Using genetic algorithm for drawing path planning in a robotic arm pencil sketching system

Abstract

The purpose of this study is to implement a robotic arm pencil sketching system through a novel drawing path planning method. Firstly, to enhance the craftsmanship and uniqueness of the artwork, the original image is transformed into an animated style using AnimeGANv2. After image processing, the image is divided into boundary and background layers for drawing path planning. In the realm of path planning, traditional methods often involve manual path design by artists, and only recently have automated path planning methods been proposed. However, these recent approaches still have some limitations. For instance, a pursuit of efficiency can lead to overly uniform strokes, making the artwork resemble a print rather than a genuine piece of art. Therefore, this study introduces a new path-planning approach based on the traveling salesman problem (TSP), utilising various algorithms for path-planning experiments. Different stroke effects generated by distinct algorithms are applied to appropriate areas. The aim is to let the generated artworks exhibit the technological appeal of algorithms while retaining the distinctive characteristics of pencil sketches. For robotic arm control, a force sensor is integrated onto the arm, and a proportional-integral-derivative (PID) controller is employed to monitor and precisely control the drawing force. Experimental results demonstrate that artworks created using genetic algorithms to generate drawing paths exhibit distinct colour-layering effects, emphasising the characteristics of pencil strokes. In comparison with previous methods, this approach effectively addresses the issue of overly uniform and rigid strokes in past techniques.