Abstract
This Sketch-Bot was developed using a 6-DOF robotic arm system, designed to automatically generate sketches of human portraits from live image input. The aim of this project is to combine computer vision, machine learning, and robotic automation to convert digital portraits into physical sketches. The process starts with recording a live image via a laptop camera, which is submitted to an online processing module to achieve a clean line sketch. The robotic arm then employs many algorithms to dominate the features and draws this reference image eight times on an A4 sheet-four times using a pencil and four times using a sketch pen-each at different speed and precision levels. All eight sketches are then assessed using machine learning models comparing them with the original line sketch in order to ascertain their accuracy. The sketch with highest accuracy ranking is chosen as the final result and displayed to users via a feedback interface. This makes it possible for people to indicate if they enjoyed the sketch and leave comments, which enables us to gauge overall satisfaction and aesthetic quality. Our method reduces the necessity of a complicated hardware setup by only using a single robotic arm.