Event-Based Star Tracking: Investigating Tracking Delay Effects on Accuracy Across Varying Signal Environments

Abstract

The safe conduct of space missions and operations depends heavily on space situational awareness (SSA). It includes tracking space objects, avoiding collisions, predicting celestial body movements, and determining attitude. The detection and tracking of stars in space, which act as anchor points for calculating spacecraft attitude, is a crucial component of SSA.

In contrast to traditional frame-based sensors, this research investigates a novel application of event-based optical sensors for the detection and tracking of stars. The promise of improved performance, including lower battery usage and increased temporal resolution, is made by event-based sensors. Event-based sensors react to changes in pixel intensity asynchronously, generating a continuous, sparse data stream with microsecond precision, unlike conventional cameras that provide periodic frames with redundant data. This method is especially useful in SSA applications when the Field of View (FOV) is largely static.

The optimization of star tracking by looking at the complete range of potential delays through which more events history can be accumulated, is the main issue this research addresses. In the past, tracking systems have been divided into two categories: online systems, which provide real-time forecasts but may be inaccurate, and offline systems, which offer greater accuracy but at the expense of system responsiveness. By defining the trade-off between system responsiveness and star location information, this research aims to close this gap and enhance accuracy while preserving functional real-time performance.

In the initial research phase, a comparative analysis between spatial and spatio-temporal star detection algorithms tailored for event-based space object data was conducted. Subsequently, the spatio-temporal algorithm underwent further development to function as a tracker, enabling an exploration of the impact of various temporal windows on system performance concerning signal-to-noise data rates and star speeds. Notably, while the detection algorithms focused on identifying the presence of stars in a single frame, the evolved tracking algorithm emphasized the temporal continuity of star positions over sequential frames. This distinction highlights a key shift from detecting isolated events to tracking the dynamic evolution of stars, offering valuable insights into the trade-off between system responsiveness and the precision of star position information. These findings contribute significantly to Space Situational Awareness (SSA), optimizing star tracking techniques and providing implications for real-world SSA applications, thus enhancing the efficacy of space missions and operations.

Date of Award	15 Dec 2023
Original language	American English
Supervisor	Baker Mohammad (Supervisor)