BIT-X1.3-GDACN-001 — Gravity-Drift-Assisted Corridor Navigation: A Typed Belief–Resource Viability Framework (F5 Closed with Scope)

This independent research report develops and closes the BIT-X1.3-GDACN-001 branch at the F5 “Closed with Scope” level. The work treats a navigation corridor not as a special region of physical space, but as the set of belief–resource states from which at least one causal, non-anticipative policy can keep all declared mission constraints satisfied under the specified disturbance, observation, actuator, delay, resource, and horizon contracts. It distinguishes the conservative geometric admissibility slice, G_geo = C_phys ⊖ Z_err, from the robust belief–resource viability kernel, K_H^I. Geometric containment of an uncertainty envelope does not by itself establish future controllability, observability, resource sufficiency, or mission viability. Known gravitational dynamics are placed in the nominal drift model, while model mismatch and unresolved perturbations enter the bounded-disturbance contract. Invariant manifolds and low-energy transfer structures may reduce corrective effort, but they do not independently certify safety, arrival, radiation exposure, navigation integrity, or control authority. The “dung beetle principle” is retained only as a biological analogy for passive celestial orientation. Its aerospace realization requires a typed observation contract covering visibility, signal-to-noise ratio, geometry, calibration, power, catalog integrity, and observability; a star cue alone does not determine the full translational state. The formal core includes monotonicity of the robust viability kernel under resource, uncertainty, constraint, and information order; a closed-form uncertainty envelope and maximum correction interval for a scalar unstable mode; directional covariance contraction under informative measurements; and explicit countermodels separating geometric containment from policy viability. The analytic core and F5 computational kernel checks passed within their declared finite-dimensional model scope. The accompanying reproducibility pack contains the locked specification, proof pack, Python kernel, a different-runtime supporting Node checker, frozen result receipts, audit ledger, dependency versions, licenses, and integrity hashes. Scope limitation: this record contains no mission telemetry replay, high-fidelity ephemeris validation, hardware-in-the-loop testing, operational safety case, proof-assistant closure, independent F7 verification, mission certification, or flight qualification. It does not establish mission-specific propellant savings and is not flight ready. Prior-art boundary: gravity assists, invariant manifolds, low-energy transfers, celestial sensing, state estimation, robust control, and viability theory are established fields. The contribution claimed here is a typed systems synthesis, separation of claim classes, scoped formal results, countermodels, and a reproducible F5 verification workflow—not a new propulsion mechanism or primitive astrodynamics/navigation law.