Landing a spacecraft on the Moon with exceptional precision, while relying solely on ground-based tracking systems, is a complex, expensive, and logistically demanding process. Modern space missions largely depend on telemetry from Earth to provide crucial initial position estimates, which significantly limits the capabilities of future missions and increases operational costs. A recent doctoral project, co-financed by the ESA Discovery program, the German Aerospace Center (DLR), and the company OHB, has developed an innovative solution that could enable spacecraft to find their way in deep space completely on their own.

Willem Oliveira's research, titled "Global Autonomous Navigation System for Planetary Approach and Landing," addresses a fundamental challenge of space travel: how to achieve precise navigation without constant support from the home planet. The motivation for such a breakthrough goes beyond mere cost savings; autonomous navigation opens the door to entirely new types of missions, such as exploring celestial bodies using swarms of small, coordinated spacecraft operating as a single organism.

Dependence on Earth: The Current Achilles' Heel of Space Missions

The current generation of spacecraft for precision landing relies on advanced systems known as Terrain-Relative Absolute Navigation (TAN). These optical navigation systems work by comparing images taken by cameras on the spacecraft with existing, detailed surface maps stored in its computer. In this way, they can determine their location with extraordinary accuracy, which is a prerequisite for landing at a predetermined point with minimal deviation. However, this entire sophisticated system has one key weakness: it is dependent on initial data sent from Earth.

Willem Oliveira, the author of the research, clarifies the core of the problem: "TAN systems require an initial state estimate, that is, rough information about where the spacecraft is, and this data can currently only be obtained via telemetry from Earth. It is precisely this link that we want to eliminate."

The existing process works as a closed loop that begins on Earth. Mission controllers send the spacecraft initial position estimates, which allows it to select the appropriate local terrain maps from its vast database. Only then can the TAN system begin to operate, using these maps and images from the onboard cameras to obtain much more precise estimates of its position and velocity. Although this data refinement loop operates autonomously, it cannot even begin without initialization from Earth, which creates a bottleneck in operations.

An Innovative Solution That Looks at the Surface and the Stars

The approach developed in this project exclusively uses optical navigation methods, which only require relatively inexpensive visual cameras for measurements relative to the surface of a celestial body. The system ingeniously combines the analysis of what is known as optical flow with measurements obtained from a star tracker (STR) over two complete orbits around a planet or moon.

Optical flow is a term that describes tracking how brightness patterns, i.e., recognizable features on the surface like craters or rocks, appear to move between successive images taken as the spacecraft moves. From these movement patterns, the system can derive key information about its own direction of motion. This data is then merged with precise measurements of the spacecraft's orientation in space, obtained from the star tracker, to estimate fundamental orbital parameters. Star trackers are, in essence, small cameras that photograph the starry sky and compare it with a star map to determine the spacecraft's orientation with incredible precision.

"We use the images we take of the surface to determine the general shape of the orbit," Oliveira explains. "This gives us sufficiently precise initial information to initialize a much more complex and precise TAN system, without the need for any signal from Earth."

Testing and Promising Results

The new approach was tested using CNav, a crater-recognition-based navigation system that can operate in different modes, depending on the accuracy of the initial state estimate. The results showed that the camera's elevation, i.e., the angle at which the camera looks at the surface, significantly affects the success rate. For elevations between 0 and 30 degrees, successful initialization of the CNav system was achieved in approximately 90% of simulated cases. However, at an elevation of 60 degrees, the success rate dropped to about 60%. The decisive factor, as it turned out, is the average surface area of the terrain visible in the images used for optical flow analysis. The larger the visible area, the more reliable the system is.

Although this project is now formally completed, research continues with the aim of creating an operational system that could be tested on a real spacecraft in the future. Moving from computer simulations to space-ready hardware is the next big step.

A Look into the Future of Autonomous Space Exploration

"The motivation for this work lies in the current need to initialize absolute navigation algorithms with a state estimate obtained from Earth," says Massimo Casasco, Head of the Guidance, Navigation and Control (GNC) Section at ESA. "What we hope to achieve is fundamentally greater autonomy on the spacecraft itself, which brings a dual advantage – not only saving operational costs but also making these operations more robust and resilient to errors."

Full autonomy means that missions can react more quickly to unforeseen circumstances, without having to wait for instructions from Earth that can take minutes, or even hours, to travel. This is crucial for the safety and success of future, more complex missions.

"We see this result as an important building block for further increasing the autonomy of spacecraft – especially for small platforms like CubeSats, which could one day navigate around the Moon on their own," said Dr. Stephan Theil, Oliveira's supervisor at DLR's Institute of Space Systems. "I sincerely hope that we will be able to implement this technology in real missions in the near future, perhaps even very soon."

The project originated through ESA's Open Space Innovation Platform and was funded through the Discovery element of ESA's Basic Activities. It represents a key step towards the autonomous operation capabilities that will be essential for future exploration of the Moon and other planets, paving the way for missions that are today only in the realm of science fiction.