The feasibility of angles-only navigation (AON)-based close-in proximity operation is studied for application to upcoming small satellite-based uncooperative rendezvous missions. For the relative motion equations described by Yamanaka-Ankersen, the Square-Root Unscented Kalman Filter (SRUKF) is adapted for relative position estimations, considering low-volume/mass, lower-power, simple optical/infrared instruments together with Lidar measurements. Considering simplicity and achievements in engineering, the multi-pulse glideslope guidance law is utilized. Using a linear covariance technique, a complete set of analytical functions are deduced for closed-loop true dispersions and estimated dispersion analysis. Monte Carlo simulation proves the offset observation model proposed provides a good solution for the range-observability dilemma: the range estimation error decreases from an initial decameter-level to a final decimeter-level. Two proximity operation mission trajectories are designed: direct v-bar quasi-linear glide approach for robotic arm capture or net capture, and glide approach (i.e., circumnavigate--glide approach for capture with attitude requirement). Using a well-designed relative approaching guidance profile, the camera operational range can be extended. This enables the operation time of the Lidar to be shortened, or even replaced by an optimal/infrared camera. This is very helpful in saving mass and power for the chaser. The variable-structure SRUKF proposed leads to a more robust trajectory: true dispersion is improved by two orders in the 100m range, and more accurate covariance prediction (i.e., all sampled trajectories are inside the ellipse) when compared to the standard SRUKF. The analysis method proposed, which could raise analytical closed-loop linear covariance, is applicable for onboard maneuver planning and real-time closed-loop control error estimation. [ABSTRACT FROM AUTHOR]