Constellation Design for Autonomous Cislunar Navigation and Target Tracking Using Relative Measurements

Ryan Menges, University of Colorado Boulder; Daniel Scheeres, University of Colorado Boulder; Derek Tuggle, The Charles Stark Draper Laboratory, Inc.
https://doi.org/10.64861/PNYM3647

Keywords: Autonomous cislunar navigation, Spacecraft-to-spacecraft tracking, Absolute and relative orbit determination, Higher-order filtering

Abstract:

High-accuracy estimation of an uncooperative cislunar target is difficult using only line-of-sight angles from an observer with an uncertain absolute state. This work proposes and numerically evaluates a three-spacecraft architecture comprising a passive target T, an observing agent A, and a cooperative pathfinder P. The agent measures only target azimuth and elevation and is assumed to have accurate attitude knowledge from an onboard star tracker, whose measurements are not modeled explicitly. The agent and pathfinder exchange range and range-rate measurements. The pathfinder neither observes, ranges, nor communicates with the target, and its absolute state is estimated rather than assumed known. All three absolute states are uncertain a priori and are estimated jointly using inter-spacecraft measurements and nonlinear Earth-Moon dynamics. The agent state enters both the T-A optical and P-A radiometric measurement models. The P-A measurements contain no direct target-state information but improve estimation of the observing agent’s trajectory. This improved observer estimate allows the same T-A angular measurements to be interpreted more accurately, improving the target absolute and target-relative estimates. Monte Carlo simulations compare the proposed architecture with an optical-only target-agent baseline and a direct target-radiometric case. The selected test case combines a near-rectilinear halo orbit and quasi-periodic orbits with multiple pathfinder phase offsets. Pathfinder assistance improves passive target estimation throughout the tested configuration set without requiring the pathfinder to measure the target or the target to cooperate. For the representative target-agent geometry with approximately 1000 km mean separation and δPT = 0.100, pathfinder assistance reduces the reported 3σ̄ scalar target absolute-state uncertainty by approximately an order of magnitude relative to the optical-only case: from 100 to 10 m in position and from 1 to 0.1 mm/s in velocity.

Date of Conference: September 15-18, 2026
Track: Cislunar SDA

Best Student Paper Award Winner 2026

View Paper