HAMLET: A Hierarchical and Adaptive Multi-Agent Framework for Live Embodied Theatrics
Researchers have introduced HAMLET, a novel hierarchical and adaptive multi-agent framework designed to create immersive, interactive theatrical experiences using large language models (LLMs). Addressing limitations in existing drama generation methods, such as lack of initiative and poor physical scene interaction, HAMLET generates a narrative blueprint from simple topics to guide real-time improvisational performance. Each actor agent utilizes an adaptive reasoning module to make decisions based on personas, memories, and goals during complex group interactions. Crucially, the framework enables embodied interactions where agents manipulate scene props, such as opening letters or picking up weapons, thereby updating the global environmental context. To ensure quality, the team developed HAMLETJudge, a specialized critic model for automated evaluation. Experimental results indicate that HAMLET successfully produces expressive, coherent, and physically interactive theatrical performances autonomously. This advancement represents a significant step forward in interactive narrative fields, leveraging AI to enhance live performance immersion without requiring detailed manual input.
Wire timeline
HAMLET: A Hierarchical and Adaptive Multi-Agent Framework for Live Embodied Theatrics
Researchers have introduced HAMLET, a novel hierarchical and adaptive multi-agent framework designed to create immersive, interactive theatrical experiences using large language models (LLMs). Addressing limitations in existing drama generation methods, such as lack of initiative and poor physical scene interaction, HAMLET generates a narrative blueprint from simple topics to guide real-time improvisational performance. Each actor agent utilizes an adaptive reasoning module to make decisions based on personas, memories, and goals during complex group interactions. Crucially, the framework enables embodied interactions where agents manipulate scene props, such as opening letters or picking up weapons, thereby updating the global environmental context. To ensure quality, the team developed HAMLETJudge, a specialized critic model for automated evaluation. Experimental results indicate that HAMLET successfully produces expressive, coherent, and physically interactive theatrical performances autonomously. This advancement represents a significant step forward in interactive narrative fields, leveraging AI to enhance live performance immersion without requiring detailed manual input.
cs.AI updates on arXiv.org