In-Situ Resource Utilization (ISRU)
ISRU is the practice of using materials available at a destination body — regolith, water ice, volatiles — instead of hauling them from Earth. Large bases, sustained crewed presence, and interplanetary supply chains are only feasible if the mass ratio stops being dominated by Earth-launched feedstock. Swarm coordination is the missing glue.
Why swarms
Monolithic bespoke rovers have dominated ISRU concepts for two decades: Curiosity-class excavators, large 3D printers, specialized refineries. They are efficient per unit but poorly matched to the actual operating regime:
- Throughput scales with parallelism, not individual size. Thousands of millimeter- to centimeter-scale micro agents processing regolith in parallel outperform one large machine at comparable mass.
- Resilience requires multiplicity. A single excavator that fails halts the base. A swarm with 10% attrition still produces 90% throughput.
- Adaptability requires decentralization. Regolith composition, slope, and terrain vary across meters. Local policies outperform global schedules when the physical environment is not uniform.
See the HMA paper for a worked example at the scale of a lunar base pad.
Macro–micro coordination
ISRU at base scale demands two tiers that must stay in lockstep:
- Micro agents. Millimeter- to centimeter-scale. Excavation, compaction, sintering precursors. Energy-constrained; short ranges; prone to individual failure; rely on local rules.
- Macro agents. Rovers, haulers, printers, charging stations. Kilogram- to tonne-scale. Longer ranges, larger energy budgets, structured schedules.
The Arboria HMA mechanism (hierarchical market auctions with energy-aware bids and depot buffering) couples the two. Buffers decouple macro / micro rate mismatches; energy-aware bids prevent depots from draining haulers into a brownout; relay rotation keeps single charging stations from becoming single points of failure.
Research questions
- Optimal fleet composition. Given a target production rate and energy budget, what is the optimal ratio of micro to macro agents? Currently set empirically — belongs in the benchmark suite.
- Depot geometry. How many depots, at what separation, for a base of size S? Depot count interacts nonlinearly with hauler count and fleet energy.
- Energy accounting honesty. Published mobility-energy figures understate thermodynamic cost by 50–100× because sintering is 1–5 MJ/kg for lunar regolith. The paper makes this scope explicit; more realistic energy modeling is an open direction.
- Wear propagation. A failure model that correlates with actual mechanical wear (not fail-stop at random rate) would change fleet-composition trade-offs materially.
Adjacent work
- Classical ISRU surveys: Sanders & Larson; IEEE Aerospace proceedings.
- Robotics constructors without swarms: NASA RASSOR, ispace Mission 2 platforms, MIT’s TERRA project.
- Market-based multi-robot coordination: Dias, Parker, and the MRTA taxonomy literature.
Arboria’s contribution is the coupling of all three: macro–micro fleets, market-based energy-aware scheduling, and published scale numbers at 10⁶+ agent mixes.