Robots learn faster and more flexibly when they can use human touch data, but humans and robots feel touch with very different sensors.
COOPER is a single AI model that both “looks better” (perceives depth and object boundaries) and “thinks smarter” (reasons step by step) to answer spatial questions about images.