Housings, seals, connectors, compensation and buoyancy must survive the verified mission depth with margin and a test record.
CONCEPT MISSION / NOT AN ACTIVE RESPONSEREV A · 15 AUG 2026
THE ABYSS
IS NOT A
JOBSITE. YET.
A deep-ocean waste field is not a place to send the existing Rovalizer. It is a reason to design a new pressure-rated, regulator-led Swovee system that begins with evidence, earns every contact and can stop without making the site worse.
NODSSUM mapped 3,355 drums in 2025. In 2026, 20 dives directly observed several at more than 4,700 metres and found that some had spilled contents. This proposal keeps the historical estimate, mapped set and directly observed set separate.
00 / SOURCE-BOUND INCIDENT RECORDWHAT IS KNOWN.
WHAT IS KNOWN.
WHAT IS NOT.
The scenario is anchored to the Northeast Atlantic NODSSUM program. It is not the separate Southern California DDT disposal story.
HISTORICAL INVENTORY>200,000 drums
A July 2026 CNRS public summary gives a historical estimate of more than 200,000 drums dumped in the Northeast Atlantic between 1950 and 1990. Other program summaries use different scopes and totals.
Official source ↗MAPPED IN 20253,355 drumsThe first NODSSUM campaign mapped 3,355 drums, collected water and sediment, and established sectors for closer study.
Official source ↗DIRECTLY OBSERVED IN 202620 dives · >4,700 mNautile inspected several drums and their surroundings. CNRS documented advanced deterioration and content spillage from some—not all—of those observed.
Official source ↗STILL UNRESOLVEDLaboratory analysis continuesOn-site activity was above the expected local level but remained low. Sitewide inventory, transport, exposure and ecological consequences are not yet closed.
Official source ↗01 / FIRST DECISIONDO NOT SUBMERGE
DO NOT SUBMERGE
THE A02.
The land Rovalizer and Scoop contribute system ideas—not a shortcut around subsea engineering. Deep water changes every load path, interface, failure mode and recovery plan.
Ship, acoustic, inertial, Doppler and visual references must produce a repeatable subsea frame with explicit uncertainty.
Manipulator force, thruster wash, tool flow and tether motion can disturb sediment or damage an unknown object.
A disabled vehicle, released tool or contaminated filter needs a predefined, practiced route back to controlled deck space.
ADEFAULT
Observe + bound
Create a repeatable, georeferenced baseline; sample without spreading sediment; mark exclusion zones; and return on a fixed monitoring schedule. This is the least disturbing response when intervention risk exceeds demonstrated benefit.
BSITE-GATED
Isolate in place
Deploy a removable local shroud, reactive mat or engineered cap around a characterized source. Material choice, footprint and placement force are released only after current, ecology and contaminant-transport models pass review.
CAUTHORITY-GATED
Recover under control
Lift or overpack an object only when its condition, handling path, surface receiving facility and failure case are engineered together. Recovery is not the automatic ‘clean’ answer at abyssal depth.
02 / SYSTEM BOUNDARYKEEP WHAT
KEEP WHAT
TRANSFERS.
Swovee's useful inheritance is mobile perception, tool changing, material traceability, planning and verification. The ocean vehicle, however, must be new.
WHAT EXISTS AS A CONCEPTWHAT MUST BE DESIGNEDWHAT IT MUST NEVER IMPLY
A02CURRENT CONCEPT / DRY
Land Rovalizer
Build shoreside staging pads, washdown curbs, equipment foundations and non-nuclear service enclosures after ordinary terrestrial qualification.
S01SURFACE CONCEPT / DRY
Swovee Scoop
Its tool-changing, sample-chain and material-accounting ideas can inform a subsea system. A terrestrial Scoop could support controlled receiving and treatment work on land.
A01-DNEW MACHINE REQUIRED
Swovee Swim Surveyor
A pressure-rated, tethered survey vehicle would map bathymetry and objects, measure the local field, collect bounded samples and maintain a traceable target register.
R01-DNEW MACHINE REQUIRED
Swovee Swim Rovalizer
A work-class subsea tool carrier would place approved mats, shrouds, anchors, grout bags or monitoring nodes with force and turbidity feedback.
S02-DOPTIONAL NEW MODULE
Swovee Swim Sampler
A closed-loop, low-flow suction and filtration cassette could collect narrowly defined sediment samples or prepare an approved margin around a containment pilot.
D01NEW SUPPORT SYSTEM
Surface Dock
A vessel-based launch, recovery, power, communications, decontamination and waste-transfer system closes the mission loop and keeps humans out of the deep work zone.
02A / SWOVEE SWIM BRIEFINGTHE OCEAN NEEDS
THE OCEAN NEEDS
A NEW FAMILY.
Swovee Swim is the working name for a proposed survey, sampling and controlled-placement system. It is not the land Rovalizer in a pressure can, and it is not a product available today.
DESIGN THESISDo not begin by printing. Begin by earning the right to place one reversible thing.
The useful Swovee loop is scan → register → compare → authorize → place → inspect. Three-dimensional material placement becomes valuable only when a site-specific remedy requires geometry that ordinary ROV tooling cannot place as precisely or document as completely.
Survey first
A 6,000-m-class survey vehicle maps geometry, current, plume indicators and uncertainty without touching a target.
Place under control
A tethered work vehicle carries force-limited manipulators and a regulator-approved placement skid. The vessel supplies power and a physical recovery path.
Sample without dredging
A closed, low-flow sampler captures a bounded core or water sample and returns filters, tooling and slurry as controlled secondary waste.
Close the surface loop
A support vessel, heave-compensated launch system, wet lab, decontamination zone and sealed receiving path make the subsea machines operationally real.
THE ROCK QUESTION
DO NOT HARVEST BOULDERS BESIDE UNKNOWN DRUMS.
Moving local rock can strike a fragile container, disturb colonized habitat, change flow paths or resuspend contaminated sediment. If an approved capping pilot needs ballast, the defensible starting point is imported, washed, graded and traceable armor aggregate—or sealed rock-filled geocontainers—delivered through a low-energy placement chute or shrouded skid.
- 01 Scan and model the target, slope, current and ecological boundary.
- 02 Place a small removable flux-control shroud or geotextile layer around—not onto—the source.
- 03 Meter qualified ballast while monitoring turbidity, force and material loss.
- 04 Rescan the geometry and return through current cycles before scaling.
LEADING PHYSICAL CONCEPT / REV 0
REMOVABLE FLUX-CONTROL SHROUD.
A split shell surrounds a characterized source without loading its crown. It is not a permanent repository or a hermetic promise; it is an instrumented device intended to reduce and measure contaminant transport.
Vent during descent so the enclosure does not become a roughly 47 MPa pressure vessel at the current investigation depth. Pressure-bearing sensors and actuators remain independently qualified.
Interlocking segments land on a load-spreading perimeter skirt. Force and object-motion limits abort placement before the shroud touches the drum.
Replaceable radionuclide- or element-selective sorbents chosen after radiochemical characterization, controlled vents, porewater ports and a low-permeability or reactive mat create measurable flow paths.
Bagged graded aggregate cells sit on the exterior skirt. Acoustic fiducials, lift points and separable locks permit inspection, cartridge exchange or removal.
WHERE ADDITIVE PLACEMENT EARNS A ROLE
Swovee Swim v1 should deliver factory-qualified shroud panels, reactive mats and cartridges. A later tool may meter small quantities of qualified anti-washout grout into contained joints, gap fillers or sacrificial formwork. It should not free-print ordinary concrete around weakened drums, pump material 4.7 km from the surface, or treat local sediment as certified feedstock.
INITIAL MATERIAL INPUTS / PREFABRICATED SHELL · SEALED SORBENT CARTRIDGES · BAGGED GRADED BALLAST · CONTAINED JOINT MEDIA · RECOVERABLE REJECT BLADDERHONEST COMPETITIVE TEST
IS SWOVEE ACTUALLY BETTER?
Not by default. Existing subsea systems already reach these depths. Swovee must prove a narrower systems advantage, not claim that mobility, AI or printing automatically wins.
TASKCURRENT LEADERENGINEERING VERDICT
Reach 4,700–6,000 mExisting research and work-class vehicles
UlyX and Victor 6000 already publish relevant depth capability. Swovee has no subsea prototype or field heritage.
Heavy lift + recoveryExisting subsea salvage systems
Purpose-built cranes, lift frames and ROV manipulators remain the credible baseline for characterized objects.
Registered adaptive placementPotential Swovee advantage—unproven
A shared site model could connect scan, approved geometry, material identity, placement force and as-left verification in one traceable loop.
Make material at depthNo demonstrated Swovee advantage
At roughly 470 bar, imported and qualified media are safer to reason about than mining aggregate or mixing ordinary concrete on the seabed.
Decide whether to interveneScientists, authorities and responsible parties
AI can compare bounded options and surface conflicts. It cannot choose the remedy or authorize contact.
03 / OPERATIONAL SEQUENCESEVEN GATES
SEVEN GATES
TO ONE TOUCH.
Autonomy proposes paths and enforces limits. It does not decide that an object is safe to disturb or that a containment action is acceptable.
- 00
Authority before autonomy
Name the responsible agency, site owner, scientific lead and radiation-safety authority. Freeze the mission polygon, allowed contacts, sample custody, environmental stops and abort responsibility.
- 01
Reconstruct the field
Fuse historical records with multibeam sonar, imaging and repeat-pass navigation. Unknown objects remain anomalies—not barrels, hazards or leaks—until evidence supports classification.
- 02
Characterize without spreading
Approach from the current-safe direction, hold altitude, measure background and take the smallest defensible water or sediment sample. Turbidity, tether load or sensor excursions trigger retreat.
- 03
Choose the least harmful response
Compare no-touch monitoring, local isolation and controlled recovery using one risk register. The system recommends; named human authorities release the action.
- 04
Run one reversible pilot
Place a removable marker, shroud, mat or small engineered cap beside one characterized target. Record force, placement geometry, water quality and material loss continuously.
- 05
Verify over time
Revisit the pilot through current cycles and seasonal windows. Scale only when monitoring shows the intervention stayed in place and did not increase transport or ecological harm.
- 06
Close the waste chain
Recover tools, filters, sample containers and damaged consumables into sealed transfer. Publish an as-left map, exceptions, evidence package and next decision—not a victory claim.
MISSION INPUTS
Nothing enters the water without a passport.
- Authorized mission polygon + no-contact zones
- Bathymetry, current and seabed-bearing model
- Object register + evidence confidence
- Dose, chemistry and ecology baselines
- Approved intervention recipe + stop limits
- Vessel power, tether, launch and recovery
- Qualified mats, shrouds, ballast and repair media
- Sealed sample and secondary-waste containers
MISSION OUTPUTS
The product is evidence plus a bounded intervention.
- Versioned, evidence-traceable site model
- Georeferenced anomaly and target catalog
- Chain-of-custody samples + sensor record
- As-placed containment geometry
- Force, turbidity, dose and material-loss log
- Exceptions, aborts and human decisions
- Recovered secondary-waste manifest
- Monitoring result + scale / hold / stop decision
04 / READINESS GATESTHE VEHICLE EARNS
THE VEHICLE EARNS
THE OCEAN.
Passing a robotics demo is not permission to enter a contaminated offshore site. Each gate requires an owner, acceptance method and recorded evidence.
SITE TRUTH
Official inventory, jurisdiction and recent field evidence agree well enough to define a bounded mission.
SURVIVE
Pressure housings, connectors, tether, seals, power and recovery pass unmanned depth-margin and soak testing.
SEE
Navigation, sonar, imaging and field sensors meet detection and repeat-positioning requirements in representative turbidity.
TOUCH
Manipulator force limiting, tool exchange and abort behavior pass fragile-object trials without puncture, drag or uncontrolled lift.
PLACE
Containment media can be deployed without unacceptable plume, washout, rebound or loss under representative current and slope.
RECOVER
A disabled vehicle, severed tool or contaminated cassette has a practiced retrieval and receiving path.
PILOT
Regulators, scientists, operator and affected communities approve one monitored, reversible field test.
05 / BUILD COALITIONTHIS IS NOT
THIS IS NOT
A ROBOTICS TEAM.
It is a public-interest environmental operation with a robot inside it. The responsible organization must be built before the machine is sent offshore.
MISSION AUTHORITY
Regulator, responsible party, environmental counsel and incident command define what may be touched and who may stop work.
OCEAN SCIENCE
Physical oceanography, benthic ecology, marine chemistry, sediment transport and bathymetric survey lead the site model.
RADIOLOGICAL SAFETY
Health physics, radiochemistry, dosimetry, contamination control and licensed waste handling own the exposure case.
SUBSEA ENGINEERING
Work-class ROV, pressure systems, tether dynamics, launch and recovery, corrosion and offshore operations own vehicle survival.
ROBOTICS + CONTROLS
Perception, navigation, force control, supervised autonomy, fault management, simulation and evidence logging own machine behavior.
MATERIALS + GEOTECH
Marine grout, geotextile, reactive media, anchoring, seabed bearing and long-term durability own containment performance.
SALVAGE + WASTE PATH
Lifting, overpacking, vessel deck handling, decontamination, transport and licensed receiving close any recovery option.
PUBLIC INTEREST
Fisheries, coastal and Indigenous representatives, environmental justice specialists and independent reviewers shape acceptable outcomes.
VERIFY THE FIELD.
THEN INTERVENE.
The public record supports the disposal history, 2025 mapping count and 2026 direct observations. It does not yet select a remedy. The game therefore uses synthetic mission geometry and treats “monitor without touching” as a valid outcome.
COULD A ROBOT
MAKE THE NEXT
DECISION SAFER?
Share this proposal with ocean scientists, subsea engineers, radiation-safety teams and public agencies who can challenge the architecture before anyone builds it.