Story
September 03, 2025
Stock image
As artificial intelligence (AI) capabilities for space advance, onboard edge processing is also improving, providing near-real-time access to data insights that were previously unobtainable. It’s a shift from information access to actionable intelligence, as AI helps manage the increasingly crowded data landscape and enhance electronics operating in the space environment. Preprocessing at the edge provides less – but more valuable – data, optimizing transmission bandwidth and reducing data latency to rapidly deliver critical information directly to the point of need. Onboard AI is also being used to monitor space systems and manage in-orbit communication, troubleshoot onboard operations, and enable dynamic mission shifts.
Artificial intelligence (AI), broadly speaking, comes with many different algorithms that are applied to various applications. The key to using AI for space is to understand what algorithms are most relevant to facilitate edge processing and computation within a space system. One example is the International Space Station, which currently has very controlled use cases of large language models (LLMs) in progress, but most space applications will involve machine learning (ML).
ML, a function very familiar to military and defense, is currently providing most of the edge-processing applications on Earth, and this concept can easily translate into the space environment. As the preferred AI method to take complex situations and boil them down to a simple mathematical probability, ML can provide guidance for many edge applications, such as providing input on what action to take within a space system’s main command and data handling (C&DH) functions.
Space ecosystems with robust data-processing capabilities continue to evolve, integrating enhanced surveillance, image processing, communications, and video analytics. These systems are being designed to not only process information, but also compute the data on board and provide only relevant information to the warfighter, giving them an extra split-second decision-making advantage during defense operations. Winning the battle goes beyond weapons and platforms; it’s about processing information faster and computing that data more accurately to provide the users with that crucial edge. (Figure 1.)
[Figure 1 ǀ AI in space applications provides enhanced strategic intelligence for critical defense operations.]
Expanding autonomous space applications
AI-based intelligence is enhancing operations within and among in-orbit autonomous space systems, leading to more intuitive computing capabilities. Powerful and compact space-rated systems are fueling new areas of autonomous space capabilities, such as advanced image, signal, and edge processing in areas as diverse as object detection, threat assessment, and earth observation.
Satellite constellations are another excellent example of autonomous operations. Simple AI/ML algorithms can engage controls to reorient the satellite to change the drag profile and change the position of the satellites or space vehicles. Not only can AI maintain the relative positions of each satellite in real time, but the AI engines in the swarm can also autonomously reconfigure a specific satellite’s mission profile to adapt to the mission requirements.
System-based operations
The most common hardware approach to AI/ML is using GPGPUs [general-purpose computing on graphics processing units], where hundreds or thousands of processors run in parallel, which is instrumental in managing the increased computational demands of today’s space-based systems. Nowhere is data input, processing, and clarity more applicable than in mission-critical military operations. GPGPU-based Âprocessing can manage the exponential number of data inputs and image resolution now used in space systems. The method is flexible to be reprogrammed for the changing missions.
GPGPU processing has helped to elevate AI to places where it can help optimize complex, highly sophisticated computing by reliably managing higher data throughput and balancing system processing for more efficient computing operations.
AI-enabled constellations
AI-based constellations are revolutionizing what is possible in space applications. IQSat, the first AI-enabled PicoSat of its kind, uses onboard AI and ML to actively interpret and analyze data directly in space, unlike traditional satellite systems that passively collect data. This AI/ML onboard capability drastically reduces the time between data collection and actionable insight, so what once took hours or days can now happen in near-real-time. Critical information, like tactically actionable updates, is delivered directly to the user.
Using Intuidex’s Higher Order-Low Resource Learning (HO-LRL) date-modeling technology, the small, rugged platform detects patterns, anomalies, and changes in behavior as they happen, not just after the fact. IQSat is small enough to be held in the palm of your hand and can be deployed in a constellation of five to thousands. (Figure 2.)
[Figure 2 ǀ The new IQSat PicoSat brings AI capabilities into a low-cost COTS-based platform for low Earth orbit (LEO) missions.]
Operating in a constellation adds more complexity to the flight software of the satellite. These additions may include constant adjustments to maintain the correct operational formation plus coordinating the multiple functions across different satellites, depending on the task at hand. These are all functions that are expertly handled using AI. In the event of a satellite malfunction, AI algorithms can detect failure and enable one of the other satellites in the constellation to provide the needed functionality and turn off the other payloads.
This new AI-enabled space-based technology reduces latency, improves response times, and ensures continuity in extreme environments where communication is limited. Using AI in this way not only enables dynamic onboard response for autonomous system management, but also gives the user unprecedented intelligence insight into critical areas such as surveillance, threat detection, object tracking, strategic monitoring, and border security.
Ensuring space system survivability
Building space electronics also centers on risk-mitigation strategies. Successfully applying AI-based technologies within in-orbit networks requires ruggedization to withstand the extreme environmental factors of space, with careful consideration of parts testing and characterization prior to launch, shielding, redundancy, power management, and EMC mitigation.
Edge AI hardware in space-rated systems requires a very different degree of survivability than what is required in automobiles or factories. Specific techniques are used to ensure that edge-processing hardware can endure the harsh environment of space and that the onboard AI/ML capabilities can help moderate and manage the system in response to environmental effects through checking for anomalies.
Radiation: The biggest and most concerning issue of space flight is radiation. For lower-orbit and shorter missions, radiation can impact the life of satellites or impact function when there are stronger solar events while being deployed. For longer-term missions in deeper orbits, resilience to prolonged radiation exposure is critical for mission success. There are several well-known techniques – such as shielding and redundancy – that can help mitigate radiation, but depending on the application, these techniques can have a varied degree of success and the extra weight is not always possible.
Another mechanism that can be used in a space-rated system is the ability to restart: Special hardware can be used that considers the impact of radiation so that it knows when to power off and back on after annealing has occurred or solar radiation has passed.
Electromagnetic compatibility (EMC): EMC can affect the vehicle itself or the multiple payloads within, causing interference and disruption. By testing and understanding the EMC profile of the different payloads as well as of the vehicle, special hardware can reduce the power of the vehicle to make sure there is no interference and that the payload or vehicle will work properly. Testing to prove survivability is adequate in most use cases.
Heavy ion damage: It is crucial to understand the effects of heavy ion damage – a special form of radiation from deep space – to the space vehicle and the payload. Testing and characterization can help prepare hardware and software to mitigate these events, a critical use of successful AI applications in space.
Up-to-date operational intelligence is paramount to the success and safety of modern defense initiatives, and AI has a unique opportunity to provide significant benefits across a range of activities. Today’s AI-enabled space-based infrastructure is helping to fuel definitive strategic shifts in modern warfare, not only by helping to manage the data available from today’s embedded systems, but also by facilitating the processing and computation of that data in near-real-time. The resulting reduced latency and optimized bandwidth usage provides critical actionable data to the end use or for the system to act autonomously.
Ralph Grundler is Director of Space Business Development and Space R&D at Aitech. Ralph has 30 years of computer and semiconductor industry experience, particularly in the development and marketing of semiconductors, IP, SoCs, FPGAs, computers, and embedded systems. Before joining Aitech, he worked at Flex Logix, a supplier of rad-hard eFPGA technology; and Synopsys, where he did marketing for interface IP subsystems working with space ASIC developers and IP prototyping hardware as well as business development for the Japanese market.
Aitech • https://aitechsystems.com/
Featured Companies
19756 Prairie Street
Chatsworth, CA 91311