Earlier works with drawing robots were motivated by ideas of delegation and autonomy to create a distance between the artist’s intentions and the final mark. In contrast, we present Companion, an artistic apparatus integrating a drawing robot with multimodal large language models (LLMs) to reintroduce the artist’s presence through collaboration. Using in-context learning (ICL) and tool use for robot control, the system engages in bidirectional interaction via drawing and speech. Our goal is to develop an agent capable of pushing shared visual storytelling into unexpected aesthetic and narrative territories. Experiments demonstrate the system’s potential as a playful cocreative partner. To validate the artistic contribution, we employed the consensual assessment technique (CAT) with a panel of seven art world experts (curators, a collector, and an artist) familiar with the authors’ prior robotic works. The results confirm that the system produces artworks with a distinct aesthetic identity and professional exhibition merit.
原文链接
分享
引用
