A maker called msana has posted his latest project in Hackster.IO. It is a robot, based on a 3D printer, with a Raspberry Pi & camera module attached. The camera takes a photo of a Sudoku puzzle placed on the print bed which is then image-processed via OpenCV to determine the exact layout of the puzzle. The puzzle is then abstracted and run through intelligence library TensorFlow to determine the solution. The robot part then writes the solution onto the grid using a pen attached via the motors. You can read more about the project here, (although I would like to have seen much more detail on the robotic part of the build). The code is available on GitHub. See it in action below: