Icarus Robotics is developing robotic labor systems for orbital environments, applying embodied AI to address the physical workforce constraints inherent to space operations. The company's technical approach centers on integrating autonomous capability with human-in-the-loop supervision, recognizing that fully autonomous deployment in unstructured space tasks remains an open problem. Core work spans learning from demonstration, where teleoperator behaviors are distilled into reusable autonomous policies, and designing control architectures that partition responsibility between onboard autonomy and ground-based human oversight.
Engineering efforts focus on closing the gap between terrestrial robotics performance and the operational demands of orbit - variable lighting, microgravity dynamics, communication latency, and constrained power and compute budgets. The team works across the full stack of embodied AI, from perception and manipulation planning to policy learning and real-time execution, with an emphasis on systems that have been validated beyond simulation. The work draws on techniques in robotics and machine learning but is grounded in the practical constraints of deploying hardware in space, where iteration cycles are long and failure tolerance is low.