The Autonomy Stack: energy, water and robots that build, maintain and grow food.

Anyground ORIJO, research edition v4, 29 October 2026. Machine-readable companion: research/v4/layers.json. Layer dossiers: research/v4/layers/{energy,water,loop,grow,build,care,mind}.json. Concept baseline: research/v3/concept.json.

How to read this report. Every quantitative claim carries a basis label: measured (a field or lab measurement in the cited source), reported (stated by an operator, agency or journalist), claimed (a vendor statement), calculated (derived by the formula shown, from cited inputs), model (a simulation result), target (an Anyground design goal) or assumed (a design assumption to be measured). Every technology carries a maturity label (COMMERCIAL, PILOT, EXPERIMENTAL, SPECULATIVE) and an estimated Technology Readiness Level (TRL 1-9). Citations are inline links of the form [layer:source-id]. Nothing in this report describes built ORIJO hardware: no physical prototype exists yet.

1. Executive summary

A space that powers, waters, builds, maintains and feeds itself anywhere is not one machine. It is a stack of eight layers that trade five currencies (kilowatt-hours, litres, grams of nutrient, robot and human minutes, and bits) through standard sockets. In ORIJO those sockets sit on one 910 x 1820 mm grid, the jo. Nine findings follow from the seven verified dossiers and a completeness pass.

1. Food, not people, dominates the energy budget. One LED farm cassette (HATA-4) draws 14.4-19.2 kWh/day calculated v3:concept, about the same as an entire efficient four-person home at about 14.5 kWh/day calculated energy:src-v3-concept. Growing a full human diet under LEDs would need 167-208 kWh per person per day calculated grow:src-johnson2021 grow:src-kusuma2020. The architecture therefore grows calories in daylight and uses LEDs only for greens, herbs and strawberries, with the photoperiod following the sun.
2. LED efficacy is nearly finished as a cost curve. The best independently tested fixtures reach about 3.0 umol/J measured, against a practical ceiling of 3.4-4.1 umol/J calculated grow:src-kusuma2020. Future kWh-per-kg gains must come from daylight, scheduling and crop choice. That is where ORIJO's design advantage lies.
3. Water is a loop, not a source. The ISS recovers 98% of its water measured water:src-nasa-98. At 90% net recovery a desert home for four needs 44 L/day of make-up, not 320 calculated water:src-anyground-v3. Atmospheric water costs 18-220 kWh/day for that residual, against 0.13-0.24 kWh/day for seawater reverse osmosis calculated water:src-lord-2021 water:src-wiki-desal. Air is therefore the last source. The best air is the farm's own humid air.
4. Nitrogen is the inverse problem. Four people excrete about 12.8 kg of urine nitrogen per year calculated loop:src-fumasoli-2016. One indoor greens rack takes up about 0.25-0.5 kg per year calculated loop:src-anyground-v3. The loop must store nitrogen in a solid "battery", such as zeolite at 86% recovery measured loop:src-beler-baykal-2011, and export it.
5. Construction autonomy is narrow, and that is good news. The robots that pay are the ones doing one repetitive, digitally specified act. Robotic layout cut labour hours by 68% measured build:p-kam2025. Robotic pile driving reaches about three times the conventional rate claimed build:src-rpd-pvmag. Generalist builders burned more than USD 2.6 bn (Katerra and Veev) build:src-katerra-cd build:src-veev-tc. ORIJO robots only mark, anchor and unfold. Unfolding a 1.1 t half-shell takes about 0.0045 kWh of potential energy calculated build:src-concept-v3, so the hard parts are control, sensing and stiction.
6. Maintenance autonomy is mostly the building's job. The most scaled autonomous maintenance robots succeed because they are captive to fixed geometry, not because they are smart. Waterless PV row cleaners claim 3.89 billion panel cleanings claimed care:src-ecoppia-site, and ISS servicing works because its parts were designed for robots care:src-mss-wiki. ORIJO's creases are rails, its ports are orbital replacement units, and every swappable part shares one handle.
7. The mind must be local, deterministic at the bottom and modest in its promises. Field studies with reliable protocols report 16% (residential) and 13% (commercial) savings from advanced control measured mind:src-khabbazi2025, not the 40% data-centre headline mind:src-dm2016. At-scale autonomy still budgets human minutes: Waymo keeps about 70 remote agents on duty for about 3,000 vehicles reported mind:src-waymo-futurism. Freight Farms' bankruptcy stranded customers whose farms ran on its cloud mind:src-freightfarms-ch7.
8. The failures are instructive and consistent. They include more than USD 2.35 bn raised by vertical farms that later failed calculated grow:src-agdive-bowery grow:src-elevenflo-plenty, Boxabl's 744 homes built against 285 delivered reported build:src-boxabl-hw, Henn-na Hotel decommissioning about half its robots reported care:src-henn-futurism, SOURCE hydropanels breaking unserviced water:src-sjv-allensworth and Biosphere 2's oxygen crash measured loop:src-severinghaus-1994. Each was a failure of capital structure, siting, service or sealing, not of physics.
9. Regulation, not physics, now gates the last step. California's direct potable reuse rule, in force since 1 October 2024, demands a validated 20/14/15-log reduction of viruses, Giardia and Cryptosporidium, four separate barriers, UV of at least 300 mJ/cm2, continuous online tracking, and a certified T5 chief operator plus a T3 operator on every shift reported water:src-ca-dpr-text. No household potable-reuse route, no code route for robot-assembled folding shells, and no route for battery banks inside the habitable space (the 2018 IRC bars listed systems above 1 kWh from habitable rooms) was found. ORIJO therefore certifies components on existing routes (NSF/ANSI 350 non-potable reuse, UL 9540 batteries, ICC/MBI 1205 factory certification, test-based approval of the fold) and labels the rest as uncertified autonomy (section 18).

Where ORIJO stands. Most components are COMMERCIAL or PILOT (TRL 6-9). The integration is EXPERIMENTAL (TRL 3-5). The unified autonomy scale (section 13) puts the 2027 bench at L1-L2, the 2029 catalogue home at L3 and the 2032 polar winter-over station at L4. Grow stays one level behind because fruit picking remains human-in-the-loop until robots exceed 90% success.

2. The stack in one picture

The table lists the layers bottom-up. Each layer's maturity mix counts the technology options assessed in its dossier.

The order matters. A layer can only be as autonomous as the layers beneath it allow. A robot cannot swap a cartridge that has no handle. An optimiser cannot shift a farm load if the bus has no priority signal. A water organ cannot release potable water without a fail-closed sensor verdict.

3. Structure (形): the folded body as fixture

Mechanism. The v3 concept defines thick-panel rigid-foldable barrel vaults (YOSHI, M02) that unfold on one degree of freedom from road-legal half-shell packs into a 26.5 m2, about 48 m3 room calculated v3:concept. Membrane hinges with an opening-face bellows (KASANE, M01) carry the seal. The vault stands on self-levelling legs on screw piles (ASHI, M03). Every floor tile is a utility socket (JO-PORT, M04). The v4 synthesis adds a robotic role for the shell's geometry. A datum socket every 910 mm along the ridges turns the building into the robot's coordinate system, and every crease that drains water can also carry a rail.

State of the art. Uniform-thickness origami can bear real loads. Zhu and Filipov's modular origami trusses reconfigure between storage and structural states, and a 1 m column carried 21 kN at 7.4 kg measured build:p-zhu2024. Rigid-panel origami habitats have housed people: SAGA's LUNARK expanded 750% and kept two crew for 60 days at down to -30 C reported build:src-lunark-saga. Softgoods habitats reached full-scale burst at 77 psi against 60.8 psi required measured build:src-life-tc. BEAM has flown on the ISS since 2016 build:src-beam-wiki.

Maturity. The ORIJO shell is EXPERIMENTAL, TRL 4 (metre-scale lab demonstrators). The weathertight, insulated hinge is the single most important risk in the system (v3 M01).

What the other layers ask of the shell. The roof must keep a continuous fall line from every facet to a first-flush gutter (water). Wings and fences must carry PV and their DC harness through the hinges (energy). Grow zones must be airtight with a lung, and fresh cement must never be exposed inside a sealed volume (loop). Rails and sockets must sit at fixed jo heights (grow, care). Latch state must be robot-readable (care, mind).

4. Energy (力): self-powering

4.1 How it works

Crystalline-silicon PV converts photons above the 1.12 eV bandgap into current. MPPT converters feed a 48-60 V DC spine directly. Bifacial east-west fences peak in the morning and evening and shed snow, which flattens the curve that storage must bridge energy:src-next2sun. Storage is split by cost per kWh:

• LFP batteries for 1-2 days of firm load.
• A small sodium-ion locker for cold keep-alive. CATL claims more than 90% capacity at -40 C claimed energy:src-catl-newatlas2026.
• Hot water for everything else. A 500 L stratified tank across 40 K holds about 23 kWh_th, bought with about 8 kWh_e at a COP of 3 calculated energy:src-v3-concept.

Distribution is DC. A Bosch study found DC uses PV 6-8% more efficiently than AC model energy:src-fregosi2015. The larger benefit is that bus voltage becomes a public signal of storage state.

4.2 State of the art, with numbers

• Batteries. Stationary pack prices fell to USD 70/kWh in 2025, down 45% year on year reported energy:src-bnef2025. Home-scale stored energy stays expensive: residential standalone storage costs USD 547-860/MWh against USD 115-254/MWh for 4 h utility batteries reported energy:src-lazard2025-pdf. That gap is the economic reason to shift loads into sun hours instead of storing for them.
• Thermal storage. Sand batteries are commercial. Polar Night Energy's Pornainen unit (1 MW / 100 MWh, about 2,000 t of soapstone) ran a severe winter without its oil boiler and cut network emissions about 70% reported energy:src-pne-essnews2025 energy:src-pne-ie2026. Rondo's 100 MWh brick heat battery has run daily from solar since October 2025 claimed energy:src-rondo-pvmag2025.
• Long-duration storage. Form Energy's 100 h iron-air units began commercial deployment in October 2025 energy:src-form-latitude2025. Modelling shows long-duration storage must reach USD 20/kWh or less to cut system costs by 10% or more model energy:src-sepulveda2021. No home-scale product is close.
• Agrivoltaics. Heggelbach reached 160-186% land-use efficiency measured energy:src-fraunhofer2019. Belgian vertical bifacial fences reached an LER of 1.00-1.18 over sugar beet, while elevated PV over wheat cost 33-46% of yield measured energy:src-reher2024.
• Off-Earth power. Lunar baselines are moving to fission. KRUSTY ran 28 h at 1.5-5.0 kWt with about 25% system efficiency measured energy:src-krusty2020. EXPERIMENTAL, TRL 5.

4.3 Failures

• Vertical farms priced their kWh like storage. Plenty filed Chapter 11 in March 2025 and emerged as a strawberry-only grower energy:src-plenty-elevenflo. Bowery ceased operations in November 2024 energy:src-bowery-yahoo. Jones Food went into administration in April 2025 despite 100% renewable power energy:src-vf-wiki.
• Storage start-ups died before their chemistry mattered. Natron ceased operations in September 2025 energy:src-sodium-wiki. Ambri went through Chapter 11 and a lender takeover in 2024 energy:src-ambri-pvmag2024.
• Small wind underdelivers. Field output deviates widely from rated curves. Tuning one Scoraig turbine raised its yield by up to 156% measured, which shows how far untuned machines fall short energy:src-leary2020.
• Polar stations remain diesel-dependent. Showa is one example energy:src-showa2020.

The lessons: buy commodity cells and innovate in integration; measure wind for 3-6 months before counting on it; publish diesel hours as a KPI instead of claiming zero.

4.4 What ORIJO adopts

• Generation: MIURA-W fold-out wings plus BYOBU-PV vertical bifacial fences, paired by latitude (BYOBU-MAKI).
• Storage: a DENCHI-B cabinet with 20-40 kWh LFP plus 2-5 kWh sodium-ion, and a CO2 heat pump into a 300-500 L tank.
• Seasonal: oversized PV curtailed into heat. Fuel or biomass backup only above the Arctic Circle. No hydrogen by default.
• Control: a rule ladder at launch, then MPC after 12 months of fleet data.

Four new mechanisms:

• HIKARI-DOKI. Gives each cassette a daily light integral instead of a clock and fills it from daylight, then PV, then battery.
• DENATSU-ROPE. Gives every port its own shed and restore voltage pair (for example comfort 49.5 V, farm 48.5 V, water 47.5 V, life safety never), so survival order holds with all computers off.
• IRORI-KURA. A reject-heat cascade tank.
• BYOBU-MAKI. The wing-and-fence pairing rule.

4.5 Autonomy ladder (human minutes per week)

energy:src-v3-concept

5. Water (水): self-watering

5.1 How it works

Streams are separated at the source:

• Grey (showers, laundry, kitchen) goes to a regeneration organ.
• Yellow (urine) goes to the loop layer.
• Brown goes to hygienisation.
• Condensate from people and plants goes back to the grey intake.

The treatment train stacks independent barriers: biology, ultrafiltration, reverse osmosis, UV-C, residual chlorine and activated carbon at the tap. Release is decided every few minutes by a soft sensor with a fixed false-positive ceiling. In the NEST Water Hub, free chlorine and ORP alone classified safe and unsafe reuse water at a false-positive rate of 2% or less measured water:src-reynaert-2023.

The master variable is recovery. Make-up = N x (D x (1 - R) + Dk), where N is people, D is daily use per person, R is net recovery and Dk is drinking and cooking water. With 4 people, 80 L, 90% recovery and 3 L, that gives 44 L/day calculated water:src-anyground-v3.

5.2 Sources, ranked by kWh per litre

5.3 Recycling state of the art

• ISS. The urine processor recovers 87% of US pretreated urine measured. The brine processor has returned about 780 of 946 L since February 2024, lifting total recovery from 93-94% to 98% water:src-ntrs-iss-2026.
• WOTA Unit. Claims 97% domestic wastewater regeneration claimed, passed TRL 6 in 2025 per the press and began household trials in January 2026 water:src-wota-site water:src-dailygalaxy-wota. Its potable status is disputed, so ORIJO treats potable release as conditional on its own guard.
• Hydraloop. Certified to NSF/ANSI 350R, recycles up to 95% of greywater and cuts mains demand up to 45% claimed water:src-hydraloop-site.
• Sneek, Netherlands. Vacuum toilets in 32 houses preserved 7.6 g N per person per day measured water:src-zeeman-2011.

5.4 Failures

• SOURCE hydropanels, Allensworth, California. Many panels broke and went unserviced, each giving about a gallon a day reported water:src-sjv-allensworth.
• ISS urine processor and toilet. Recovery was cut from 85% to 75% in 2009 over urine chemistry. Purge-pump failures and a toilet stand-down followed in 2024-26 water:src-ntrs-iss-2026.
• Autarky toilet. Treated water regrew E. coli, Pseudomonas and Salmonella in storage without a residual disinfectant water:src-nguyen-2017.
• Lab-to-field gap. Lab MOF projections of 2.8 L/kg/day water:src-kim-2017 against field yields of 0.1-0.7 L/kg/day water:src-hanikel-2019.

The lessons: sell service and spares, not panels; confirm every chemical dose with a sensor; carry a residual disinfectant in every store; size from single-cycle field data.

5.5 What ORIJO adopts

• MIZU-1: a WOTA-class regeneration organ.
• KAWAYA-U: urine diversion.
• Rain on every facet.
• PV-RO or PV-ED wherever any saline source exists.
• KUMO-12 sorption racks only in no-source deserts, regenerated with heat-pump or PV-T waste heat in a closed vapour loop. A PoliTo patent claims 2.7 kWh_th/L closed-loop against 4.7 open-loop claimed water:src-pat-polito.

Four new mechanisms:

• KUMO-NAE. Harvests the farm's humid air.
• SHIO-BUKURO. A brine bladder after the ISS brine processor.
• MIZU-MAMORI. The fail-closed release controller.
• AMA-TSUYU. A rain-and-dew roof coat.

Autonomy ladder. L1 is a certified greywater appliance (30 min/week). L3 is source-separated streams with a soft-sensor interlock (10 min/week). L4 is 95-98% closure with brine drying and robot cartridge swaps (5 min/week). L5 needs regulators to accept sealed sensor logs as compliance, which does not exist yet.

6. Loop (環): self-closing

6.1 How it works

The loop is a ledger before it is a machine. NASA's BVAD reference adult consumes 0.895 kg O2 and 0.8 kg of dry food per day and produces 1.085 kg CO2 and 1.42 kg of urine water reported loop:src-bvad-2022. A per-element ledger (O2, CO2, water, N, P, K, Na) reconciles every stream daily. A residual beyond +/-10%, or a missing hygienisation log, automatically blocks release to edible crops.

Yellow loop. Stored urine is settled with magnesium to remove phosphate as struvite (95% average P removal, measured) loop:src-wilsenach-2007. It then passes clinoptilolite, which takes up 97% of ammonium and releases 88% on elution, 86% overall measured loop:src-beler-baykal-2011. The commercial Aurin route uses nitrification plus distillation. Its distillation step costs 71 W per capita of primary energy, plus 13 W for nitrification calculated loop:src-fumasoli-2016. That is affordable at hotel scale on PV surplus, not in a home.

Brown and green loop. Kitchen scraps are dried overnight and fed to black soldier fly larvae. In a pilot, 190 kg of food waste gave 79 kg of larvae measured loop:src-gligorescu-2020. Frass and faeces are hygienised at 55 C for 3 days (reported rule) loop:src-cfr-503. Dry residue becomes biochar (+25% yield in the tropics, no average effect in temperate soils, measured) loop:src-jeffery-2017 or mycelium insulation infill loop:src-elsacker-2019.

Air loop. On Earth the atmosphere is a free buffer, so ORIJO does not seal the house. Lettuce produces only 7.78 g O2/m2/day, so supplying one person's oxygen would take about 115 m2 of it calculated loop:src-bvad-2022. Leafy greens are not an air system. Instead, sealed grow zones receive living-room CO2 through a bellows lung, and plant CO2 drawdown is logged as a crop-health signal. That trick comes from SAM at Biosphere 2 loop:src-sam-b2-2026.

6.2 The analogue lineage

6.3 Failures

• Biosphere 2. Oxygen fell from 21% to 14% in 16 months because soils respired and fresh concrete absorbed CO2 loop:src-severinghaus-1994.
• Insect protein. Ynsect, Aspire, Enorm, AgroLoop and Goterra together had raised about USD 777 M before failing reported loop:src-insectinstitute-777m.
• Household digesters. Montana household digesters ended at pH 5.3-6.1 in a cold climate measured loop:src-ebel-2025.
• MELiSSA nitrite excursions that inhibited downstream photosynthesis loop:src-garcia-gragera-2021.
• NASA's BIO-Plex was discontinued before integrated crewed testing loop:src-porterfield-2025.

The lessons: separate atmospheres; treat BSF as an internal service, not a revenue line; use online nitrite sensing with automatic bypass; and ship each loop as a standalone product that pays for itself before integrating.

6.4 What ORIJO adopts

• TSUBO-Z: a 20-25 kg zeolite nitrogen-battery cartridge.
• KAMADO-2: a dry buffer, a larvae tower and a logged 55 C hot box.
• FUIGO: the CO2 bellows lung.
• WARA-ITA: straw-mycelium infill cassettes regrown at each SHIKINEN renewal.

Aurin's Swiss authorisation for edible crops (provisional in 2016, edible crops in 2018) loop:src-eawag-aurin and a 201-reference meta-analysis showing no systematic yield penalty for excreta-derived fertilisers loop:src-schroeder-2026 make the regulatory path real, but narrow.

Autonomy ladder. L1 is manual bins and dry toilets (90 min/week). L3 is automated dosing and self-harvesting larvae (20 min/week). L4 is daily per-element reconciliation, automatic routing and robot cartridge swaps (10 min/week).

7. Grow (農): self-feeding

7.1 How it works

Fully automated food production is intralogistics plus a crop twin. Plants ride standard carriers past fixed stations. A model-predictive controller sets climate and light. Robots do the night work.

Physics sets the boundary. Lighting energy is E = DLI x 365 x area / (PPE x 3.6) kWh/yr, where DLI is the daily light integral in mol/m2/day and PPE is fixture efficacy in umol/J. At DLI 15 and 3.0 umol/J, lettuce needs 7.2-10.6 kWh/kg of fresh weight for light alone calculated grow:src-kusuma2020. A full HATA-4 cassette (6.62 m2 of tier area) needs 9.2 kWh/day of lighting calculated grow:src-anyground-v3. Above about 12.6 mol/m2/day, extra light adds no significant lettuce weight while efficiency falls measured grow:src-yan2019.

7.2 State of the art

• Control is solved enough. In the Autonomous Greenhouse Challenge, an AI team earned EUR 24.78/m2 against 21.18 for expert growers on cucumber measured grow:src-hemming2019. In the second edition, all five AI teams beat the human reference mind:src-hemming2020.
• Logistics is solved. Spread's Techno Farm was designed for 30,000 heads a day target with 50% less labour and 98% water recycling claimed grow:src-fruitnet-spread2017.
• Harvesting is not solved. Across 50 harvesting robots, average success was 66% at 33 s per fruit measured grow:src-bac2014. SWEEPER managed 61% in a crop adapted for it and 18% in a normal one measured grow:src-arad2020.
• Protection pays first. Thorvald night UV-C rovers claim a 70-100% pesticide reduction claimed grow:src-saga-site.
• Sealed analogues are the honest worst case. EDEN ISS grew 268 kg in 286 days on 12.5 m2 (21.4 kg/m2 per season) measured grow:src-zabel2020.
• Calories need area. Supplying dietary calories takes 40-50 m2 of high-light crops per person reported grow:src-johnson2021.

7.3 Failures

More than USD 2.35 bn was raised by the named failures calculated: Bowery, Plenty, 80 Acres, Monarch and Iron Ox grow:src-agdive-bowery grow:src-agfunder-plenty grow:src-agfunder-80acres grow:src-tc-monarch grow:src-robotreport-ironox. Infinite Acres, the technology arm tied to 80 Acres, went bankrupt with its parent in September 2026 grow:src-hortidaily-infiniteacres. Iron Ox used robots for logistics a conveyor could do. Monarch's autonomy claims ended in a dealer lawsuit.

The lessons: separate the hardware business from any farming operation; never finance cassettes with growth-story debt; publish the autonomy level per module; and use robots only where coordinates are unknown or change.

7.4 What ORIJO adopts

HI-NO-WAKE (the sun split) assigns each crop to a daylight vault, an LED cassette or an outdoor bed by kWh per kg of nutrient delivered. Other elements:

• KAGO-MICHI: half-jo carriers on passive rollers with one stationary vision-and-cutter cell.
• YOAKE-UV: a pre-dawn UV-C bar on the ridge rail.
• TSUBO-KABE: a fruit wall where every truss hangs at a fixed jo coordinate.
• MUSHI-DAI: a larvae bench fed only crop residue.

One HATA-4 covers greens for about 9-13 people at 100 g/day calculated grow:src-anyground-v3.

Autonomy ladder. L2 is carriers through stations (60-90 min/week per cassette). L3 adds a crop twin and whole-tray greens harvest (30-45 min/week; v3 target 30 or less). L4 needs more than 90% fruit-picking success in a robot-designed crop (10-20 min/week, target). L5 is a Lunar Palace-class sealed loop and is SPECULATIVE without resident crew.

8. Build (建): self-building

8.1 How it works

The factory does the work and the package carries the complexity. A licensed container micro-factory (TATAMI-KOUJOU) CNC-cuts, laminates, wires and fold-tests panels, following capital-light precedents:

• AUAR: up to 180 homes/yr per micro-factory claimed build:src-auar-trr.
• Reframe FAB1: under USD 5 M of equipment for up to 250 single-family homes/yr reported build:src-reframe-trr26.

On site, three robotic acts remain:

1. Mark. RTK staking rovers are commodity tools, at about 8-9 mm claimed build:src-civ-site.
2. Anchor. This is the most under-automated step build:p-melenbrink2020. Romu showed small robots can drive posts and sheet piles build:p-romu2020. Built Robotics' RPD 35 drives about 300 solar piles a day against about 100 conventionally claimed build:src-rpd-pvmag.
3. Unfold. FBR's Hadrian X compensates its boom with a laser tracker at 250 Hz or more, preferably 1,000 Hz reported build:src-hadrian-pat. ORIJO borrows that two-loop structure for its fold controller. From NASA's ARMADAS, which assembled and reconfigured a 256-cell lattice using the lattice itself for indexing measured build:p-gregg2024, it borrows structure-relative metrology.

8.2 Failures

• Katerra raised more than USD 2 bn and filed Chapter 11 in 2021 build:src-katerra-cd.
• Veev raised USD 600 M and closed in 2023 build:src-veev-tc.
• Boxabl had 744 homes built against 285 delivered by August 2025, with a 2024 net loss of about USD 51 M reported build:src-boxabl-hw build:src-boxabl-bo.
• 3D printing shook out: ICON cut 25% of its staff build:src-icon-tc25 and Diamond Age was liquidated build:src-icon-3dp25.
• BEAM's first expansion stalled. The second attempt took about 7 h of staged air pulses build:src-beam-wiki.
• Hadrian took about 9 years from patent priority to sale calculated build:src-hadrian-pat build:src-hadrian-ie25.

The lessons: stay asset-light; ship anchoring, autonomy and a permit pack with every kit; do not print shells; add a dwell-time rule and staged unfolding against stiction; and budget years of field iteration for any new site robot.

8.3 What ORIJO adopts

• ORI-SEQ: a closed-loop fold sequencer.
• SUMI-ITO: an anchoring rover named after the carpenter's ink line, which logs a torque-depth curve per pile as proof of capacity.
• KUSABI: a self-tightening wedge latch that proves its own state.
• TATAMI-KOUJOU: the two-container factory cell.

Site labour for a four-person home is 30-38 person-hours over 2-3 days (calculated from v3 targets) build:src-concept-v3.

Autonomy ladder. At L1 (machine-assisted), the same home needs 30-38 person-hours on site. L2 (robotic anchoring with logged torque) brings that to 15-20. L3 (supervised autonomy, the shell unfolding under closed-loop control) brings it to 6-10. L4 adds autonomous re-levelling and storm refold. L5 is a self-growing settlement, the lunar pre-deployment case.

9. Care (守): self-maintaining

9.1 How it works

A fully automated habitat does not need a humanoid that fixes everything. It needs a few captive robots doing scheduled tasks on fixtures the building provides, organ-level swaps instead of repairs, a fleet ledger that learns real failure rates, and a remote operator for exceptions.

Space shows the pattern. Canadarm2 and Dextre work because ISS orbital replacement units carry grapple fixtures and standard bolts care:src-mss-wiki. Astrobee needed crew for only about 220 of more than 1,200 console hours (about 80% crew-free, calculated). Its own team judged that it had too many replaceable units and that localisation in a changing interior was its biggest problem care:src-astrobee-tfr.

9.2 State of the art

• Waterless PV row robots. Each covers about 9.3 m2/min reported care:src-ecoppia-ketura. Soiling still costs at least 3-4% of global solar output care:src-ilse-joule.
• Legged inspectors. Spot and ANYmal run for about 90 min per charge reported care:src-spot-spec care:src-anymal-spec, so docks are essential.
• Learned manipulation is improving fast. Diffusion Policy averaged 46.9% better than the prior state of the art measured care:src-dp. Toyota's large behavior models match single-task performance with under 30% of the task data in simulation calculated care:src-lbm.
• Home humanoids still lean on people. 1X NEO costs USD 20,000 or USD 499/month and was mostly teleoperated when the WSJ tested it care:src-neo-newatlas care:src-yt-neo-wsj. General home manipulation in 2026 still needs remote humans.
• Fleet reliability data changes the spares plan. Bayesian updates of ISS life-support data showed 18% of replaceable units failing more often than predicted (one about 7.8x) and saved 3.9-6.0 t of spares for a Mars-class mission measuredcalculated care:src-owens-iss.

9.3 Failures

• Henn-na Hotel decommissioned about half of its 243 robots because they created work care:src-henn-futurism.
• iRobot filed Chapter 11 and was taken over by its contract manufacturer care:src-irobot-md.
• Ozmo, a facade-cleaning robot, still needs a rooftop supervisor, and its lidar struggles on glass care:src-ozmo-trr.
• Hotel staff shelve robots that add workload care:src-chloe.

The lessons: keep robots back-of-house; sell building uptime, not robots; use fixtures, not visual maps; measure staff minutes added, not only saved.

9.4 What ORIJO adopts

• KAWARI handle. One robot handle and lead-in standard for every swappable organ, cartridge and robot wear part.
• HOKI. A waterless broom on the wing hinge rail.
• MIMAMORI. A 3 kg patrol head with thermal, ultrasonic, NH3/VOC and RGB sensing.
• DAICHO. The fleet failure ledger.
• MADOGUCHI. The remote desk.

The sizing for a home is about 245 robot-minutes and about 6 remote-operator minutes per week. That puts roughly 200 homes on one operator FTE (calculated from targets) care:src-astrobee-tfr. With organ swaps, availability is set by mean time to repair, not mean time between failures. Taking the 500 h starting prior for robot MTBF (a pessimistic assumption), a 0.5 h swap gives 99.9% availability, while a 48 h on-site repair gives about 91% calculated.

10. Mind (心): self-governing

10.1 How it works

The mind is a stack of five parts:

1. A semantic nervous system. Field devices publish to an MQTT Sparkplug bus, described by a Brick/223P graph mind:src-balaji2016.
2. A deterministic safety kernel. It owns every actuator.
3. Optimisers that propose setpoints. Rules, grey-box MPC and warm-started reinforcement learning.
4. A fleet coordinator. Open-RMF lets robots share rails, doors and chargers mind:src-openrmf-news.
5. A remote desk that answers questions instead of steering.

LLMs translate intent into constraints and never write to actuators. A 2026 study found LLM schedulers fabricated schedules under constraint conflicts mind:src-jonah2026.

10.2 State of the art

• Data centres to buildings. Google's ML cooling cut cooling energy by up to 40% in recommendation mode mind:src-dm2016 and by about 30% after nine months of autonomous control mind:src-dm2018. The same approach in commercial buildings saved 9% and 13% measured mind:src-luo2022.
• Homes. In one occupied house, MPC saved 18.1% and RL saved 20.9% of heat-pump energy, with overlapping confidence intervals measured mind:src-mulayim2025.
• Field evidence base. Of 104 field demonstrations, 71% used protocols that may be unreliable, and only 13 reported deployment costs measured mind:src-khabbazi2025.
• Unattended operation. Antarctic geophysical observatories have run unattended on about 100 W since 1996. They coast about 4 days on batteries at about 50 W, and their designers recommend oversizing by 5x or more and tolerating the loss of any two of three sources reported mind:src-melville2014. Halley VI now runs autonomously through winter mind:src-bas-halley.
• Regulation. The IMO's code for autonomous ships becomes a template for remotely supervised autonomy: voluntary from 1 July 2026, mandatory targeted for 2032 mind:src-imo-mass.

10.3 Failures

• Freight Farms filed Chapter 7, its farm cloud was phased out, and growers reverse-engineered their own controls mind:src-freightfarms.
• Johnson Controls had its own IT encrypted by ransomware mind:src-jci2023.
• Lunar Gateway was paused in March 2026 mind:src-gateway-ifl.
• Waymo faced a February 2026 inquiry over where its remote assistants sit mind:src-waymo-futurism.

The lessons: the full stack runs locally with escrowed firmware; the safety kernel accepts only signed envelopes; never tie the autonomy stack to one programme; publish the human-minutes ledger.

10.4 What ORIJO adopts

• BANSHO safety kernel. About USD 1,500 per KURA-3 (estimate).
• KIROKU fold-passport graphs. Docking merges them, removing commissioning labour.
• YOBIDASHI remote-assistance queue.
• FUYU-GOMORI island mode. A 4-14 day coast target with five survival tiers.

At L3, 15-60 human minutes per site per week puts 40-160 homes on one operator calculated. Waymo's ratio is about 1:43 reported.

11. Interfaces: what crosses between layers

The JSON holds the full 8 x 7 ordered matrix: 56 flows, each with a unit, a typical value, a basis and a design rule. The flows that shape the architecture:

Three patterns recur:

1. Every layer exports heat or water to another. The tank and the grey intake are the two universal sinks.
2. Every safety decision lives in the lowest layer able to make it: the voltage ladder, the water guard, the hygienisation log, the latch sensor. Higher layers only schedule.
3. Every robot task depends on a structural affordance: a rail, a socket or a handle.

12. Balances: home, hotel, farm

The formulas, sources and labels for every line are in layers.json -> balances. Summary:

12.1 Desert home, 4 people (target L3)

Energy: 25.3 kWh/day calculated

At desert sun (6 peak sun hours, performance ratio 0.75) this needs 5.6 kWp. The installed 8.5 kWp (MIURA-W plus BYOBU-PV) leaves about 13 kWh/day of surplus for the heat bank and sorbent regeneration. Firm night-capable load is 9.4 kWh/day, so two firm days need a 22 kWh battery, which falls inside DENCHI-B's 20-40 kWh.

Water calculated

• Treated: 320 L/day (160 showers and washing, 60 laundry, 48 kitchen, 12 drinking and cooking; dry toilet).
• Recycled: 288 L/day.
• Make-up needed: 44 L/day, sourced as:
• rain, 8.8 L/day annual mean;
• KUMO-12, 2.4-12 L/day target;
• a residual of 23-33 L/day, from a monthly 0.7-1.0 m3 delivery or a brackish-well PV-RO at 0.13-0.24 kWh/day.

The honest gap: a desert home at 90% recovery is not water-independent from air alone. At 95% recovery, make-up falls to 28 L/day.

Food

• 2.6-3.8 kg/week of leafy greens on 2.8 m2 of tier area calculated.
• 0.3-0.5 kg/week of herbs on 0.5 m2 (assumed yield).

Against a 2.8 kg/week greens demand, that is a surplus. Calories from the site are under 5% calculated.

Nutrients

• Urine N: 35 g/day (design) to 44 g/day (theoretical).
• 30 g N/day captured on zeolite; 0.7-1.4 g N/day used by the farm; about 29 g N/day exported.
• P: 4.0 g/day excreted, 3.8 g/day recovered as struvite.
• K: about 15 g/day, low confidence (estimated from the ratio in Aurin's product).

Waste

Robot minutes: about 60/day

Human minutes: about 66/week

12.2 TABI-NO-YADO hotel, 6 keys

• Energy: 82-91 kWh/day, needing 27-30 kWp at 4 peak sun hours. The v3 default of about 17 kWp is short (conflict flagged). Battery: 58 kWh.
• Water: 1,300 L/day treated; 172 L/day make-up, roughly covered by rain on 126 m2 at 600 mm/yr.
• Food: breakfast greens of 4.2 kg/week from one HATA-4. Cherry tomato (4.2 kg/week) needs a YOSHI-6G vault that the v3 default lacks.
• Nutrients: 7.8 person-equivalents give 69 g N/day, of which 59 g is recoverable.
• People: housekeeping (370 human min/week) dominates and is not robot-replaceable in 2026. Building care is 30 min on site plus 10 remote; remote operations up to 120 min/week. Site build: 74-78 person-hours at L1, 42-44 at L2.

12.3 HATA-MURA farm module

• Energy conflict: six lit cassettes need 86-115 kWh/day against 51-77 kWh/day from about 17 kWp. Light only as many cassettes as the design month covers (2-4) and lean on the 79 m2 of daylight vaults.
• Food: greens of 37-53 kg/week if all cassettes are lit, 18-27 kg/week with three; duckweed 0.5-0.9 kg dry/week.
• Water: 78-157 L/day of transpiration recaptured; net loss 11-19 L/day.
• People: about 180 operator minutes/week.
• Nutrients: the farm is a nitrogen sink. One person's urine N can feed 6-13 racks.

12.4 Conflicts found and resolved

13. One autonomy scale for the whole site

MTBI is the mean time between human interventions. "Island days" are days without resupply or network.

The scale combines the mind ladder (MTBI, island days, human minutes) mind:src-khabbazi2025 with closure criteria from the water water:src-ntrs-iss-2026, loop loop:src-fu-2021-biorxiv and grow layers. A site is rated at the lowest level any layer meets.

Where ORIJO pilots sit.

14. A day in the life: desert home, early October

The values below come from the home balance. The battery state of charge (SOC) trajectory is illustrative, calculated from the balance rather than simulated hour by hour.

15. Build sequence: container to first harvest

16. Frontier combinations: implementable now

These cross-layer combinations are non-obvious but buildable with today's parts. Prototype costs are estimates.

A household-scale model of food-water-energy interactions syn:hussien-2017-wef supports treating these couplings as a design object rather than an afterthought. No published system combines more than two of them.

17. Failures atlas: what investors should remember

1. Vertical farms. Bowery, Plenty, Jones Food, 80 Acres and Infinite Acres, plus Freight Farms' cloud. Debt-funded lit commodity crops; a vendor tied to one operator dies with it. Use daylight for calories and local control. grow:src-agdive-bowery grow:src-hortidaily-infiniteacres mind:src-freightfarms-ch7
2. Katerra and Veev. Stay asset-light; use one grammar. build:src-katerra-cd build:src-veev-tc
3. Boxabl. The fold is solved; site and permission are not. build:src-boxabl-hw
4. 3D-printing shake-out. Automating one trade does not make a house cheap. build:src-icon-tc25
5. BEAM stall. Stiction after packed dwell. build:src-beam-wiki
6. SOURCE in Allensworth. Service beats hardware. water:src-sjv-allensworth
7. ISS urine processor and toilet. Dose sensors and swappable wear parts. water:src-ntrs-iss-2026
8. Biosphere 2. Do not share air with soil; no fresh concrete inside sealed volumes. loop:src-severinghaus-1994
9. Insect-protein collapse. Internal service, not revenue. loop:src-insectinstitute-777m
10. Henn-na and iRobot. Back-of-house robots inside a service contract. care:src-henn-futurism care:src-irobot-md
11. Harvest robots. Design the crop for the robot. grow:src-bac2014
12. Natron and Ambri. Buy commodity storage. energy:src-sodium-wiki energy:src-ambri-pvmag2024
13. Antarctic observatory generators. No combustion in the critical path at polar sites; oversize 5x. mind:src-melville2014
14. AI headline shrinkage and LLM fabrication. Promise 10-20%, measure with switch-back tests, keep LLMs off actuators. mind:src-luo2022 mind:src-jonah2026
15. Autarky. Residual disinfection in every store. water:src-nguyen-2017

18. Certification paths: what a regulator will accept

Regulation, not physics, now gates the last autonomy step. Every layer has a certification route for a non-autonomous product; none has a route for an unattended household system that makes drinking water, holds large battery banks indoors, sells produce or is assembled by robots without per-site inspection. Strategy: certify components on existing routes (NSF/ANSI 350 non-potable, UL 9540, ICC/MBI 1205 plant certification, ICC-ES helical piles), certify novel structure by full-scale testing, and keep learned software out of certified safety functions.

Three consequences.

• Water autonomy is split in two. The DPR train is the engineering benchmark for MIZU's potable release: four barriers, three mechanisms, fail-closed on the 16/10/11-log floor, which the rule requires at all times, while the full 20/14/15 must hold for at least 90% of production time reported water:src-ca-dpr-text. But the rule covers municipal systems with certified operators. Water L3 in the 2029 catalogue home is therefore redefined as unattended non-potable recycling plus drinking water from rain, air or delivery through a certified point-of-use train (balances, conflict 'Water autonomy level vs regulation').
• The shell is certified like a factory product, not like a 3D print. 3D printed concrete has been assembling its route in layers: an optional IRC appendix (2021), ICC-ES acceptance criteria AC509 (2023), and an ICC consensus standard under development since October 2023 build:src-icces-ac509 build:src-3dprint-icc-walls build:src-irc2024-appbm. Researchers still call the absence of approval procedures the main brake build:src-vasilic2025. ORIJO's nearer route is ICC/MBI 1205 plant certification, which needs 100 percent inspection of at least one initial module build:src-thau-1205 build:src-evstudio-icc-mbi, plus design-by-testing for the hinge and vault, as the Eindhoven printed bridge did build:src-salet2018. Each licensed micro-factory carries its own certification cost.
• Safety functions stay deterministic. The EU Machinery Regulation applies from 20 January 2027 and adds rules for AI-powered safety functions and for cyber-safety mind:src-ec-machinery. ORIJO keeps learned models in the proposing layer, and the certified safety function stays in the local kernel (section 10).

Not verified in this pass: NSF/ANSI 350 numeric effluent limits and test duration (NSF pages 403); NFPA 855 / IRC R328 residential ESS aggregate limits (snippets only); EU 2020/741 Class A numeric limits; Japanese routes: MLIT structural type approval for folding shells; MHLW position on WOTA-type household potable reuse; ICC 1150-2026 text and adoption; Food-code rules for a hotel serving its own produce (US FDA Food Code, Japan Food Sanitation Act).

19. Environments: where each layer breaks first

The v3 concept defines eight environments v3:concept. The table names the layer that limits autonomy first at each one, and the analogue evidence behind it.

Arctic is the nearest hard proof. Naurvik in Gjoa Haven, Nunavut, has grown basil, strawberries and cherry tomatoes in three 168 ft2 container pods since 2019. It runs mostly on wind and solar and is operated by local technicians reported grow:src-vfd-naurvik. It publishes no yield, kWh/kg or cost figures, and that gap is ORIJO KITA's opening. In container farming the survivor sells service: Growcer bought Freight Farms' assets in July 2025 and offers a fully supported farm for under CAD 1,000 per week reported grow:src-vfd-growcer-2026. The polar failure to remember is the PENGUIn AGO observatories, whose propane generators froze below about -40 C mind:src-melville2014.

Lunar is a physics case, not a market. On the Moon the night lasts about 14 days, which forces fission or very heavy storage. NASA's Fission Surface Power target is 40 kWe for 10 years under 6,000 kg, and a 2025 directive asks for at least 100 kW by late 2029 target energy:src-fsp-wiki energy:src-sciam2025. ORIJO-specific lunar elements remain SPECULATIVE (TRL 2-3). Their value is that the same organ interfaces serve a polar station.

20. Cost ledger: what is priced and what is not

Known prices with sources. Components are cheap; the unknowns are integration, certification and service. No ORIJO bill of materials exists yet.

Unpriced, and needed before a catalogue price:

• Price of a household potable or near-potable recycler (WOTA Unit price not published)
• Sorbent (KUMO) cost per L/day of capacity and lifetime
• ORIJO shell cost per m2 from a licensed micro-factory
• Installed off-grid PV and storage cost at remote sites (logistics dominate)
• Certification costs: NSF/ANSI 350 test, UL 9540 listing, ICC/MBI 1205 plant certification, full-scale fold testing
• Service cost per site-year (the item that killed SOURCE in Allensworth and Freight Farms customers)

The ledger says the investable risk is not hardware cost. A 30 kWh LFP pack is about USD 2,100 at 2025 pack prices calculated. A field-ready quadruped was USD 74,500 at launch. Freight Farms' Chapter 7 left USD 7.01 million of liabilities against USD 0.59 million of assets reported mind:src-freightfarms mind:src-freightfarms-ch7. The costs that decide survival are service per site-year and certification, and neither is published for any comparable.

21. Investor view

21.1 Why now: cost curves with numbers

The investable point is the combination. Generation, storage, sensing, compute and robot bodies all became cheap in the same five years. Meanwhile the dominant food-energy term (LED efficacy) has stopped falling, so advantage now goes to architectures that use sunlight and schedule light, not to better lamps.

21.2 Moats

• The jo grid as one coordinate system. Shell, carriers, rails, datums and ports share one 910 mm pitch.
• Catalogue modules with fold passports. These remove commissioning labour.
• Fleet data from identical siblings. Labelled failure and performance data that single buildings never have.
• Integration IP across layers. Voltage-rope priority, DLI-to-sun scheduling, reject-heat cascade, farm-air water and the zeolite N battery.
• The SHIKINEN lease. It aligns Anyground's incentive with fewer human minutes.
• Public honesty ledgers. Human minutes, litres per kWh and autonomy level per module.

21.3 Risks

• Structure: the KASANE hinge is unbuilt.
• Energy: LED cassettes overrun the array.
• Water: no multi-year independent safety data for household potable reuse, and no online pathogen-specific sensing.
• Loop: no certified automated household urine-to-fertiliser appliance.
• Robots: no public home-robot MTBF data, and rail locomotion is untested in ice and sand.
• Mind: LLMs are unproven in actuation paths, and remote operations carry cyber exposure.
• Regulation: potable reuse needs certified operators every shift in California, and no route was found for household potable reuse, robot-assembled shells or unattended battery banks (section 18).
• Capital structure: this is where most of the precedents actually died.

21.4 Milestones

22. Visual briefs for the /layers/ section

Eight still briefs and two film briefs are in layers.json -> visual_briefs. They use the v3 global style: photoreal, Japanese minimalism, washi-white panels, hinoki, charred cedar, bronze, no text.

Stills

Films

All are AI concept visuals, not photographs of hardware.

23. Method and limits

Inputs. Seven layer dossiers, each produced by a SerpApi and Elicit sweep with primary reading and an adversarial verification pass, plus the v3 concept. Together the dossiers now register 544 unique source URLs, 216 papers, 195 projects and 207 numbers.

This synthesis. It added 9 SerpApi calls and 2 Elicit queries under the budget-guarded helper (341 of 380 SerpApi units used in total at that point; 116 Elicit queries logged across all layers). Ten new sources were read on 29 October 2026 for the why-now table and the frontier combinations.

Completeness pass (same day). A critic pass checked the stack against named programmes, regulation, environments and cost. Coverage of the named programmes (BVAD, MELiSSA, EDEN ISS, the Autonomous Greenhouse Challenge, ICON Olympus, Form Energy, WOTA, SOURCE, Aurin, Open-RMF, Spot and ANYmal, Iron Ox, AeroFarms, Bowery, Lunar Palace 365 and Biosphere 2) was already complete. The pass added sections 18-20 and spent 16 SerpApi calls (layer "gaps") and 3 Elicit queries. The changes are logged in research/v4/gaps-log.md.

Verification. All 201 source ids cited in layers.json or in this report are registered in layers.json -> sources, and every dossier-prefixed id was checked programmatically against its dossier registry. (The first synthesis registered 110 ids; 49 ids cited only in the report were added in the completeness pass, together with 42 new ones.)

Limits

• Balances are calculated from design targets and assumptions, not measured hardware.
• Hourly behaviour (section 14) is illustrative.
• The K balance is low-confidence.
• Several vendor numbers are claims (WOTA 97%, Ecoppia counters, RPD 3x, Atoco litres).
• Li et al. 2020's PV-cooling numbers could not be read past a truncated abstract and are not quoted.
• The Unitree price comes from the vendor page, not an independent purchase.
• The lidar price is a single executive's quote relayed by two outlets.
• No independent audit of any vertical farm's kWh/kg, or of any home robot's MTBF, was found.
• NSF/ANSI 350 limits, NFPA 855 residential caps and EU 2020/741 class limits were not read from primary text (section 18 lists them as unverified).
• Arctic farm evidence (Naurvik, Growcer) is journalistic and carries no yield or energy data.

ORIJO itself remains unbuilt. The value of this document is that it shows where each layer's evidence ends.

References cited

• build:p-gregg2024: Ultralight, strong, and self-reprogrammable mechanical metamaterials (Science Robotics 9(86):eadi2746, 2024)
• build:p-kam2025: Comparison of dusty robotics and traditional layout methods (Construction Robotics, 2025)
• build:p-melenbrink2020: On-site autonomous construction robots: Towards unsupervised building (Automation in Construction 119:103312, 2020)
• build:p-romu2020: Autonomous anchoring for robotic construction (Automation in Construction 120:103391, 2020)
• build:p-zhu2024: Large-scale modular and uniformly thick origami-inspired adaptable and load-carrying structures (Nature Communications 15, 2024)
• build:src-3dprint-icc-walls: 3DPrint.com (19 Feb 2025) ICC to Set Standards for 3D Printed Walls
• build:src-auar-trr: AUAR and ABB bring robotic Micro-Factories to U.S. homebuilding (The Robot Report, 2024-09-26)
• build:src-beam-wiki: Bigelow Expandable Activity Module (Wikipedia, citing NASA CRS-8 overview, The Verge, SpaceNews)
• build:src-boxabl-bo: Boxabl secures US$3.5b valuation while fewer than 300 Casitas have been delivered (Built Offsite, 2025-09-19)
• build:src-boxabl-hw: Boxabl valuation at $3.5 billion tests housing fundamentals (HousingWire, 2026-07-30)
• build:src-civ-site: Civ Robotics home page (CivDot+)
• build:src-cobod-bod2: COBOD BOD2 product page (specifications)
• build:src-concept-v3: Anyground research/v3/concept.json (ORIJO mechanisms M01-M13, modules, typologies, value chain)
• build:src-cuby-hw24: Cuby Breaks Ground On Mobile Micro Factory U.S. Home Model Output (HousingWire, 2024-04-02)
• build:src-esa-dshape: Building a lunar base with 3D printing (ESA, 2013-01-31)
• build:src-evstudio-icc-mbi: EVstudio (31 Oct 2024) ICC/MBI 1200, 1205 and 1210 Standards for Offsite Construction
• build:src-hadrian-ie25: World's most advanced bricklaying robot completes factory tests (Interesting Engineering, 2025-10-13)
• build:src-hadrian-pat: US11842124B2 Dynamic compensation of a robot arm mounted on a flexible arm (Fastbrick IP; priority 2016-07-15)
• build:src-icc1150-store: ICC Store listing: ICC 1150-2026 Standard for Automated Construction Technology for 3D Printing Walls (search snippet only, not read)
• build:src-icces-ac358: ICC-ES AC358 Helical Pile Systems and Devices (April 2025 edition per purchase link)
• build:src-icces-ac509: ICC-ES AC509 3D Automated Construction Technology for 3D Concrete Walls (approved Aug 2023, editorially revised Aug 2024 per purchase link)
• build:src-icon-3dp25: ICON Secures $56M Amid Construction 3D Printing Sector's Growing Pains (3DPrint.com, 2025-02-17)
• build:src-icon-nasa: NASA Has Awarded ICON $57.2M to Develop 3D Printed Construction on the Moon (3Dnatives, 2022-12-02)
• build:src-icon-tc25: ICON, a builder of 3D-printed homes last valued around $2 billion, cuts about 25% of staff (TechCrunch, 2025-01-09)
• build:src-irc2024-appbm: ICC Digital Codes - 2024 IRC Appendix BM 3D Printed Building Construction (formerly 2021 IRC Appendix AW); search snippets only, page 403; UpCodes snippet: only materials listed in a UL 3401 report may be used
• build:src-katerra-cd: What does Katerra's demise mean for the contech and modular industries? (Construction Dive, 2021-10-13)
• build:src-life-tc: Successful failure: Sierra Space's inflatable habitat blows up as planned (TechCrunch, 2024-01-22)
• build:src-lunark-saga: SAGA Space Architects - LUNARK project page
• build:src-reframe-trr26: Reframe Systems raises $40M to scale its robotic microfactories for home building (The Robot Report, 2026-08-31)
• build:src-rpd-pvmag: Robotic solar pile driver three times faster than humans (pv magazine USA, 2023-04-27)
• build:src-salet2018: Salet, Ahmed, Bos et al. (2018) Design of a 3D printed concrete bridge by testing, Virtual and Physical Prototyping (abstract via Elicit)
• build:src-thau-1205: Tiny House Alliance USA (1 May 2026) How ICC/MBI 1205 will raise the price of housing (advocacy; quotes 1205 plant-certification text; reports Virginia and Utah adoption)
• build:src-vasilic2025: Vasilic (2025) Standardization Aspects of Concrete 3D Printing, RILEM Technical Letters (abstract page read)
• build:src-veev-tc: Prefab home builder Veev reportedly shutting down after reaching unicorn status last year (TechCrunch, 2023-11-27)
• care:src-anymal-spec: ANYbotics ANYmal product page
• care:src-astrobee-tfr: Alexandrov et al., Astrobee: Free-Flying Robots for the International Space Station (IEEE Trans. Field Robotics 2026, NTRS 20260001396)
• care:src-chloe: Salminen & Milenkovic, 'Her Name Was Chloe and She Delivers Stuff to Your Room' (NordiCHI 2020)
• care:src-dp: Chi et al., Diffusion Policy: Visuomotor Policy Learning via Action Diffusion (RSS 2023 / IJRR 2024)
• care:src-ecoppia-ketura: World's first self-cleaning solar park in the Arava Valley (Times of Israel, 2014-03-25)
• care:src-ecoppia-site: Ecoppia home page (fleet counters)
• care:src-gecko-trr: Gecko Robotics reaches unicorn status with latest funding (The Robot Report, 2025-06-12)
• care:src-henn-futurism: Japan's Robot Hotel Just Laid Off Half Its Robots (Futurism, 2019-01-15)
• care:src-ilse-joule: Ilse et al., Techno-Economic Assessment of Soiling Losses and Mitigation Strategies for Solar Power Generation, Joule 2019
• care:src-irobot-md: Roomba maker files for Chapter 11 bankruptcy (Manufacturing Dive, 2025-12)
• care:src-lbm: TRI LBM Team, A Careful Examination of Large Behavior Models for Multitask Dexterous Manipulation (arXiv 2507.05331; Science Robotics 2026)
• care:src-mss-wiki: Mobile Servicing System (Wikipedia; secondary summary of NASA/CSA data)
• care:src-neo-newatlas: 1X's first robot housekeeper is available to pre-order in the US (New Atlas, 2025-10-29)
• care:src-owens-iss: Owens & de Weck, International Space Station Operational Experience and its Impacts on Future Mission Supportability (ICES-2018-198, NTRS 20200002558)
• care:src-ozmo-trr: Skyline Robotics deploys Ozmo window cleaning robot in New York City (The Robot Report, 2024-08-28)
• care:src-spot-slashdot: Slashdot (16 Jun 2020) Boston Dynamics starts selling its Spot robot for $74,500 (Spot Explorer kit contents)
• care:src-spot-spec: Boston Dynamics Spot product page
• care:src-stretch-spectrum: Stretch 3 Brings Us Closer to Realistic Home Robots (IEEE Spectrum, 2024-02-15)
• care:src-yt-neo-wsj: I Tried the First Humanoid Home Robot. It Got Weird. (WSJ, YouTube)
• energy:src-ambri-pvmag2024: pv magazine USA (2024-08-05) Liquid metal battery storage specialist Ambri emerges from restructuring
• energy:src-bnef2025: BloombergNEF (2025) Lithium-ion battery pack prices fall to $108 per kilowatt-hour
• energy:src-bowery-yahoo: Yahoo Finance / Agriculture Dive (2024-11-07) Celebrity-backed indoor farming company Bowery closes
• energy:src-catl-newatlas2026: New Atlas (2026-04-28) Production EV packing sub-zero-operation sodium-ion batteries on its way (CATL Naxtra)
• energy:src-ceccanti2026: Ceccanti, Bischi, Antonelli, Baccioli (2026) Can sunlight replace LEDs in vertical farms? arXiv:2607.24385
• energy:src-currentos: Current/OS Foundation website (DC microgrid set of rules, reference projects)
• energy:src-form-latitude2025: Latitude Media (2025-10-29) Form's first 100-hour batteries are hitting the grid
• energy:src-fraunhofer2019: Fraunhofer ISE press release (2019) Agrophotovoltaics: high harvesting yield in hot summer of 2018 (Heggelbach)
• energy:src-fregosi2015: Fregosi et al. (2015) A comparative study of DC and AC microgrids in commercial buildings across different climates and operating profiles, IEEE ICDCM
• energy:src-fsp-wiki: Wikipedia: Fission Surface Power (requirements, Phase 1 awards)
• energy:src-iea-pvps2026: IEA PVPS Snapshot of Global PV Markets 2026 (report page: ~698 GW new PV in 2025; cumulative nearly 3 TW, up from 2.3 TW end-2024)
• energy:src-krusty2020: Poston, Gibson, Sanchez, McClure (2020) Results of the KRUSTY Nuclear System Test, Nuclear Technology
• energy:src-lazard2025-pdf: Lazard (June 2025) Levelized Cost of Energy+ (LCOE+), Levelized Cost of Storage Version 10.0, p.19 'LCOS Comparison ($/MWh)'
• energy:src-leary2020: Leary, Piggott, Howell (2020) The Scoraig wind trials: in situ power performance measurements of locally manufactured small wind turbines, Wind Engineering 45(3)
• energy:src-next2sun: Next2Sun vertical bifacial agri-PV product site
• energy:src-nfpa855-snippet: Central Hudson (3 Mar 2020) Energy Storage Testing, Codes and Standards - NFPA 855 (2020) dwelling limits 40 kWh utility closets / 80 kWh garages and outdoors (search snippet only, not read)
• energy:src-nicolello-ess: Nicolello (WSAFC, c. 2018) Fire Safety for New Battery Technologies (slides; 2018 IFC thresholds and 2018 IRC RB171-16 battery rules; PDF text read)
• energy:src-pea-wiki: Wikipedia: Princess Elisabeth Antarctica
• energy:src-plenty-elevenflo: ElevenFlo (2026-01-03) Plenty Unlimited: $1.9B-peak vertical farmer reorganizes in 53 days
• energy:src-pne-essnews2025: ESS News (2025-11-25) Finland's Polar Night to build 250 MWh sand battery for district heating provider
• energy:src-pne-ie2026: Interesting Engineering (2026-04-28) The world's largest sand battery just survived its worst-case winter
• energy:src-reher2024: Reher et al. (2024) Potential of sugar beet and wheat production in vertical bifacial, tracked, or elevated agrivoltaic systems in Belgium, Applied Energy 359:122679
• energy:src-rondo-pvmag2025: pv magazine (2025-10-21) Rondo Energy's 100 MWh heat battery powered by 20 MW of onsite solar
• energy:src-sciam2025: Scientific American (2025-08-05) NASA's plan for a nuclear reactor on the Moon could be a lunar land grab
• energy:src-sepulveda2021: Sepulveda, Jenkins, Edington, Mallapragada, Lester (2021) The design space for long-duration energy storage in decarbonized power systems, Nature Energy 6:506-516
• energy:src-showa2020: Study of a clean microgrid for the Japanese Antarctica Showa Base (2020) E3S Web Conf. 160:01001 (Elicit record)
• energy:src-sodium-wiki: Wikipedia: Sodium-ion battery (CATL, Faradion 0 V shipping, Natron closure, IRENA cost projection)
• energy:src-v3-concept: Anyground ORIJO v3 concept.json (internal; M08 IRORI spine, DENCHI-B, HATA-4, BYOBU-PV, MIURA-W)
• energy:src-vf-wiki: Wikipedia: Vertical farming (Jones Food Company administration 2025; LED efficiency)
• grow:src-agdive-bowery: Agriculture Dive (7 Nov 2024) Celebrity-backed indoor farming company Bowery closes, lays off 187 workers
• grow:src-agfunder-80acres: AgFunderNews (3 Aug 2026) Indoor ag heavyweight 80 Acres Farms to cease operations
• grow:src-agfunder-plenty: AgFunderNews (27 Mar 2025) Plenty, an autopsy: It promised the moon, but plants don't run on pitch decks
• grow:src-anyground-v3: Anyground v3 concept.json (M07 HATA loom, HATA-4, UKI-P, KAMADO-B, YOSHI-6G, M09 TSURU)
• grow:src-arad2020: Arad et al. (2020) Development of a sweet pepper harvesting robot (SWEEPER), J. Field Robotics
• grow:src-bac2014: Bac, van Henten, Hemming & Edan (2014) Harvesting Robots for High-value Crops: State-of-the-art Review and Challenges Ahead, J. Field Robotics 31(6)
• grow:src-elevenflo-plenty: ElevenFlo (Jan 2026) Plenty Unlimited reorganizes in 53 days: petition 23 Mar 2025, plan confirmed 15 May 2025, emerged 29 May 2025 with Richmond strawberry farm (secondary source)
• grow:src-fda-preharvest: US FDA - FSMA Final Rule on Pre-Harvest Agricultural Water (effective 5 Jul 2024)
• grow:src-fda-psr: US FDA - FSMA Final Rule on Produce Safety (current as of 13 Nov 2024)
• grow:src-fruitnet-spread2017: Fruitnet (23 May 2017) Largest automated farm to open in Japan (Spread Techno Farm Keihanna)
• grow:src-gai-carbon-g2: Global Ag Tech Initiative (18 Feb 2025) Carbon Robotics introduces G2 product line
• grow:src-hemming2019: Hemming et al. (2019) Remote Control of Greenhouse Vegetable Production with Artificial Intelligence - Greenhouse Climate, Irrigation, and Crop Production, Sensors 19(8):1807
• grow:src-hortidaily-infiniteacres: Hortidaily (Sep 2026) Infinite Acres declared bankrupt in the Netherlands
• grow:src-johnson2021: Johnson, ..., Wheeler (2021) Supplemental Food Production With Plants: A Review of NASA Research, Front. Astron. Space Sci. 8:734343
• grow:src-kusuma2020: Kusuma, Pattison & Bugbee (2020) From physics to fixtures to food: current and potential LED efficacy, Hortic. Res. 7:56
• grow:src-riggio2019: Riggio, Jones, Gibson (2019) Risk of Human Pathogen Internalization in Leafy Vegetables During Lab-Scale Hydroponic Cultivation, Horticulturae 5(1):25 (abstract via Elicit; MDPI 403)
• grow:src-robotreport-ironox: The Robot Report (2 Nov 2022) Reports of layoffs at Iron Ox
• grow:src-saga-site: Saga Robotics official site (Thorvald UV-C)
• grow:src-tc-monarch: TechCrunch (19 Nov 2025) Monarch Tractor preps for layoffs and warns employees it may shut down
• grow:src-vfd-growcer-2026: Vertical Farm Daily (8 Jun 2026) Rising food prices drives demand for container farming in Canada (Growcer)
• grow:src-vfd-naurvik: Vertical Farm Daily (7 Jul 2025) Inside an arctic greenhouse (Naurvik, Gjoa Haven, Nunavut)
• grow:src-yan2019: Yan, He, Niu et al. (2019) Growth, nutritional quality, and energy use efficiency of hydroponic lettuce as influenced by DLI, HortScience
• grow:src-zabel2020: Zabel et al. (2020) Biomass Production of the EDEN ISS Space Greenhouse in Antarctica During the 2018 Experiment Phase, Front. Plant Sci. 11:656
• loop:src-anyground-v3: Anyground v3 concept.json (mechanism M06 MOTTAINAI loop; modules KAWAYA-U, KAMADO-B, HATA-4, UKI-P)
• loop:src-barta-lmlstp: Barta D, Henninger D, Edeen M, Lewis J - Closure of Regenerative Life Support Systems: Results of the Lunar-Mars Life Support Test Project (NTRS 20080026014)
• loop:src-beler-baykal-2011: Beler-Baykal B, Allar AD, Bayram S - Nitrogen recovery from source-separated human urine using clinoptilolite and preliminary results of its use as fertilizer, Water Sci. Technol. (2011), abstract via OpenAlex
• loop:src-bvad-2022: Ewert, Chen, Powell (eds.) - Life Support Baseline Values and Assumptions Document, NASA/TP-2015-218570/REV2 (Feb 2022), Tables 3-31, 4-91, 4-92 and section 4.14
• loop:src-cfr-503: 40 CFR Part 503 Appendix B - Pathogen treatment processes (PFRP composting time-temperature)
• loop:src-eawag-aurin: Eawag - Aurin - Duenger aus Urin (project page; Google snippet: provisional approval since Feb 2016, approved by the Swiss Federal Office for Agriculture for edible plants since 2018)
• loop:src-ebel-2025: Ebel RH, Eberly JO, Grimberg BI, Seipel TF et al. - Household-scale anaerobic digestion of food waste - a community case study from Bozeman, Montana, Front. Sustain. Food Syst. (2025), abstract via OpenAlex
• loop:src-elsacker-2019: Elsacker E, Vandelook S, Brancart J, Peeters E et al. - Mechanical, physical and chemical characterisation of mycelium-based composites with different types of lignocellulosic substrates, PLoS ONE 14:e0213954 (2019), abstract via OpenAlex
• loop:src-fu-2021-biorxiv: Fu Y-M et al. - Establishment of a closed artificial ecosystem to ensure human long-term survival on the moon (Lunar Palace 365), bioRxiv 2021
• loop:src-fumasoli-2016: Fumasoli A, Etter B, Sterkele B, Morgenroth E, Udert KM - Operating a pilot-scale nitrification/distillation plant for complete nutrient recovery from urine, Water Sci. Technol. 73(1):215-222 (2016), PubMed abstract
• loop:src-garcia-gragera-2021: Garcia-Gragera D, Arnau C, Peiro E et al. - Integration of Nitrifying, Photosynthetic and Animal Compartments at the MELiSSA Pilot Plant, Front. Astron. Space Sci. 8:750616 (2021)
• loop:src-gitelson-198990030-6): Gitelson II, Terskov IA, Kovrov BG, Lisovskii GM et al. - Long-term experiments on man's stay in biological life-support system, Adv. Space Res. 9(8):65-71 (1989), PubMed abstract
• loop:src-gligorescu-2020: Gligorescu A, Fischer CH, Larsen P, Noergaard J et al. - Production and Optimization of Hermetia illucens Larvae Reared on Food Waste and Utilized as Feed Ingredient, Sustainability 12:9864 (2020), abstract via Semantic Scholar
• loop:src-insectinstitute-777m: Insect Institute - $777 million invested in five failed insect farming companies: what happened
• loop:src-jeffery-2017: Jeffery S et al. - Biochar boosts tropical but not temperate crop yields, Environ. Res. Lett. (2017), abstract via OpenAlex
• loop:src-nitta-2010: CEEF: Closed Ecology Experiment Facilities (Institute for Environmental Sciences, Aomori), 2010 overview (Tako Y, Arai R, Tsuga S et al.), abstract via OpenAlex
• loop:src-popsci-aurin-2025: Popular Science - Swiss startup turns urine into plant fertilizer (8 Dec 2025)
• loop:src-porterfield-2025: Porterfield DM, Tulodziecki D, Wheeler R - Critical investments in bioregenerative life support systems for bioastronautics and sustainable lunar exploration, npj Microgravity (2025), abstract via OpenAlex
• loop:src-rodriguez-2025: Rodriguez EE, Dickman RA, Kennedy BC, Aga DS - Comparative Exposure Assessment of Crops Grown by Urine-Derived Fertilizer and Crops Irrigated with Reclaimed Water, Environ. Sci. Technol. (2025), abstract via OpenAlex
• loop:src-sam-b2-2026: A Study of Super-Dwarf Pea for Bioregenerative Life Support at the Space Analog for the Moon and Mars (Biosphere 2 Test Module, 2026), abstract via OpenAlex
• loop:src-schroeder-2026: Schroeder C, Haefner F, Krause A et al. - Recycling fertilizers from resource-oriented sanitation systems: product-level characterization and meta-analysis of agronomic performance, Front. Environ. Sci. (2026), abstract via OpenAlex
• loop:src-severinghaus-1994: Severinghaus JP, Broecker WS, Dempster WF, MacCallum T et al. - Oxygen loss in Biosphere 2, Eos 75 (1994), abstract via OpenAlex
• loop:src-wheeler-199600880-N): Wheeler RM, Mackowiak CL, Stutte GW, Sager JC et al. - NASA's Biomass Production Chamber: a testbed for bioregenerative life support studies, Adv. Space Res. 18(4-5):215-224 (1996), PubMed abstract
• loop:src-wilsenach-2007: Wilsenach JA, Schuurbiers CAH, van Loosdrecht MCM - Phosphate and potassium recovery from source separated urine through struvite precipitation, Water Research 41(2):458-466 (2007), PubMed abstract
• mind:src-balaji2016: Balaji, Bhattacharya, Fierro et al. (2016) Brick: Towards a Unified Metadata Schema for Buildings, ACM BuildSys
• mind:src-bas-halley: British Antarctic Survey: Halley VI Research Station facility page
• mind:src-dm2016: Google DeepMind blog (20 Jul 2016) DeepMind AI reduces Google data centre cooling bill by 40%
• mind:src-dm2018: Google DeepMind blog (17 Aug 2018) Safety-first AI for autonomous data centre cooling and industrial control
• mind:src-ec-machinery: European Commission - Machinery (Regulation (EU) 2023/1230 applies on a mandatory basis from 20 January 2027; AI-powered safety functions; cyber-safety)
• mind:src-freightfarms: HortiDaily (5 May 2025) Freight Farms' bankruptcy filing reveals $7M in liabilities
• mind:src-freightfarms-ch7: Vertical Farm Daily (1 May 2025) Freight Farms files for Chapter 7 bankruptcy, leaving growers disconnected (headline)
• mind:src-gateway-ifl: IFLScience (24 Mar 2026) NASA pauses Lunar Gateway, announces ~$20B moon base plan
• mind:src-hemming2020: Hemming, de Zwart, Elings et al. (2020) Cherry Tomato Production in Intelligent Greenhouses - Sensors and AI for Control, Sensors 20(22):6430
• mind:src-imo-mass: IMO press briefing (22 May 2026) IMO adopts first global Code for autonomous ships (MSC 111)
• mind:src-isa62443: ISA: ISA/IEC 62443 Series of Standards (overview page)
• mind:src-iso10218-2025: Control Engineering (19 Feb 2025) Revised industrial robot safety standards now available (ISO 10218-1/-2:2025; snippet; page 403) and A3/Robot Report coverage noting ISO/TS 15066 content absorbed into ISO 10218-2:2025
• mind:src-iso13482: ISO 13482:2014 Robots and robotic devices - Safety requirements for personal care robots (catalogue page; 403 on fetch)
• mind:src-jci2023: Cybersecurity Dive (28 Sep 2023) Johnson Controls hit by 'severe' cyberattack
• mind:src-jonah2026: Jonah, Moses et al. (2026) LLMs for Agentic Home Energy Management, arXiv:2607.04569
• mind:src-khabbazi2025: Khabbazi, Pergantis, Reyes Premer et al. (2025) Lessons learned from field demonstrations of MPC and RL for residential and commercial HVAC: A review, Applied Energy 399:126459 (arXiv:2503.05022)
• mind:src-liu2022: Liu, Balaji, Gao et al. (2022) Safe HVAC Control via Batch Reinforcement Learning, ICCPS 2022
• mind:src-luo2022: Luo, Paduraru, Voicu et al. (2022) Controlling Commercial Cooling Systems Using Reinforcement Learning, arXiv:2211.07357 (DeepMind, Google, Trane)
• mind:src-melville2014: Melville, Stillinger, Gerrard, Weatherwax (2014) Sustainable energy at the 100 W level for scientific sites on the Antarctic Plateau: lessons learned from the PENGUIn-AGO project, Rev. Sci. Instrum. 85:045117, doi:10.1063/1.4871555
• mind:src-mulayim2025: Mulayim, Pergantis, Reyes Premer, Chen, Qu, Kircher, Berges (2025) Comparative Field Deployment of RL and MPC for Residential HVAC, arXiv:2510.01475
• mind:src-openrmf-news: The Robot Report (8 Jan 2024) New version of Open-RMF hits ROS 2
• mind:src-plato2009: Ashley, Allen et al. (2009) The PLATO observatory: robotic astronomy from the Antarctic plateau, Proc. IAU (6-tonne self-contained, deployed Dome A Jan 2008)
• mind:src-waymo-futurism: Futurism (23 Feb 2026) Here's how many remote operators Waymo has per self-driving taxi (citing Waymo letter to Sen. Markey and Waymo ops blog)
• syn:chen-2026-mof-battery: Chen W-C et al. 2026 Symbiotic heat sharing in MOF-based atmospheric water generators for battery thermal management and sorbent regeneration, Nat Commun (abstract via OpenAlex)
• syn:hussien-2017-wef: Hussien WA, Memon FA, Savic DA 2017 An integrated model to evaluate water-energy-food nexus at a household scale, Environ Model Softw (abstract via OpenAlex)
• syn:ifr-2026: IFR press release (24 Sep 2026) Five million robots now operate in factories globally; >600,000 installed in 2025 (+11%)
• syn:jetson-2024: NVIDIA blog (17 Dec 2024) Jetson Orin Nano Super: 67 INT8 TOPS for USD 249 (was 499)
• syn:li-2020-pv-awh: Li R, Shi Y, Wu M, Hong S, Wang P 2020 Photovoltaic panel cooling by atmospheric water sorption-evaporation cycle, Nat Sustain 3:636-643 (abstract partially read via Elicit)
• syn:lidar-ct-2025: CleanTechnica (20 Mar 2025) Lidar's wicked cost drop, quoting China Daily 6 Mar 2025 (30,000 -> 1,000 yuan)
• syn:opis-2023-10: pv magazine / OPIS (27 Oct 2023) Solar module prices dive to record low; CMM USD 0.140/W
• syn:opis-2026-10: pv magazine / OPIS (16 Oct 2026) China TOPCon solar cell prices edge lower after Golden Week; China Module Marker USD 0.107/W
• syn:so101-bom: TheRobotStudio SO-ARM100/SO-101 README: follower-arm parts USD 121.94; leader+follower USD 229.88
• syn:unitree-r1: Unitree R1 product page (accessed 2026-10-29): R1 AIR USD 4,900, R1 USD 5,900, ~29 kg, 26 DoF
• v3:concept: Anyground ORIJO v3 concept (research/v3/concept.json)
• water:src-almassad-2022: Almassad et al. 2022 - Environmentally adaptive MOF-based device enables continuous self-optimizing AWH, Nat Commun
• water:src-ann-japan-purifiers: Asia News Network / Yomiuri (2 Sep 2026) Japan government to boost water supply capacity during disasters
• water:src-anyground-v3: Anyground v3 concept.json (internal: MIZU-1, KUMO-12, M05, M06)
• water:src-ca-dpr-page: California State Water Resources Control Board - Direct Potable Reuse (adoption 19 Dec 2023; effective 1 Oct 2024; OAL approval 6 Aug 2024)
• water:src-ca-dpr-text: SWRCB SBDDW-23-001 Direct Potable Reuse final regulation text, CCR Title 22 sections 64669.00-64669.130 (June 2024, 63 pp.; read sections 64669.35 and 64669.45)
• water:src-dailygalaxy-wota: The Daily Galaxy 2026 - Japan built a home machine that purifies and reuses almost all household water (WOTA Unit)
• water:src-eu-reuse-2020-741: European Commission - Water reuse (Regulation (EU) 2020/741, applies from 26 June 2023, agricultural irrigation only)
• water:src-fathieh-2018: Fathieh et al. 2018 - Practical water production from desert air, Sci Adv
• water:src-hanikel-2019: Hanikel et al. 2019 - Rapid cycling and exceptional yield in a MOF water harvester, ACS Cent Sci
• water:src-hydraloop-site: Hydraloop official site (H300/H600 specs, certifications)
• water:src-kim-2017: Kim et al. 2017 - Water harvesting from air with metal-organic frameworks powered by natural sunlight, Science
• water:src-klemm-2012: Klemm et al. 2012 - Fog as a fresh-water resource: overview and perspectives, Ambio 41:221
• water:src-li-2024-ohw: Li Q. et al. 2024 - AWH system based on the Optimal Harvesting Window design, Science Bulletin
• water:src-lord-2021: Lord et al. 2021 - Global potential for harvesting drinking water from air using solar energy, Nature 598:611
• water:src-mit-pved-2024: MIT News 2024 - Solar-powered desalination system requires no extra batteries (Bessette, Pratt, Winter, Nature Water)
• water:src-nasa-98: NASA Achieves Water Recovery Milestone on International Space Station (2023, updated 2026)
• water:src-nguyen-2017: Nguyen et al. 2017 - Controlling bacterial pathogens in water for reuse in the Blue Diversion Autarky Toilet, Front Environ Sci
• water:src-ntrs-iss-2026: Williamson, Wilson, Luong, Chen - Status of ISS Water Management and Recovery, ICES-2026-425
• water:src-pat-polito: CN111278532B - Method for producing water from air based on low-temperature heat (Politecnico di Torino)
• water:src-reynaert-2023: Reynaert et al. 2023 - Predicting microbial water quality in on-site water reuse systems with online sensors, Water Research
• water:src-sjv-allensworth: SJV Water 2024 - Allensworth residents feel abandoned by company that makes water out of thin air
• water:src-uc-deathvalley: University of California News 2023 - Hand-held water harvester powered by sunlight
• water:src-watergen-site: Watergen official site
• water:src-who-potable-reuse-2017: WHO (2017) Potable reuse: guidance for producing safe drinking-water, ISBN 978-92-4-151277-0 (catalogue page)
• water:src-wiki-desal: Wikipedia - Desalination (energy table citing IWA and Panagopoulos 2020)
• water:src-wota-site: WOTA Corp. official site (WOSH, WOTA BOX, WOTA Unit)
• water:src-zeeman-2011: Zeeman & Kujawa-Roeleveld 2011 - Resource recovery from source separated domestic waste(water) streams; full scale results, WST