In a three-week university module, we built a speech-to-drawing system that turns spoken requests into custom pen illustrations. The goal was to explore novel experiences using natural language and AI models: Using GPT-3.5, Stable Diffusion, and a custom-built plotter, it generates unique images drawn on paper for visitors to keep.
LLM Drawing Robot
About
Video
Process
Interaction
We aimed at creating a straightforward speech-input GUI inspired by social media voice messages to encourage natural, human-like interaction. This interface was central to our goal of exploring novel ways to engage with LLMs by replicating AI sentience through conversational exchanges.
Results
Learnings
The resulting prototype serves as a proof of concept for pseudo-sentience in AI. This foundation is core to the novel developments in HRI enabling more natural and meaningful interactions between humans and robots.