The Future of AI in Space

Abstract

Casual observers of NASA spacecraft, systems, and missions might assume that artificial intelligence has long been integral to what NASA does. However, the reality of high-stakes space missions must balance bold concepts with careful engineering, especially risk management. New capabilities, whether based on AI or other technologies, are adopted only when they offer overwhelming benefits to a mission. Even then, a new capability's risks must be well understood and aggressively retired. And yet, space exploration, whether by robotic spacecraft or astronauts, is not for the faint of heart or vision. Since 1998, when we initiated this department via a special issue, AI has made steady progress in the space realm. In particular, two successful flight-technology experiments-the Remote Agent Experiment (RAX) in 1999, and the Autonomous Sciencecraft Experiment (ASE), deployed in 2003 and still functioning on the Earth Observing One (EO-1) platform-validated appropriate uses of AI-based capabilities in future robotic missions. These capabilities also support NASA's renewed emphasis on robotic and human exploration of the Moon, Mars, and beyond