On a clear night, the Moon hangs 384,400 kilometers away — nine times farther than the geostationary arc where the world's most valuable satellites cluster. Beyond it lies cislunar space: the volume bounded roughly by the Moon's sphere of influence, a realm so vast it dwarfs every orbital regime humanity has ever managed. The problem is that almost none of it is watched. As the 2020s accelerate into a lunar renaissance — NASA's Artemis program, China's ILRS ambitions, a wave of commercial landers, and dozens of planned orbital platforms — cislunar space is rapidly becoming crowded. Yet the sensors, algorithms, and governance frameworks that have protected Earth-orbit operations for six decades are failing almost completely at these distances. Cislunar space domain awareness (SDA) is, by any engineering measure, the hardest tracking problem our species has ever attempted. AI-generated image 384,400 km Moon's mean orbital radius — 9.1× farther than GEO ~10,000× Cislunar volume vs. the GEO belt 25+ Planned CLPS lunar surface deliveries by 2028 50+ Potential cislunar satellites within a decade Mag 17.9 GBOSS detection limit after 2025 upgrade 3 Collision-avoidance maneuvers Chandrayaan-2 needed in 4 years among 6 orbiters The Geometry Problem: A Volume Nobody Told You About Start with raw numbers. Geostationary orbit sits at 42,164 kilometers from Earth's center — a thin shell roughly 75 kilometers deep where telecommunications satellites park. The Moon orbits at 384,400 kilometers, meaning its orbital radius is 9.1 times larger than GEO. Volume, however, scales as the cube of radius. That single ratio translates to a cislunar volume approximately 750 times larger than the sphere enclosing GEO — and when you include the full sphere of influence out to roughly the Moon's Hill sphere, the multiplier climbs to 1,000–10,000× depending on the regime you're monitoring. The practical consequence is immediate: a sensor network that could adequately cover GEO would be overwhelmed trying to cover cislunar. Earth-based telescopes observing GEO objects at ~36,000 km already push the limits of angular resolution for meter-class objects. At lunar distance, the same object would appear roughly 9× smaller in apparent size and roughly 81× dimmer (inverse-square law), demanding dramatically larger apertures and longer integration times. Geometry also creates structural blind spots. The lunar exclusion zone — the region within 10–35 degrees of the lunar limb — floods sensitive sensors with scattered moonlight, rendering ground-based optical systems essentially blind to objects transiting near the Moon. Similarly, solar exclusion zones of 30–50 degrees around the Sun create seasonal and orbital-geometry windows where large portions of cislunar space become unobservable from any single ground station. No single observatory, and no current network, covers the full cislunar volume at any given time. Even when objects are geometrically visible, the angular rates involved create additional challenges. A satellite in a Near-Rectilinear Halo Orbit (NRHO) sweeps through a complex three-dimensional path that changes apparent velocity dramatically depending on where it is in its orbit — near the lunar periapsis it moves rapidly; near apolune it nearly stalls. Designing a survey strategy that catches objects in all phases of such orbits requires careful scheduling that ground-based networks were never optimized for. Scale Reality Check If you compressed the GEO belt to the size of a basketball court, cislunar space would fill a volume the size of several city blocks. The Space Surveillance Network was designed for the basketball court. Nobody has seriously built infrastructure for the city blocks — until now. The Physics Problem: When the Math Stops Working AI-generated image Even if sensors could observe every corner of cislunar space, the computational tools used to track objects in low Earth orbit (LEO) and GEO would fail. The workhorse of Earth-orbit catalog management is the Two-Line Element set (TLE) — a compact representation of an object's orbital state based on Keplerian two-body dynamics. In LEO, TLEs degrade over hours to days due to atmospheric drag and perturbations, but the underlying Keplerian framework holds reasonably well. In cislunar space, TLEs are essentially useless from the moment of injection. The gravitational influences of Earth, the Moon, and the Sun interact simultaneously and non-linearly — the classic three-body problem that has no closed-form analytical solution. Objects on near-rectilinear halo orbits (NRHOs), Distant Retrograde Orbits (DROs), and low-energy transfer trajectories evolve in ways that two-body approximations cannot capture over any meaningful timeframe. The mathematical consequences are severe. In standard orbit determination, uncertainty propagation is treated as roughly Gaussian — a probability ellipsoid that grows predictably in time. In the chaotic dynamics of cislunar regimes, that assumption breaks down within hours. Research from Purdue University (2021) demonstrated that conventional Kalman filters and their derivatives lose track custody — meaning the true object state falls outside the filter's uncertainty estimate — within just 1.5 to 2 orbital revolutions for typical cislunar trajectories. After that point, you don't know where the object is; more critically, you don't know that you don't know. Non-Gaussian Uncertainty: The Real Killer When uncertainty distributions become non-Gaussian — banana-shaped, bimodal, or wrapped around unstable manifolds near libration points — standard conjunction analysis breaks down entirely. The probability of collision between two cislunar objects cannot be reliably calculated using the same methods that work for LEO satellites. This is not a software problem or a data problem; it is a fundamental feature of the dynamical environment. New approaches under active research include Gaussian mixture models (GMMs) , polynomial chaos expansions, and particle filter methods that can represent multi-modal uncertainty distributions. But these are computationally expensive and require observation cadences far higher than cislunar sensors currently provide. The feedback loop is vicious: poor tracking geometry means sparse observations, sparse observations mean rapid uncertainty growth, and rapid uncertainty growth means conjunction analysis becomes nearly meaningless. The three-body problem also creates what dynamicists call sensitive dependence on initial conditions : tiny errors in a spacecraft's state vector at injection grow exponentially over time. Two objects that appear to be on safe parallel trajectories today may diverge or converge in ways that standard propagators can't predict. This is not hypothetical — it is why every operational mission in cislunar space requires intensive, dedicated tracking support from its operators, resources that won't be available to passively catalog the growing population of defunct hardware. Cislunar Orbit Types and Their Tracking Challenges • NRHO (Near-Rectilinear Halo Orbit): NASA Gateway's planned home — highly elongated, passes close to the Moon at periapsis, extremely sensitive to perturbations, requires continuous station-keeping • DRO (Distant Retrograde Orbit): Naturally more stable than NRHO, but large and poorly covered by ground networks; used for long-duration storage or staging • Low-Energy Transfers: Exploit three-body dynamics for fuel efficiency; highly chaotic during transit, near-impossible to catalog with classical two-body methods • Libration Point Orbits (L1/L2/L4/L5): Quasi-stable, used for relay satellites and future SDA platforms; require continuous station-keeping, evolve through manifold structures • Lunar Frozen Orbits: Somewhat stable low lunar orbits tuned to lunar gravity harmonics; still subject to solar perturbations over months Sensors Today: Built for a Different War The United States Space Surveillance Network (SSN) — now operat