Artistic robotic painting automates the process of creating an artwork. This complex and challenging task includes several aspects: creating algorithms for rendering brushstrokes, reproducing the exact shape of a brushstroke, and developing the principles of mixing paints. This work contributes to the previously unsolved problem of accurately reproducing colors of brushstrokes by means of artistic paints. The main contributions of this paper include: the development of a novel 4-component data-driven mathematical model for artistic paint mixing; the design and implementation of a novel robot capable of accurately dosing and mixing acrylic paints thanks to the improved syringe pumps and the innovative paint mixer; the development of a novel pneumatic system for paint release with a build-in clogging detection mechanism. The capabilities of the designed robotic system are demonstrated by painting four artworks: replicas of Claude Monet’s and Arkady Rylov’s landscapes, a synthetic image generated using the StyleGAN2 neural network trained on Vincent van Gogh’s artistic heritage, and a synthetic image generated using the Midjourney neural network. The obtained results can be useful in various applications of computer creativity, as well as in artistic image replication and restoration, and also in colored 3D printing.
