Adaptive UKF-Based Navigation for ISS Rendezvous and Proximity Operations

This work presents an adaptive Unscented Kalman Filter (UKF) navigation architecture for rendezvous and proximity operations (RPO) with the International Space Station (ISS). The filter dynamically adjusts process and measurement covariances using residual-based trust metrics, enabling resilient state estimation under thrust dispersion, initial injection offsets, and orbital perturbations including drag, J2, and solar radiation pressure. This Monte-Carlo campaign evaluates navigation robustness across varied maneuver-execution thresholds, revealing three distinct regimes of behavior: high-gain continuous correction, under-actuated sparse correction, and a mid-range operational corridor. Within this corridor, ∼100-400 maneuver events yield consistent 97-98% 3σ relative-navigation accuracy with ΔV tightly clustered at 0.324-0.329 km/s. The study identifies a scenario (~214 maneuvers) as the most realistic cold-gas-like operational configuration, balancing correction frequency, estimator stability, and fuel usage. These results demonstrate that adaptive UKF navigation can maintain performance across a wide control spectrum while exposing a tunable region suitable for autonomous ISS-class RPO.