A sunlight lamp illuminates a satellite model covered in golden thermal insulation. A cup-shaped thruster extends from the model’s centre, reflecting a rainbow of colours. A few metres to the left, not captured in this image, a camera is brought closer and closer to the model, scanning the scene in this simulation of a rendezvous in space.

Two companies from Luxembourg, LMO and ClearSpace, have used the European Space Agency’s Guidance, Navigation and Control (GNC) Rendezvous, Approach and Landing Simulator – GRALS – for the testing of their autonomous satellite navigation technology.

GRALS is part of ESA’s Guidance, Navigation and Control Test Facilities at the agency’s technical centre, ESTEC, in the Netherlands. It consists of a duo of robotic arms, mounted on long rails. One holds the navigation unit of a ‘chaser', the other holds a satellite model as the ‘target’, suspended in the centre of a dark room, simulating space.

As part of the Development of In-Orbit Servicing Space Situational Awareness Payloads (DIOSSA) activity, LMO and ClearSpace are developing a navigation system for rendezvous scenarios in space.

“The number of uncooperative human-made objects in space that are in need of servicing is rapidly increasing,” explains Sabrina Andiappane from ClearSpace. “Whether these are satellites that ran out of fuel, reached the end of their operational life, or were damaged by debris, they are progressively cluttering Earth’s orbital environment.”

“In order for us to be able to refuel, repair, or deorbit them, we first need to be able to identify and approach them,” notes Alexander Finch from LMO.

“Working out where another satellite is relative to you is actually quite difficult. Imagine trying to walk to a friend’s house, go in through the front door, and then lightly tap your friend on the nose – you need to be able to see what you are doing. We’re trying to provide the ‘eyes’ for satellites to do just that, by developing what is called a ‘Vision-Based Navigation’ (VBN) system.”

VBN systems enable satellites to identify and approach or avoid other objects in space with the help of AI – in a way similar to self-driving cars.

The satellite model shown in the image was developed by ClearSpace and includes typical features of a real-life satellite. “At first, we used a smaller model to simulate a greater distance between the target and the camera. This larger model was used for the later, more close-up stages of a rendezvous,” adds Alexander. “These real-world models complement and validate computer-generated images we use to train our AIs.”

“To test VBN systems in the final, close-proximity phases of a rendezvous simulation, ClearSpace developed a larger physical model of a satellite,” says Sabrina. “High-fidelity models like this one complement and validate the computer-generated scenarios to ensure that autonomous servicing technologies are robust and reliable in real-world conditions.”

Joris Belhadj, ESA’s GNC system engineer, comments: “Although GRALS already enables representative real-world hardware testing in simulated space scenarios, we continue developing the facility to keep meeting the needs of European industries. In the future, we will offer new features and accommodate scenarios that are more complex, taking another step closer to the reality these instruments encounter in space.”

The DIOSSA activity was funded by LuxIMPULSE, Luxembourg’s national programme for research and development.

[Image description: This is a close-up photo of a satellite model, seen from the side. A prominent feature is a cone-shaped nozzle extending from the centre of the model. The nozzle has an iridescent surface with shades of blue, purple, red and gold. The rest of the model is covered in crinkled gold-coloured thermal insulation, which is shiny and reflective. The background is out of focus, but horizontal metal rails mounted on the wall behind the model are visible. The lighting is bright and highlights the textures of the nozzle and the insulation.]