Towards Fluent Interaction with Cyber-Physical Architecture

Paper Title

Towards Fluent Interaction with Cyber-Physical Architecture

Publication Info

• Topic area: Human-robot interaction in shape-changing architectural environments.
• Keywords: Cyber-physical architecture, shape-changing environments, human-robot interaction, multimodal interaction, speculative design, elicitation study, user intent modeling, adaptive environments, privacy, personalization.

Background and Problem

• Problem / challenge: Current research on shape-changing environments lacks a deep understanding of user interaction techniques and the values that make these systems desirable. Prior work has focused more on technical feasibility than on practical and relational user needs.
• Significance: Shape-changing environments have the potential to transform living and working spaces, offering adaptive solutions for urbanization, accessibility, and resource optimization. Understanding how users interact with these systems is critical for designing environments that are both functional and trusted.
• Motivation and related work: Previous studies explored shape-changing interfaces but were limited to explicit reconfiguration techniques and lacked insights into multimodal, naturalistic interactions. This paper builds on foundational work in tangible computing, intelligent environments, and multimodal interaction, addressing gaps in user-centric design for cyber-physical architecture.

Solution

• Proposed approach: The authors conducted two complementary studies: speculative design workshops to uncover aspirational values and a Wizard-of-Oz elicitation study to examine practical interaction techniques with a shape-changing robotic wall.
• Novelty:
  1. Identification of eleven design values for shape-changing environments, highlighting tensions between automation and user autonomy, and personalization versus public ownership.
  2. Empirical characterization of multimodal interaction strategies, introducing a modal entropy metric to quantify user behavior diversity.
  3. Proposal for a modality-agnostic, dynamic model of user intent to enable more fluent interaction with robotic environments.
• Procedure and key techniques:
  1. Speculative design workshops (N=20): Participants ideated and storyboarded scenarios for robotic walls, revealing functional and relational values.
  2. Wizard-of-Oz elicitation study (N=12): Participants performed tasks with a robotic wall controlled by a human operator, exploring multimodal interaction strategies (voice, gesture, touch).
  3. Data analysis: Qualitative coding of workshop outputs and video recordings, semantic embedding of user actions, and entropy-based metrics for interaction diversity.

Results

• Concrete findings:
  • Eleven design values emerged, spanning functional (e.g., resource optimization, privacy) and relational (e.g., trust, engagement) dimensions.
  • Multimodal interaction was prevalent, with 8 out of 12 participants adopting mixed strategies (voice, gesture, touch) by the end of the study.
  • Modal entropy metrics revealed diverse interaction styles, with some users favoring single modalities and others blending multiple channels.
• Advantage over baselines: The study demonstrated that multimodal strategies and implicit context interpretation are essential for natural interaction, surpassing the limitations of prior explicit-command systems.
• Experiments / evaluation:
  • Workshops: 20 participants (designers/architects) created scenarios and discussed values.
  • Elicitation study: 12 participants performed block-placement tasks with a robotic wall, under free-choice and forced-choice modality conditions.
  • Metrics: Modal entropy quantified interaction diversity; semantic embeddings visualized user actions.
• Limitations and future work:
  • Limited participant diversity (e.g., age, abilities) restricts generalizability.
  • Laboratory setting may not fully capture real-world interaction dynamics.
  • Future work includes co-design with disabled users, exploration of long-term adaptation, and development of load-bearing, malleable surfaces.

Summary

This paper explores human interaction with shape-changing architectural environments through speculative design workshops and a Wizard-of-Oz elicitation study. It identifies eleven design values, highlights the importance of multimodal interaction, and introduces modal entropy as a metric for user behavior diversity. The findings suggest that robotic environments must infer user intent from implicit context to achieve natural and fluent interaction. Future research directions include co-design for accessibility, load-bearing structures, and intent modeling for adaptive environments. These insights contribute to the design of cyber-physical systems that are collaborative, responsive, and trusted partners in everyday life.