# RETURN — research notes and formative test plan

Private working document. Sources checked 19 September 2026. This is a design exploration with simulated telemetry, not a flight-qualified interface. No participant sessions, suited trials, or physical readability tests have been conducted for this document.

**Purpose:** help the wearer judge when to return, identify the most urgent action, and make an understandable handover to a crewmate.

## Evidence ledger

- **SOURCE:** a finding supported by a linked primary source.
- **CALCULATION:** derived from the specified screen geometry and viewing distance.
- **DESIGN HYPOTHESIS:** a proposed design choice that requires evaluation.
- **PLANNED TEST:** an activity not yet performed. A planned test is never a result.
- **OBSERVATION:** reserved for dated, attributable participant evidence; none recorded yet.

## Research that changes the design

| Source and finding | Application and limit |
| --- | --- |
| **SOURCE — NASA, Appendix F, F.5.1:** text uses a minimum character visual angle of 0.25°, with ≥0.4° preferred; the measure refers to uppercase glyph height. Sans serif text and stable-width numerical data support discrimination. [Display Standard](https://www.nasa.gov/reference/appendix-f-vol-2/) | **DESIGN HYPOTHESIS:** start important labels near or above 0.4° and enlarge urgent content. Measure rendered glyphs, not just the font-size setting. This Latin-uppercase reference does not validate Chinese readability. The Chinese interface deliberately departs from F.5.1's ASCII suggestion and needs language-specific review. |
| **SOURCE — NASA, V2 10020, 10113:** distinguish stale/missing/unavailable/unknown data and present critical information without repeated navigation or mental calculation. [Crew Interfaces, §§10.2.1, 10.2.6](https://www.nasa.gov/reference/10-0-crew-interfaces-vol-2/) | **DESIGN HYPOTHESIS:** show return margin with its basis. Remove an invalid estimate and name the missing input; label retained history as history. A numerical stale threshold must be set by the mission's sensor and task requirements, not this prototype. |
| **SOURCE — NASA, V2 10114/10175:** use distinct visual and audio alerts and prioritize concurrent alerts. Appendix F assigns yellow to caution and red to warning/emergency, and treats event activity, acknowledgment, and annunciation as separate states. [Crew Interfaces, §10.3](https://www.nasa.gov/reference/10-0-crew-interfaces-vol-2/); [Display Standard, F.6.4](https://www.nasa.gov/reference/appendix-f-vol-2/) | **DESIGN HYPOTHESIS:** use a labeled yellow caution state for emerging concerns; reserve the dominant red critical state for the simulated immediate-danger scenario. “Critical” is product language, not a claim of equivalence to NASA's emergency/warning classification. Acknowledgment must not imply recovery. Add helmet-audio behavior only after system integration is defined. |
| **SOURCE — NASA, V2 10063/10068:** labels belong next to the items they identify; controls used without sight need spatial or tactile distinction. [Crew Interfaces, §§10.4.3.2, 10.4.4.1](https://www.nasa.gov/reference/10-0-crew-interfaces-vol-2/) | **DESIGN HYPOTHESIS:** keep each label beside its physical key. A rescuer view rotates the reading content only if the correct key association is maintained. Identical bumps are not evidence that the keys are reliably distinguishable in pressure gloves. |
| **SOURCE — NASA, V2 11024:** suits can reduce dexterity, tactility and mobility; pressurization can reduce capability further. V2 11035/11036 address displayed and monitored suit pressure, oxygen partial pressure and carbon-dioxide partial pressure. [Spacesuits, §§11.2.1, 11.3.3–4](https://www.nasa.gov/reference/11-0-spacesuits-vol-2/) | **DESIGN HYPOTHESIS:** use four discrete inputs with visible feedback, without touch gestures or hold-to-repeat critical actions. Keep raw environmental data reachable behind the return estimate. A thick terrestrial glove can reveal gross problems but cannot validate a pressure-glove interface. |
| **SOURCE — NASA/JPL, 26 March 2025:** fine Martian dust adheres to suit materials; Mars exposure also involves UV and cold, and material degradation remains under study. [Perseverance suit-material research](https://www.nasa.gov/missions/mars-2020-perseverance/perseverance-rover/how-nasas-perseverance-is-helping-prepare-astronauts-for-mars/) | **DESIGN HYPOTHESIS:** test reduced contrast, partial occlusion and off-axis reading; do not promise that a palette survives actual visor, dust and lighting conditions. A screen overlay is only a legibility stressor, not a physical dust simulation. |
| **SOURCE — ESA, 27 October 2020:** Apollo lunar dust obscured visors, clogged mechanisms and damaged suit layers. [Dust-proof material research](https://www.esa.int/Enabling_Support/Space_Engineering_Technology/ESA_seeking_dust-proof_materials_for_lunar_return) | This is **lunar analogy**, not Mars-specific evidence. It motivates checking both reading and actuation after contamination. Do not transfer lunar particle sharpness or abrasion rates to Mars. |

These references inform a concept; selective use does not establish compliance with the entire NASA standard. Suit-specific hazard limits, alert timing, communication protocols and emergency procedures remain outside this design's evidence base.

## Actual-size calculation

**CALCULATION.** Assume a rectangular active area, square pixels, no display cutout, 6.7-inch active diagonal and a 2400:1080 aspect ratio. Bezel dimensions are unknown.

`diagonal = 6.7 × 25.4 = 170.18 mm`

`long edge = 170.18 × 2400 / √(2400² + 1080²) = 155.191 mm`

`short edge = 69.836 mm; pixel density = 15.4648 px/mm ≈ 392.807 ppi`

For viewing distance `d = 450 mm`, glyph height is `h = 2d × tan(angle/2)`:

| Visual angle | Actual glyph height | Native pixels |
| --- | ---: | ---: |
| 0.25° | 1.963 mm | 30.37 px |
| 0.40° | 3.142 mm | 48.58 px |
| 0.50° | 3.927 mm | 60.73 px |
| 0.60° | 4.712 mm | 72.88 px |

The 48.58-pixel value is **glyph height**, not CSS or Figma font size. If a measured font has a 0.70 cap-height/em ratio, a 0.4° capital needs about `48.58 / 0.70 = 69.41` native pixels of font size. This ratio is illustrative; measure the chosen font. The prototype uses a 1200 × 540 logical drawing mapped 2× to 2400 × 1080; divide native sizes by two. Browser devicePixelRatio does not establish physical size.

Print or calibrate the active rectangle to **155.2 × 69.8 mm**, check a ruler reference, and measure from eye to screen. A large desktop presentation is not an actual-size check. Test Chinese independently with fluent readers; do not compress translated labels to fit English boxes. Physical contrast through the intended visor and display stack remains unmeasured.

## Planned 18-minute formative session

Recruit 5–8 participants in iterative rounds if available, including fluent Chinese readers and a range of hand sizes; report the actual sample honestly. Use authorized private participants only. Lay readers can screen interpretation problems; they cannot substitute for EVA crew and human-factors specialists. Obtain agreement before any recording.

Prepare an actual-size print or calibrated display, four physical substitute keys or a four-key keyboard, ruler, timer, and a private note sheet. Place it on the left forearm, with the right hand operating keys. Record exact substitutions. No loose powder is needed: use a removable visual-occlusion sheet for an optional reading check.

| Time | Moderator script and task | Record |
| --- | --- | --- |
| 0–2 min | “We are testing the interface, not you. All values and events are simulated. You may stop at any time.” Explain four physical inputs; do not teach the solution. | Language, vision correction, setup, display dimensions, actual distance and relevant experience. |
| 2–4 min | Show home for five seconds, then cover it. “What is the main time telling you? When would you start returning? Was anything wrong?” | Exact interpretation; confusion between work margin, return duration and survival time. Do not correct until answers are recorded. |
| 4–7 min | “Use the buttons to find the rover's bearing, distance and estimated return time.” No screen tapping. | First chosen key, total presses, wrong presses, assistance and completion time. Ask whether the arrow proves a safe route. |
| 7–10 min | Introduce a caution, then a critical event. Ask “What changed? What is the next available action?” Repeat once with a key held during escalation. | Whether severity and identity are noticed; whether old input triggers a new action; whether fresh release-and-press is understood. This is input testing, not an emergency drill. |
| 10–13 min | A second reader approaches from the opposite side. “Who is this? What is wrong? How recent is the information? Has anyone confirmed receiving the request? What was last confirmed?” | Reading direction, key-label mapping, human-confirmed versus sensor-verified confusion. Use procedure placeholders only. |
| 13–16 min | Make position data stale, then show a crewmate alert while retaining the wearer's own limitation. “Which information can you trust now?” | False confidence in old ETA, alert ownership errors, and whether the wearer's constraint disappears from attention. |
| 16–18 min | Ask “What was hardest to read or operate? What did you expect each key to do?” Repeat one failed task after a small revision if time allows. | Participant words, concrete hesitation/error, proposed change; mark any repeat as a learning-affected retest. |

For a light/dark comparison, counterbalance order across participants and keep content and viewing conditions constant. A small convenience sample yields formative observations, not statistical evidence that one polarity is superior. Thick-glove and partial-occlusion repetitions should be separate, recorded conditions; do not add them while claiming to isolate another variable.

**Revision rule:** any survival-time misreading, false-current stale data, wrong-person interpretation, or unintended action during escalation requires investigation before adding visual polish. This is a project screening rule, not a flight acceptance criterion. Other repeated reading or navigation issues are ranked by consequence and frequency in the observed sample.

## Blank evidence record

| Version / participant / date | Task and setup | Exact observed behavior or quote | Interpretation (separate from observation) | Change | Retest result |
| --- | --- | --- | --- | --- | --- |
| Not yet tested | — | No observations recorded | — | — | — |

AI source checking, code checks and reviewer critique must be labeled separately from user evidence. Never rewrite “we predict this will be hard” as “users struggled.” Do not invent participant names, quotations, forearm photographs or elapsed human work time.

## Five questions for human-factors review

1. Which validated inputs, uncertainty bounds and mission reserve rule may determine a return estimate, and when must it become unavailable?
2. What are the mission's alert classes, escalation/clearance conditions and priorities when wearer and crewmate hazards coexist?
3. What does the communications protocol actually prove: transmission attempted, transport receipt, human acknowledgment, or responder commitment?
4. What are the physical key centers, travel, force and glove properties, and which independent emergency controls remain available if the display fails?
5. Which crew languages, visual capabilities, visor/lighting conditions and approved procedure provenance must the final validation cover?