EBS-EKF: Accurate and High Frequency Event-based Star Tracking

1Kitware Inc. 2University of Dayton

We present EBS-EKF, an event-based star tracking algorithm that combines a novel centroiding technique with an extended Kalman filter and is validated using the first ground-truthed dataset of event streams from real stars.

(a) Our centroiding technique accounts for event camera behavior in low light, enabling more accurate tracking.

(b) We develop a data collection setup of a EVK4-HD event camera that is rigidly mounted and synchronized with a space-ready APS star tracker. The insets show APS (top) and event camera (bottom) pixels for 8 Cygni, a subgiant star in the constellation Cygnus.

(c) We demonstrate that our attitude estimates (red) are more accurate than existing methods (purple), and that we can operate above the 3 deg/sec cutoff of the APS star tracker, highlighting the utility of our method for high-frequency star tracking.

EBS-EKF Overview


Abstract

Event-based sensors (EBS) are a promising new technology for star tracking due to their low latency and power efficiency, but prior work has thus far been evaluated exclusively in simulation with simplified signal models. We propose a novel algorithm for event-based star tracking, grounded in an analysis of the EBS circuit and an extended Kalman filter (EKF). We quantitatively evaluate our method using real night sky data, comparing its results with those from a space-ready active-pixel sensor (APS) star tracker.

We demonstrate that our method is an order-of-magnitude more accurate than existing methods due to improved signal modeling and state estimation, while providing more frequent updates and greater motion tolerance than conventional APS trackers. We provide all code and the first dataset of events synchronized with APS solutions.

Results

Overview of related works. Color indicates high (green), medium (yellow) and low (red) performance. ā€œsā€ denotes seconds. Compared to previous works, ours is the only one to 1) be evaluated on night sky data that is open source, 2) incorporate 3D state estimation, 3) require absolute orientations updates only on initialization, and 4) have subpixel centroiding precision

Metadata and tracking results for our real world dataset. Metrics are reported as (across, about) mean squared difference in arcseconds. Colors rank methods with the first (green), second (yellow), or third (red) lowest across and about means. Across all of the data runs, EBS-EKF is a magnitude or more accurate than the next closest algorithm. The intensity-dependent offset usually improves the accuracy due to the higher centroiding precision.

Sample Data Recordings

Examples of recorded tracks from our dataset. From left to right, these correspond to "SmoothSine", "Velocity Sweep 1", "Multipose 1" in the above table, respectively.


Live Demo


A large advantage of the EBS-EKF is that it is lightweight and can be run on many different systems. Here is an example of EBS-EKF in operation, running in realtime on a Rasberry Pi with an update speed of 350Hz. Video is recorded at 5x speed for brevity. Code to run EBS-EKF is provided in the Gitlab

Pictured is our demo setup. A Prophesee EVK4-HD event camera and a Rasberry Pi are mounted on a motorized tripod. A lens with a focal length of 35mm is used in our setup, and the EBS-EKF software is running on the Rasberry Pi with an update speed of 350Hz. A TC66 power meter is used to monitor realtime power consumption, which is visualized in the demo GUI.

BibTeX


      @article{reed2025ebs,
        title={EBS-EKF: Accurate and High Frequency Event-based Star Tracking},
        author={Reed, Albert W and Hashemi, Connor and Melamed, Dennis and Menon, Nitesh and Hirakawa, Keigo and McCloskey, Scott},
        journal={arXiv preprint arXiv:2503.20101},
        year={2025}
      }