Nebula Public Library

Mission Control deployed a low-level implementation of the OPS-SAT SmartCam model using a Field Programmable Gate Array (FPGA), comparing against a high-level CPU model using Tensorflow Lite. Experiments showed that the FPGA implementation reproduced the precision and accuracy of the high-level model, while running at a slower speed. Further optimizations of the FPGA are expected to close the gap in timing and unlock new methods for deploying deep learning on spacecraft.

With the “democratisation of space”, and the proliferation of large non-geostationary orbit (NGO) constellations, the pace at which human-made objects are being deployed in orbit demands urgent action in defining and supporting a global Space Traffic Management (STM) plan. Promptly identifying, tracking, and cataloguing of resident space objects (RSOs), in particular after orbit insertion or break-up and fragmentation events, as the recent Russian anti-satellite weapon (ASAT) test put in evidence, is of critical importance to enable a safe and sustainable use of space.

A proof of concept of an attitude control system based on Artificial Intelligence, as an alternative of current state-of-the-art systems based on model approaches, has been developed and tested in a simulator and in the EM of the OPS-SAT. The system relies on Computer Vision and Deep Reinforcement Learning algorithms to command the OPS-SAT reaction wheels in order to actively modify its attitude, so that the on-board optical camera keeps framed and focused on a target without the aid of any other instrument, such as the star tracker, the ADCS or the GPS receiver.

OPS-SAT mission is a key element for operational validation in flight (or IOD/IOV) of the DeepCube service of Deep Learning at the edge that our company is preparing with the support of the GSTP programme (GSTP Make, ESA-CNES) and the R&D performed in the CORTEX project (Permanent Open Call EOEP-4, ESA/Phi-Lab, https://esacortexproject.agenium-space.com). The goal of the proposed experiment is to execute on-board the inference of the simplified models defined in those projects.