Spacefarer AI

Spacefarer AI is Mission Control’s platform for safely and reliably deploying artificial intelligence on spaceflight and robotic hardware, enabling ‘onboard intelligence’ for autonomous spacecraft and mission objectives. It has validated space heritage, with deep learning neural networks deployed on two missions (Earth orbit and the lunar surface), including the MoonNet AI model used for lunar surface feature classification for navigation and instrument targeting during a 2023 lunar mission. The platform is built around three components: a Data & Model Pipeline for building explainable, tested neural networks for rovers, satellites and stations; a Deployment Toolkit for deploying models to Systems-on-a-Chip from various hardware manufacturers with hybrid CPU/FPGA optimization and no vendor lock-in; and a Data Explorer for visualizing AI outputs and intelligent overlays for human-AI collaboration. Mission Control has applied Spacefarer AI to Earth observation and wildfire detection (under a Canadian Space Agency smartEarth program contract), terrain classification for autonomous ground vehicles, and ESA-funded federated learning research for distributed lunar rover computing, working with partners including NASA, ESA, the Canadian Space Agency, Canada’s Department of National Defence, Honeywell Aerospace, Xiphos and ispace.