Borja Del Campo Lopez, Deimos Space UK Ltd.; Emma Kerr, Deimos Space UK Ltd.; Jaime Nomen Torres, DEIMOS Space; Chris Dorn, Inverse Quanta Ltd; Stuart Eves, SJE Space Ltd
Keywords: sensor, optical, wide field-of-view, software, image processing
Abstract:
Optical tracking of LEO objects has been extensively studied and exercised. In this paper, a novel surveillance method using an array of collocated fixed Field of View telescopes and the development of a dedicated image processing software is described. The Low-Cost Low-Earth Orbit Surveillance Sensor Array (LCLEOSEN) has been developed to target a similar market as conventional radar systems through the reduction of the cost per sensor while maintaining a similar performance. This paper briefly discusses the use cases for such a sensor, the potential site locations considered, and a method for ranking site selection. The design of the sensor and its dedicated image processing chain is also discussed and, finally, the prototype implementation is described and the results from the test campaign using the prototype are presented and analysed.
With the aim of minimising the overall system cost, the system is composed of Commercial Off-The-Shelf hardware components including a wide FoV lens and a high-sensitivity CMOS sensor per telescope in the array, a computing server for processing data, and a mount and dome for the telescope array. The selection of the lenses will be decided based on the performance assessment of a prototype. Two different lenses were implemented in the prototype to allow analysis of their real-world performance. Using an array of telescopes with full sky coverage allows correlation of objects detected by the different telescope in their different Field of Views in order to improve accuracy and decrease false detections and omissions. This also results in much longer tracklets for LEO objects, which can improve the eventual orbit determination solution greatly.
Considering each telescopes sensor is designed to capture an image every 2-3 seconds, the system has been designed to have near real-time processing capability. This near-real time processing avoids storage issues, as there will terabytes of data collected each night by the full-scale system. Custom built software has been developed to autonomously process the data and has two main functions: processing the captured images and detecting objects in them. The software module in charge of detecting objects has to efficiently detect objects and distinguish between false detections and actual detections while keeping near real-time performance. During the prototype test campaign, it was found that different software solutions were required for different lenses, to optimise the efficiency and results from the system. The results in this paper discuss the filter implemented and the effect it has on the detection rate.
While a final system design has been defined using the result of the prototype test campaign, it is possible that this system could be fully customised to a customer’s needs. By implementing different lenses, and tuning the software, it would be possible to provide a custom design to meet requirements.
The original project this work builds on was done under a grant awarded by UKSA in the ‘Advancing research into space surveillance and tracking’ call.
Date of Conference: September 14-17, 2021
Track: Optical Systems & Instrumentation