Expanding data storage in space is critical to enabling more powerful artificial intelligence systems to process large amounts of information directly from orbit.
More powerful hard drives reduce the need to download data to the ground for processing, facilitating faster and more insightful analysis from space.
“This is useful for both AI training and inference,” said Jon Trantham, principal engineer at data storage specialist Seagate.
“We think the fact that there is a large pool of data available and the potential to expand it with AI processing hardware is a new application to space.”
Satellite-to-satellite links, which allow spacecraft to relay data to each other, are also key to speeding communications with the ground, allowing low earth orbit (LEO) satellites to pass over authorized ground stations. Now you can send information without waiting.
Transom has been Seagate’s technical lead for the mission testing a 4-terabyte solid-state drive mounted on the outside of the International Space Station for the past two months.
The storage device is slightly larger than a standard computer hard drive and is being mounted on a small satellite provided by partner and potential customer BAE Systems.
Seagate has been supplying the hard drives installed on the ISS for more than 20 years, but this is Seagate’s first storage device designed specifically for the harsh environment of space.
“The main news from the telemetry results is that the actual environment is more benign than we feared,” Transom said.
“Cold temperatures can be very stressful for solid state drives.”
While stress testing continues, the company plans to refine the device next year ahead of commercial release to LEO carriers in 2026, he said.
Seagate aims to strike a balance between cheaper, less rugged off-the-shelf storage devices and more expensive military-grade hardware.
“This is much larger than most of the alternatives that currently exist,” Trantham added. “Many of them are in the gigabyte capacity range. Terabytes are relatively new here, so that should enable new use cases.”
Although these new use cases are uncertain at this time, Transom anticipates demand for climate data, imagery, and video processing applications.
Onboard autonomy
Michele Faragali, chief technology officer at Canadian software startup Mission Control, said that in addition to the real-time use of space data, bringing AI to spacecraft will require a high degree of autonomy to support real-time operations. said that it is important for the mission.
Onboard autonomy is essential when communication with the operator is delayed or limited, and when data volumes exceed affordable downlink capacity. Timely action is required when data is easily lost or when complex operations require quick and accurate decision making.
“Onboard AI can therefore benefit a diverse class of space missions, from robotic exploration to disaster response using Earth observation satellites,” he continued.
Faragali emphasized that advances in onboard autonomy through AI could pave the way for new types of space missions.
These include robotic underground exploration of Europa and the autonomous coordination of multiple space-based assets to observe large areas using a method known as “chip and cue.” In this method, one sensor identifies an area of interest (chip) and directs the other sensors to investigate. Further (cue).
This article first appeared in the November 2024 issue of SpaceNews Magazine.
