Loris camera & AI system update: first light & first paper!

Commissioning activities on SpIRIT are progressing positively, with all University of Melbourne systems staying healthy after more than 135 days in space.

Today, we are very pleased to share that a scientific paper on the Loris camera and Artificial Intelligence system has been accepted for publication and presentation at the AI4Space Workshop at the 2024 Conference on Computer Vision and Pattern Recognition that will be held in Seattle (WA – USA) in June. The conference is one of the premier events in the field and we are thrilled about the opportunity to present the innovative design of Loris.

The research and publication has been led by Dr. Miguel Ortiz del Castillo at the Melbourne Space Laboratory with substantial cross-disciplinary contributions from colleagues across multiple departments and faculties at the University of Melbourne. The paper is available for download in an open-access format on arxiv.

One of the innovative aspects of Loris is the use of the JPEG-XL progressive compression standard, that allows us a highly efficient downlink of quality images from the on-board cameras while minimising bandwidth. The image below is among the first views of Earth from SpIRIT and only required downlink of 60kB of data.

The team is looking forward to continuing to demonstrate in-orbit capabilities of Loris and of the other payloads on board SpIRIT.

Earth as seen from SpIRIT

Image caption: RGB image of Black Sea coastline acquired from SpIRIT on March 31st 2024, 9:09 UTC. The image was captured by Loris camera 5 while the satellite was orbiting above the Adriatic Sea. The image covers a region of approximately 600km along the horizontal direction and shows part of the Earth limb as well in the bottom right corner. The native 8M pixel resolution was compressed to just over 60kB used to downlink the image shown thanks to the progressive JPEG-XL algorithm implemented by Loris. Image credits and copyright: University of Melbourne – Melbourne Space Laboratory.