Australian AI Technology Could Revolutionize Lunar Surface Exploration
A robotic arm observes a 3D-printed lunar landscape in Adelaide’s Australian Institute for Machine Learning (AIML) lab.
By Sofia McLeod
AI researchers in the South Australian city of Adelaide are hoping to use their own proprietary space technology to revolutionize how spacecraft survey the lunar surface. While the initial focus is on lunar surveys, the technology's versatility could extend to mapping any planetary surface, provided there's existing data from previously mapped catalogs for reference.
In a groundbreaking paper accepted for publication in the journal Astrodynamics, the authors introduce STELLA (Spacecraft crater-based localization for lunar mapping), an innovative AI-powered Crater-Based Navigation (CBN) pipeline specifically designed for long-duration lunar mapping missions. Without GPS in space, spacecraft rely on their own navigation systems, and traditional methods like radio-ranging can have significant errors, sometimes reaching several kilometers. This is where CBN comes in, utilizing vision-based navigation techniques that employ images of the Moon's cratered surface to pinpoint a spacecraft's position with remarkable accuracy.
The STELLA system, as demonstrated in the paper, outperforms conventional lunar surveying methods, showcasing its potential to revolutionize lunar exploration. Here's how it works: A camera onboard the spacecraft captures an image of the Moon's surface, and CBN identifies craters within that image, matching them to known craters from pre-existing catalogs. These matches enable the system to determine the spacecraft's precise location. This process is purely about position estimation; CBN doesn't create new maps but accurately determines the spacecraft's position using craters as landmarks.
The system's capabilities are further enhanced by its ability to operate across various regions of the Moon, including the poles and large craters. In orbital scenarios, STELLA can leverage position estimates made using images captured before and after entering shadowed regions to infer the spacecraft's orbital trajectory. This means the spacecraft can still estimate surface positions while passing over permanently shadowed regions, such as those at the Moon's south pole.
The paper presents the first comprehensive study of a vision-based spacecraft navigation system tailored for long-term lunar orbital missions. Once STELLA obtains a 'good' image of the lunar surface, containing enough clearly identifiable craters, it becomes autonomous, analyzing the entire surface pattern of visible craters. This vision-based navigation system operates similarly to how humans use sight to understand their surroundings, achieving meter-level accuracy.
AI plays a pivotal role in this technology, enabling CBN to adapt to the challenging conditions of long-term lunar orbital missions. It allows the system to recognize craters regardless of lighting conditions, camera angles, or terrain types. The technology is being developed to meet the positioning needs of the upcoming Japanese TSUKIMI (Lunar Terahertz Surveyor for Kilometer-scale Mapping) mission, scheduled for a 2028 launch.
TSUKIMI's primary objective is to survey the Moon from lunar orbit, seeking resources like water. The crater-based positioning algorithm developed by Adelaide University will be instrumental in pinpointing the exact locations of these resources. The AI pipeline's capabilities extend beyond lunar science, enabling accurate maps of key resources and precise planning for lunar infrastructure and habitats.
In summary, Australian AI technology is poised to revolutionize lunar surface exploration, offering unprecedented accuracy and autonomy in spacecraft navigation. This development promises to significantly enhance our understanding of the Moon and pave the way for more efficient and effective lunar missions.
