CSIRO has unveiled Vetra, a modular edge computing infrastructure designed for real-time robot learning. By processing data locally rather than relying solely on distant cloud servers, the system enables faster response times and safer operations for physical AI and robotics research. To address the high energy and cooling demands of AI, Vetra employs carbon dioxide-based cooling and closed-loop liquid systems. This design minimizes water consumption and is projected to reduce carbon dioxide emissions by approximately 225 tonnes per year.
