Exploring Collaborative Movement Improvisation Towards the Design of LuminAI—a Co-Creative AI Dance Partner

Title of the Paper

Exploring Collaborative Movement Improvisation Towards the Design of LuminAI — a Co-Creative AI Dance Partner

Paper Information

• Research Domain: Human-Computer Interaction (HCI), Dance Improvisation, Collaborative Creative Artificial Intelligence
• Keywords: Co-creativity, Collaborative Creative Agents, Dance Improvisation, Movement Improvisation, AI Agents, Computational Creativity

Research Background and Problems

• Identified Issues or Challenges:
  1. Human-computer interaction systems often lack collaborative creativity and fail to achieve real-time co-creation with humans.
  2. Existing research primarily focuses on computationally generated art or human-AI co-creation in music and painting, with limited exploration of embodied domains like dance.
  3. The design of AI capable of effectively participating in real-time physical collaboration remains underexplored.
• Significance of the Research:
  • Dance improvisation is a complex form of connection through movement, coordination, and creativity, which holds critical importance for artistic expression, cultural heritage, interaction research, and technology design.
  • Designing AI agents capable of engaging in physical improvisational collaboration can deepen our understanding of human-machine interaction in complex embodied environments.
• Motivation and Related Work:
  • This study is inspired by the previously developed LuminAI prototype (an AI system generating dance movements based on improvisation theory) and aims to explore new possibilities for improving and optimizing this technology through collaborative research with professional dancers.
  • Insights from participatory cognition theories (e.g., "participatory sense-making") on improvisational movement frame the motivation and theoretical background of this study.

Solution

• Proposed Method:
  • Conducting focus group studies with 24 dance students to explore their experiences in movement improvisation, using thematic analysis to summarize key patterns and behaviors in collaborative movement improvisation.
  • Proposing an "Interrelated Model of Improvisational Dance Input" and providing design recommendations for the AI dance partner LuminAI.
• Innovative Contributions:
  • Introducing a dancer-centered human-computer interaction design framework, systematically analyzing factors influencing dancers' decision-making, generation strategies, and guiding principles in collaborative improvisation.
  • Incorporating movement, creativity, and social improvisation models into AI system design to build highly participatory and dynamically responsive collaborative AI agents.
• Key Technologies and Steps:
  1. Data Collection: Gathering insights from focus group discussions and dance improvisation tasks to understand dancers' movement choices during interaction.
  2. Thematic Analysis: Using participant feedback to summarize and conceptualize three main components:
     • Immediate influences (from self, partners, and the environment).
     • Generation strategies (e.g., imitation, repetition, transformation, amplification).
     • Collaborative heuristic principles (e.g., partner awareness, emphasis on connection).
  3. System Design Recommendations: Integrating research themes with LuminAI functionality to provide directions for improvement.

Research Outcomes

• Specific Results:
  1. Model:
     • Proposed the "Interrelated Model of Improvisational Dance Input," encompassing three dimensions: "immediate influences, self-generation strategies, and collaborative principles."
  2. Behavioral Insights:
     • The importance of bodily perception and emotions in improvisational creation.
     • The guiding role of partner and environmental interaction in creative behaviors.
     • Common generation strategies used by dancers in improvisational creation (e.g., imitation, mirroring, amplification).
  3. Design Recommendations:
     • Presented a set of 10 key design recommendations centered on LuminAI (e.g., perceiving human emotions, utilizing multi-sensory inputs, dynamic competitive strategies).
• Advantages:
  • The model provides theoretical support for enhancing LuminAI's collaborative capabilities.
  • Compared to existing systems that only generate movements, this model better supports dynamic and meaning-making interactions between humans and machines.
• Experimental Evaluation:
  • Provided data from focus group studies and dance improvisation tasks, constructing the theory through semi-structured questionnaires and mirroring experiments.
• Limitations and Future Directions:
  1. Limitations:
     • The sample is limited to 24 dance students from one university, lacking diversity in dance styles and cultural backgrounds.
     • Practical functionality testing of LuminAI based on the design recommendations has not yet been conducted.
  2. Future Directions:
     • Expanding the participant pool to include diverse cultural backgrounds and professional levels.
     • Further exploring collaboration between AI and non-normative bodies (e.g., users with assistive devices).
     • Enhancing LuminAI's real-time responsiveness and perceptual complexity to validate the model's applicability in real-world interactions.

Output Format

• Based on the analysis of the paper, the results primarily focus on thematic insights and design implications, integrating theoretical models with human-centered technological design.