Toward transforming space exploration with artificial intelligence neuromorphic computing

With space missions venturing farther from Earth and becoming increasingly complex, the integration of neuromorphic systems is imperative. Inspired by the architecture and function of the human brain, neuromorphic computing offers significant advantages in terms of power efficiency, real-time processing, and autonomous decision making. This review explores the emerging role of artificial intelligence (AI) neuromorphic computing in space exploration, highlighting its current state and potential future strategies. Recent advancements have demonstrated its potential for long-duration missions that demand autonomous decision making and real-time data processing. For example, in the BrainStack project, neuromorphic processors operated with a power consumption of less than 1.5 W (W), exhibiting approximately 90 % greater energy efficiency over 15 W consumption by conventional von Neumann processors. However, several challenges persist, including physical constraints imposed by the space environment, such as radiation exposure, extreme temperatures, and vacuum conditions, as well as greater integration with existing spacecraft systems and further improvements in power efficiency for long-duration missions. Therefore, we identify environmental and technical challenges, as well as future directions for deploying neuromorphic technologies in space. To achieve enhanced energy efficiency and operational reliability, future strategies should focus on improving hardware and software capabilities, optimizing spiking neural network algorithms, and incorporating neuromorphic computing into mission design and validation phases. This review is intended to serve as a reference for advancing AI neuromorphic computing as a strategic technology that enables sustainable mission execution and autonomous exploration in extreme space environments.