The Autonomous Lunar Logistics Fleet
A coordinated fleet architecture for cargo handling, excavation, inspection, construction, rescue, maintenance, and surface mobility before and between crew visits.
- Difficulty
- Advanced
- Evidence
- Modeled
- Readiness
- TRL 2 / 9
- Review
- Author reviewed
- Sources
- 4
- Reading time
- 12 min
Abstract
The lunar institution will be built and maintained mostly while no crew is present. I therefore define a fleet of complementary machines rather than one universal rover: scouts map terrain and change, haulers move standardized cargo, excavators handle regolith, manipulators assemble utilities, inspectors create condition evidence, and crew rovers provide protected mobility and rescue. Shared navigation, charging, tools, grasp points, cargo records, geofences, and work authority allow independently built machines to cooperate. Autonomy is allocated by consequence. Routine traverse and handling can proceed locally, while irreversible actions require verified conditions and explicit authority. Every task is represented as a state machine with safe pause, reversal, or recovery. The fleet's productivity is measured at the completed service interface, not in kilometers driven or soil moved. A cargo delivery is complete only when its identity, condition, location, connections, and commissioning evidence are trustworthy.
Machines occupy the institution between crews
A crewed sortie lasts days or weeks. Surface assets must survive months or years. During that interval, dust settles, cables move thermally, batteries age, structures experience impacts, and new cargo arrives. If every inspection, connector, reset, and unloading operation waits for people, the institution remains mission-bound.
I divide robotic territory by function.
| Fleet role | Typical work | Key design property |
|---|---|---|
| Scout | Mapping, route verification, hazard and change detection | Low mass, endurance, diverse sensing |
| Hauler | Cargo, regolith, waste, rescue tow | Standard payload restraint and reliable mobility |
| Excavator | Grading, trenching, berms, feedstock | Reaction management and wear tolerance |
| Manipulator | Assembly, connectors, tools, recovery | Force control and verified completion |
| Inspector | Imaging, metrology, thermal and electrical survey | Repeatable pose and traceable measurements |
| Crew rover | Pressurized mobility and refuge | Life support, redundancy, rescue and manual control |
No vehicle needs every capability. Common physical and information interfaces let the fleet combine them. A hauler positions a module. A manipulator connects it. An inspector independently verifies the load path and utilities.
Standardize the work before the robot
Robots struggle when equipment is designed for human intuition. Small unlabeled fasteners, flexible cables, hidden latches, ambiguous connector states, uncontrolled loose items, and force-sensitive operations make autonomous work brittle.
I require external equipment to include machine-readable identity, fiducials, standardized grasp points, captive fasteners, alignment guides, defined tool reactions, guarded connectors, drain and dust behavior, positive mechanical state, and test points. The interface should advertise not just geometry but allowable force, contamination limit, electrical state, and recovery method.
Cargo pallets need the same discipline. Each carries mass, center of gravity, handling points, keep-out zones, thermal limits, hazards, destination, power state, required orientation, and commissioning plan. A vehicle should reject a lift if the observed article does not match the digital record.
Autonomy follows consequence
Communication delay to the Moon is short enough for supervision but not for every traction or manipulator control loop. Local systems must avoid hazards, regulate force, track energy, and stop safely. Earth operators can approve goals and interpret complex faults.
I use four authority bands:
- continuous local control for stability, traction, collision avoidance, and joint safety;
- autonomous execution for surveyed routes and reversible routine tasks;
- supervised autonomy for lifting, connection, excavation near assets, and configuration change;
- explicit human authorization for explosive devices, pressure-boundary changes, reactor controls, crew rescue commitment, and irreversible disposal.
The boundary can move only after evidence. Successful repetitions under varied conditions can expand autonomy. A novel or degraded state moves authority back toward deliberate review.
Every task has entry conditions, allowed transitions, completion evidence, timeout, safe pause, and recovery. A connector task is not complete because motor current peaked. It may require alignment image, insertion depth, latch indication, electrical isolation test, continuity, insulation, and communication handshake.
Mobility is an energy and terrain service
Range depends on grade, soil, wheel slip, temperature, payload, speed, navigation detours, and reserve. I screen mission energy as
The safe range is not the distance obtained by dividing nameplate battery energy by a flat-course test. The vehicle must preserve energy for an alternate route, communication outage, wheel degradation, cold survival, and recovery tow.
Charging points should publish connector, voltage, alignment, thermal, access, and reservation. Mobile batteries can rescue a stranded machine. A tow interface allows another vehicle to recover an immobilized unit without improvised rigging. Routes receive grades, obstacles, bearing, illumination, communication, traffic class, and current condition.
Dust changes mobility and maintenance. Wheels and tracks disturb it. Electrostatic charging and low gravity can keep particles mobile. Worksite plans separate clean interfaces from excavation traffic and require cleaning before docking or habitat approach.
Mixed fleets need traffic law
Geofences protect landing zones, reactors, science preserves, cables, unstable slopes, and crew areas. They must be versioned and distributed with integrity. A stale map can be worse than no map if a vehicle confidently enters a newly active worksite.
Right-of-way rules should be simple: crew rescue and hazard response first, crew vehicles next, vehicles carrying unstable or suspended loads next, ordinary cargo after that. Machines announce route intent, braking limits, localization quality, and degraded state. A local coordinator can schedule narrow corridors without becoming a single point that prevents emergency stop.
Give every machine the same terrain, asset, hazard, and configuration basis.
Coordinate mobility through energy-aware corridors and local services.
Make force, alignment, latching, and recovery states observable.
Records who may act and what proves the physical result.
Productivity ends at commissioning
An excavator that moves tonnes but delivers inconsistent feedstock can starve a process plant. A hauler that reaches a destination but damages a hidden connector has not delivered useful cargo. I measure end-to-end services:
- certified regolith delivered to a process hopper;
- cable route prepared and inspected;
- cargo transferred, connected, tested, and registered;
- degraded unit isolated, exchanged, and returned for analysis;
- crew route surveyed, marked, and maintained;
- landing zone cleared and declared ready.
Fleet availability also depends on maintenance. External modules, wheels, batteries, cameras, joints, and tools should be exchangeable by other machines. A disabled robot should become repair inventory, not a permanent obstacle.
The first uncrewed campaign should deliberately fail vehicles and interrupt tasks. The important demonstration is not perfect autonomy. It is a fleet that notices incomplete work, preserves the site, reassigns resources, recovers hardware, and eventually closes the service without asking a future crew to clean up hidden uncertainty.
Evidence register
Sources
- Lunar Surface TechnologyNASA
- Moon to Mars Architecture WorkshopsNASA
- Moon Base SystemsNASA
- LunaNet Interoperability SpecificationNASA, ESA, and JAXA
Unresolved work
Open problems raised here
Demonstrate uncrewed lunar cargo turnaround
Can a mixed robotic fleet inspect a landing zone, unload standardized cargo, verify it, transport it, connect utilities, commission the payload, and recover from an interrupted task without local crew?
Qualify shared robotic worksites
What traffic, perception, command, geofence, tool, and authority standards let machines from several providers work in the same lunar district without unsafe interference?
Continue the work
Challenge, correct, or extend this report.
Bring evidence, expose a weak assumption, or propose the next experiment. Confirmed work can become a revision to the report.
Authorship and provenance
A named, versioned record.
- Author
- Mingkai
- Rights holder
- Mingkai
- First published
- August 19, 2026
- Current revision
- v1.0 / 2026-08-19
Suggested citation: Mingkai. “The Autonomous Lunar Logistics Fleet.” Physical Extremes, 2026, https://mingkaibuilds.github.io/Frontiers/reports/deep-space/autonomous-lunar-logistics/
© 2026 Mingkai. Physical Extremes™. All rights reserved.