Orbital debris tracking is one of the most important measures for collision avoidance. Current technology is capable of tracking debris that are 10 centimeters or larger in size. However, the number of small (untrackable) debris is higher in numbers, and still dangerous. In early 1990s, NASA and DoD collaborated on a set of ground based laboratory hypervelocity impact (LHVI) tests called SOCIT. These tests helped engineers and scientists to understand the effects of orbital debris impact and provided a statistical representation of the debris environment. As the technology advanced, many metallic components were replaced by carbon-fiber reinforced polymers (CFRP), Kevlar, and high performance materials. This change in material compositions of modern satellites have led to the different expectations in fragmentation characteristics. Thus, the DebriSat project took place in 2014. The DebriSat project was created by NASA, DoD, The Aerospace Corporation, and the University of Florida (UF) to provide characteristic data to improve orbital debris modeling capabilities. A test article was designed and developed to be representative of a modern-day LEO satellite and subjected to a LHVI at the test facility at Arnold Engineering Development Complex (AEDC) in Tennessee. The post-impact test system and fragments were collected and transported to UF for fragment identification and characterization. During characterization fragments are individually assessed for material, shape, mass, and size (cross-sectional area, volume, and characteristic length). The UF team established various dimensional analysis instruments for material assessment and interconnected these instruments within a database referred to as the Debris Categorization System (DCS). Around 2022, a bias related to selection of metal fragments was identified in the DCS. Thus, an investigation took place to address the cause and propose a solution was devised. After a survey, it was determined that the primary bias arose from the thin fragment processing approach for volume computation. To resolve this bias, an algorithm was developed to compensate the volume overestimation and utilized to “relabel” the dataset associated with metal labels (i.e., primarily stainless steel and aluminum). In addition, a supervised Machine Learning (ML) model was trained based on the new material definitions obtained from the relabeling activity. When applied to a dataset labeled as “metal”, the ML model was able to detect stainless steel fragments from the dataset by over 90% accuracy. However, the ML model was not able to obtain similar high accuracy scores for other materials (aluminum, MLI, glass, epoxy, etc.). This manuscript discusses the methods of volume optimization and presents a concept model with the ability for detecting all types of single-material fragments.
Őndeş et al. (Thu,) studied this question.