Novel generative algorithms based on AI achieve superior results in text-based image generation or image-to-image translation. Additionally, human-robot collaboration is often required to solve various problems in numerous domains. The main goal of this thesis is to enable collaboration of human and robot for the purpose of visual arts creation (e.g. drawing, painting, or sculpting). Such a collaborative setup can result in novel tangible artistic results. This way, the creativity of humans and generative power of artificial intelligence can be combined to discover novel creations in visual arts. The vision of artificial intelligence will be enabled by a color camera. The acquired images will be processed by neural networks and by generative algorithms based on stable diffusion. Optionally, text-based input can also be provided during the collaboration. The output image from AI generation will be further processed to identify the series of strokes for robot arm that will then execute these strokes. Optionally, the ceiling-mounted projector may be used to indicate future intentions of AI before drawing. The human-robot collaboration can be achieved in an alternated fashion to ensure high safety of the setup. The implementation of robot manipulation will be done using Robot Operating System (ROS) and the AI algorithms will be implemented in python. The developed methods will be evaluated qualitatively in an expert study.

Type: Master thesis

For more information please contact Peter Kán