 Photonic receive — on the same substrate

Electronic control of light,
not mechanical motion of optics.

Rotating prisms are how you learn the optics, not how you ship it. Every macroscopic moving part is mass, latency, wear and a thermal path. The progression runs from motorised prisms through a fast steering surrogate to integrated photonics, an optical phased array, and eventually a metasurface front end — and the software written for the first stage is meant to survive to the last.

▲Lumiskur is not required for POC 0 and does not exist yet. The optical bench is the beginning of that organ, and nothing here should be read as a component you can order.