Michael Squires, Leo Labs; Nathan Griffith, LeoLabs; James Rowland, LeoLabs; Matthew Stevenson, LeoLabs
Keywords: maneuvers, delta-v, characterization, low false alarm, detection
Abstract:
In the context of operational space missions, space situation awareness (SSA) is critical for operators in order to maintain a usable orbit and complete the mission as designed. Collisions or radio communications interference by other live missions may prevent successful completion. While collisions can occur with any resident space object, one key aspect of space situational awareness is monitoring the actions of other live space missions which entails both the monitoring of those mission’s nominal orbit and any orbit changes. Of these two elements, it is particularly challenging to monitor orbital changes induced by maneuvers. Detecting and characterizing any maneuvers that may be present in these live, potentially uncooperative, missions is thus of paramount importance to the operator.
In this paper we describe a series of configurable algorithms to detect and characterize maneuvers in an automated fashion, and a corresponding series of analyses supporting their successful employ.
The first element in our automated detection and characterization algorithms is the configurable maneuver detector. The configuration options for the detector pertain to the number of tracklets (sets of coherent radar measurements) present in a detection window and signal hypothesis sensitivity. The tracklet window size setting varies the detection latency from low, using the first radar measurements past the putative maneuver event, to a high latency with a low false positive detection rate which uses more radar data. In both cases the sensitivity precludes any appreciable rate of false negatives. The low latency single tracklet detection window is especially useful for high interest objects where an elevated false positive detection rate is acceptable. The low false positive rate higher latency modes work best for passively monitoring a large catalog of objects where frequent alerts are undesirable. Both modes are demonstrated in this paper.
The configurable maneuver characterizer models maneuvers as impulsive, with adjustable fit regularization and fit bounds. These parameters allow one to search the solution space most likely to contain the correct post-maneuver object state. Successive widening of the bounds and disabling of regularization parameters allows the system to fit more unlikely varieties of maneuvers. The output of the maneuver characterizer includes an improved estimate of maneuver time and estimated impulsive velocity vector. Metrics from the fit, including quality of fit and magnitude of the maneuver can be utilized to veto any false positive detection. Additionally, we show that incorporating the estimated maneuver in a true positive case, allows the state produced by the nominal orbit determination system to quickly converge to the true state post maneuver.
We present a series of analyses applying these algorithms to LeoLabs radar data. We show validation against third party information including International Laser Ranging Service (ILRS) ephemeris with maneuver truth. We also show the application of the maneuver detector on known non-maneuverable objects of various classes to demonstrate low false alarm rate and apply our algorithms to an operational set of non-cooperative high interest objects, highlighting likely maneuvers based upon appreciable changes to object Keplerian orbital elements. The maneuver detection and characterization system is shown to perform well in all the studied cases.
Date of Conference: September 19-22, 2023
Track: SDA Systems & Instrumentation