MarioChart: Autonomous Tangibles as Active Proxy Interfaces for Embodied Casual Data Exploration

Paper Title

MarioChart: Autonomous Tangibles as Active Proxy Interfaces for Embodied Casual Data Exploration

Publication Info

• Topic area: Tangible user interfaces for situated data visualization
• Keywords: Tangible user interfaces, active proxies, situated visualization, data exploration, spatial memory, autonomous tangibles, embodied interaction, user study, data analytics, human-computer interaction

Background and Problem

• Problem / challenge: Existing tangible user interfaces (TUIs) for situated visualization lack empirical evaluation of the effects of active manipulation of tangible proxies on task performance. Prior systems like Uplift required manual repositioning of proxies, limiting usability.
• Significance: Understanding the benefits and drawbacks of active proxies can inform the design of intuitive, engaging systems for casual data exploration, with potential applications in education, public engagement, and collaborative analytics.
• Motivation and related work: Prior work demonstrated TUIs' cognitive and collaborative advantages but focused on static proxies or graphical interfaces. Situated visualization with tangible proxies has been underexplored, particularly in scenarios requiring active manipulation of referents.

Solution

• Proposed approach: The MarioChart system, an active proxy interface combining autonomous tangible proxies with a tabletop display and dashboard for situated data visualization.
• Novelty:
  1. Introduction of a conceptual design space for situated visualization, defined by axes of passive vs. active proxies and spatial vs. abstract representations.
  2. Development of the MarioChart system, featuring autonomous tangible proxies that self-relocate to their correct positions on a map.
  3. Empirical evaluation of active proxies compared to a graphical tablet interface, focusing on task performance, memory, and user experience.
• Procedure and key techniques:
  1. Tangible proxies (scale models of buildings) are tracked in six degrees of freedom and autonomously repositioned using desktop robots.
  2. Interaction design includes filtering, comparison, and drill-down tasks via physical manipulation of proxies.
  3. User study (n=12) compared MarioChart with a tablet interface across four task types, measuring performance, recall, workload, and engagement.

Results

• Concrete findings:
  • Active proxies improved short-term spatial memory (mean score: 0.333 vs. 0.083 for tablet, Cohen’s d = 0.746).
  • Faster completion of referent-data tasks with active proxies (72.40s vs. 95.56s for tablet, Cohen’s d = -0.538).
  • No significant differences in long-term memory, physical fatigue, mental workload, or user engagement between interfaces.
• Advantage over baselines: Active proxies provided better spatial cognition and faster access to referents in specific tasks, though overall performance was comparable to the tablet interface.
• Experiments / evaluation:
  • Tasks: Bookmarking, data-referent matching, referent-data matching, and spatial data-referent matching.
  • Metrics: Completion time, correctness, recall scores, Borg CR10 (physical load), PAAS (mental load), and user engagement scale.
  • Dataset: Campus sustainability data (electricity, emissions, water consumption).
• Limitations and future work:
  • Limited to five tangible proxies due to hardware constraints.
  • Camera-based tracking imposed holding constraints.
  • Small sample size (n=12).
  • Future work should explore larger proxy sets, alternative tracking methods, and tasks emphasizing referent shape recognition.

Summary

This paper introduces MarioChart, an active proxy interface for situated data visualization, combining autonomous tangible proxies with a tabletop map and dashboard. The system enhances short-term spatial memory and task performance in referent-data matching compared to a tablet interface, though no significant differences were found in long-term memory, workload, or engagement. The study provides a baseline for active proxy systems and highlights their potential for casual data exploration. Future research should address scalability, tracking constraints, and tasks emphasizing referent shapes.