A Chinese Calligraphy-Writing Robotic System Based on Image-to-Action Translations and a Hypothesis Generation Net

This paper attempts to use a delta robot’s structure and reliable coordinates develop self-learning Chinese calligraphy-writing system that requires precise control. Ideally, achieve human-like behavior, robot can learn stroke trajectories autonomously present the beauty of calligraphy characters. Unfortunately, state-of-the-art approaches have not yet considered presentation resulting from angles rotation tilt brush. presents an integrated consisting processing module, hypothesis generation net (HGN) learning model, robot, image capture module. Our approach utilizes both module information HGN model automatically produce five degrees freedom action instructions. Based on instructions, completes writing. Then, provides feedback writing for error calculation coordinate correction. We utilize mean absolute percentage verify performance results. A correction algorithm linear regression were used improve results (less than 2% error). After several cycles, written approached target sample finally. Consequently, produced by prove our proposed is capable