Density adaptive path based on stacked rotation convolution model in ultra-precision robotic optical polishing

Deterministic optics fabrication using sub-aperture tools has been vital for manufacturing precision optical surfaces, industrial robotic polishing, which is a more economical and intelligent method is required in modern fabrication process. However, the challenge of robotic polishing lies in the widely used spiral and raster paths, which may leave excess waviness from the tool path, and the unavoidable constant removal layer is added to obtain positive dwell time which cause low polishing accuracy. The waviness can be removed by either using smoothing tools sequentially or randomizing the tool path. However, process efficiency and accuracy are not well considered in the existing tool-path planning. A density adaptive path based on a stacked rotation convolution model to ensure polishing accuracy and efficiency while avoiding waviness generation is proposed in this study, and then the dwell time is calculated by anti-aliasing space-variant deconvolution. The robotic polisher experimental results confirm that the root mean square (RMS) of the final surface figure has been successfully reduced and stabilized at 7.355 nm, and the convergence effect at unit wavelength has been significantly augmented with an improvement of 367%, reducing the measurement from 28% to 131%; in addition, no obvious mid-spatial frequency (MSF) peak was generated in the PSD analysis of density adaptive path polishing results. Henceforth, the polishing accuracy, efficiency, and MSF error of robotic polishing can be greatly enhanced.