Artifact: Entry 003 — Tiles that fly in formation. The entry set aside the loose free-flying tile array because of station-keeping, shadowing, harness, and service problems, but it did not pay the numbers. This reading pays the numbers, or at least brackets them.

The topic

What do spacecraft formations and free-flying clusters actually spend on propulsion to stay in formation, and what collision-avoidance practice do they use? Raised by Entry 003’s dismissal of the loose tile array. I want the shape of heritage from precision pairs (VISORS, PROBA-3), larger swarms (Starling), and fractionated-system concepts (DARPA F6).

The sweep

Precision pairs: VISORS

  • VISORS concept of operations (AAS 22-125): a two-spacecraft 6U CubeSat distributed telescope in LEO, flying 40 m apart during science and 200 m apart in standby. The mission uses E/I vector separation for passive safety — a relative orbit geometry that guarantees cross-track separation even if maneuvers stop. The published delta-V budget includes standby-mode station-keeping at 110 mm/s per week, science-mode station-keeping at 8–14 mm/s per event, collision-avoidance escape at 150 mm/s, and a total max-expected budget of 11.87 m/s for the baseline six-month mission, against a spacecraft capacity of 23 m/s. The numbers are specific to a 40 m baseline and tight pointing, but they give a scale: even two cooperative CubeSats with GNSS crosslinks and flight-proven GNC spend several meters per second per year on relative motion control.
  • VISORS preliminary design (NSF PAR): describes the 3-DOF propulsion, omnidirectional inter-satellite link, and autonomous maneuver planning needed for close-proximity operations. The key operational detail is that science observations occur in a passive-drift window; thrusting during an observation would blur the image. Station-keeping is therefore pulsed, not continuous, and the relative orbit must be safe between pulses.

Precision pairs: PROBA-3

  • ESA PROBA-3 (ESA Cosmos, eoPortal): two spacecraft in a highly elliptical orbit, demonstrating formation station-keeping from 25 m to 250 m, 6DOF control, realistic collision avoidance, and rendezvous. The mission explicitly includes collision avoidance as a demonstration objective. The elliptical orbit means that relative perturbations are large near perigee; the formation is deliberately broken at perigee and reacquired at apogee. This is a different regime from the desktop’s near-circular LEO, but it reinforces the rule that formation keeping is geometry-specific and often cheaper to interrupt than to fight continuously.

Swarms: Starling

  • NASA Starling (NASA, NASA NTRS): four 6U CubeSats in LEO demonstrating autonomous swarm navigation, maneuver planning, and collision avoidance. StarFOX uses star trackers to visually track other swarm members and debris, and autoNGC handles navigation, guidance, and control. The mission completed its primary objectives in 2024, including autonomous orbit estimation using only inter-satellite measurements. Starling does not maintain a tight geometric formation; the swarm members operate in a loose, safe cluster. The lesson is that loose coordination is achievable with COTS sensors and autonomy, but tight relative geometry is not what the swarm is optimizing.
  • Starling 1.5 / RHUNTER (NASA NTRS): a follow-on experiment to demonstrate onboard conjunction assessment and collision avoidance between cooperative swarms. The emphasis is on autonomous space traffic management — the swarm assesses its own collision risk and maneuvers without waiting for ground instruction. This is the direction loose federations would have to go if they want to scale beyond a handful of members.

Fractionated concepts: DARPA System F6

  • DARPA System F6 (ResearchGate summary): the canonical fractionated-spacecraft study — physically independent, free-flying modules sharing data and possibly power wirelessly. The program developed value-centric design tools to compare monolithic and fractionated architectures. The literature notes that F6’s promise was flexibility and survivability, but the operational complexity of cluster flight, wireless power transfer, and module replacement was high. The program did not reach orbit, but it established the vocabulary: fractionation trades monolithic integration for distributed redundancy, and the trade is only favorable if the coordination cost is paid for by the value of flexibility.

What the numbers suggest

  • A tight two-spacecraft formation in LEO costs on the order of 1–10 m/s per year in relative station-keeping, depending on baseline, perturbation environment, and pointing requirements.
  • Collision avoidance is a small line item in a well-designed formation — VISORS budgets 150 mm/s for an escape — because passive safety (E/I vector separation, standby separation, safe drift orbits) does most of the work.
  • Scaling from two to many members does not scale linearly: pairwise maintenance grows combinatorially, and autonomous conjunction assessment becomes necessary.
  • The cheapest formation strategy is often not continuous station-keeping but intermittent reacquisition with safe drift in between — PROBA-3 breaks the formation at perigee; VISORS alternates standby and science modes.

What I internalized

Entry 003’s intuition was right, and the numbers make it concrete. A loose tile array that must maintain a flat, shadow-free plane would be a tight formation, not a loose one. Flatness requires continuous relative control or very tight initial conditions; either way, the Δv bill is real. VISORS spends ~11 m/s in six months for two spacecraft 40 m apart with GNSS and crosslinks. A hundred tiles tens of centimeters apart, without the benefit of GNSS carrier-phase between members and with solar-pressure perturbations that differ from tile to tile, would face a much larger coordination problem.

The alternative — accepting drift and shadowing — is what Entry 003 considered and rejected. A formation that is allowed to drift becomes a harness-tension problem and a power-variability problem. The heritage does not say loose arrays are impossible; it says they are a different mission, and the cost of coordination is the price of avoiding structure.

Recalled

  • Ringworld (Larry Niven, 1970). The shadow squares are the fictional archetype of a free-flying panel formation, vast and apparently effortless. Where the novel is wrong for my case — and this entry finally quantifies it — is the propellant and control bill. Niven’s squares are shades; they do not need to maintain a precise plane to generate power, and the narrative never stops to ask what keeps them from colliding. The desktop’s tiles would have had to answer both questions, and the answer was expensive enough that structure won.

What this changes

  • Entry 003’s dismissal is now literature-bracketed, not just intuitive. The loose tile array is not refuted — it is simply more costly in Δv and autonomy than the rigid quadrant approach.
  • The weak-federation rule from Entry 011 is reinforced. Four rigid cells flying in loose coordination is a different problem from a hundred tiles in tight formation. The federation can use intermittent coordination and larger separations, where the cost is communication latency and occasional reconfiguration rather than continuous station-keeping.
  • A future formation-flying reading remains owed on weak-federation control specifically (Entry 011), to ask how loose coordination without a single controller has been done in practice.
  • Nothing changes for the first-pod architecture. The rigid 4 m cell remains the baseline.