The technical work performed in the context of DeDUST on the algorithm development side consists of several parts.
The first part is the development of image processing algorithms that allow for the detection of space debris in star tracker images. This includes optimizing the already running algorithms on the star tracker, to make "room" on the processing electronics to handle the extra computational workload for the DeDUST algorithms.
Second, the observations that are generated from part one are processed to perform characterization of the debris element.
Finally, a full simulation framework is made to function as a digital twin for the DeDUST project.
In the image processing part, an improved blob detection method has been developed, which is able to detect bright areas in the star tracker image that can correspond to space debris elements. These elements typically have a higher relative velocity and show up on the sensor as streaks. The detection of streak-like blobs was a novel development within arcsec. After blob detection, the debris elements are tracked over multiple consecutive images using a kalman filter to estimate position and velocity. This leads to increased precision, as well as more accurate error estimates. Finally, extra information is extracted from streak-like blobs. This information exists of location, direction, length and total brightness.
The debris characterization part consists of the development of an orbit determination pipeline to go from debris observation to debris orbit prediction. This is done by processing the raw observations to get angular measurements, then feeding the angular measurements into a batch least squares precise orbit determination. Using these algorithms, accurate predictions can be made about the observed debris.
Finally, a simulation framework was set up to predict and characterize results for the DeDUST network. A combination of orbital dynamics, photometric brightness simulation and observation simulations allow us to determine an expected number of observations for a satellite with a given orbit and a given attitude profile.
Next, the database framework was developed. A database that gathers and makes available the SSA data was developed and an API was made based on conversations with users. The database is set up and running using ground-based SSA data gathered by arcsec's star trackers. The database allows to track what each satellite operator is providing and what each SSA data users is requesting, so that a payment scheme can be built upon it.
Finally, the full ecosystem is set up and tested. The algorithms are implemented on ground-based star trackers that have observed the night sky for several weeks, gathering real SSA data. The star trackers are able to carry out their star tracker functionality and SSA capability in parallel, without loss of performance. The SSA data has been automatically fed into the database and has been made available to first pilot users. We are currently working on demonstrating DeDUST for the first time in orbit.
