A removal machining robot was already developed for foamed polystyrene materials. The proposed robotic CAM system provided an interface without using any robot languages between operators and the machining robot, so that a teaching-less operation could be realized. In this paper, a preprocessor with a depth sensor for converting stereolithography data (STL data) into cutter location source data (CLS data) is first proposed for smart robotic removal machining. Due to the preprocessor, no commercially provided CAM system is needed for making a desired trajectory. The STL was proposed by 3D System as a file format mainly used for 3D printers. The STL includes numerous triangular patches for constructing 3D objects. The preprocessor enables the machining robot to run through a zigzag path or a circular spiral path calculated from STL data. Then, a smart smoother and generator are introduced for smoothing coarse point cloud data (PCD) and for generating STL data from the smoothed PCD, respectively. The performance of the developed approaches are verified through experiments of smoothing, generating and machining.