Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Interactive Augmented Reality Interface for Personalized Proxemics Modeling

Massimiliano Nigro, Amy O'Connell, Thomas Groechel, Anna-Maria Velentza, Maja Matarić

TL;DR

This paper tackles personalized proxemics modeling for older adults in human-robot interaction by combining an AR-based data collection interface with an active transfer learning approach. The authors frame proxemics prediction as a domain adaptation problem, using SocNav1 as the training domain and a per-user application domain, and implement an ordinal regression neural network with smoothing. Through two IRB-approved user studies, they show that fine-tuning with ATL data reduces testing error by about 26.97% on average and that AR data approximates physical robot interactions while older adults report positive usability and trust. However, RS sampling yielded better performance than ATL in their tests due to greater angular sampling diversity; findings inform AR-based personalization pipelines for socially assistive robotics.

Abstract

Understanding and respecting personal space preferences is essential for socially assistive robots designed for older adult users. This work introduces and evaluates a novel personalized context-aware method for modeling users' proxemics preferences during human-robot interactions. Using an interactive augmented reality interface, we collected a set of user-preferred distances from the robot and employed an active transfer learning approach to fine-tune a specialized deep learning model. We evaluated this approach through two user studies: 1) a convenience population study (N = 24) to validate the efficacy of the active transfer learning approach; and 2) a user study involving older adults (N = 15) to assess the system's usability. We compared the data collected with the augmented reality interface and with the physical robot to examine the relationship between proxemics preferences for a virtual robot versus a physically embodied robot. We found that fine-tuning significantly improved model performance: on average, the error in testing decreased by 26.97% after fine-tuning. The system was well-received by older adult participants, who provided valuable feedback and suggestions for future work.

An Interactive Augmented Reality Interface for Personalized Proxemics Modeling

TL;DR

This paper tackles personalized proxemics modeling for older adults in human-robot interaction by combining an AR-based data collection interface with an active transfer learning approach. The authors frame proxemics prediction as a domain adaptation problem, using SocNav1 as the training domain and a per-user application domain, and implement an ordinal regression neural network with smoothing. Through two IRB-approved user studies, they show that fine-tuning with ATL data reduces testing error by about 26.97% on average and that AR data approximates physical robot interactions while older adults report positive usability and trust. However, RS sampling yielded better performance than ATL in their tests due to greater angular sampling diversity; findings inform AR-based personalization pipelines for socially assistive robotics.

Abstract

Understanding and respecting personal space preferences is essential for socially assistive robots designed for older adult users. This work introduces and evaluates a novel personalized context-aware method for modeling users' proxemics preferences during human-robot interactions. Using an interactive augmented reality interface, we collected a set of user-preferred distances from the robot and employed an active transfer learning approach to fine-tune a specialized deep learning model. We evaluated this approach through two user studies: 1) a convenience population study (N = 24) to validate the efficacy of the active transfer learning approach; and 2) a user study involving older adults (N = 15) to assess the system's usability. We compared the data collected with the augmented reality interface and with the physical robot to examine the relationship between proxemics preferences for a virtual robot versus a physically embodied robot. We found that fine-tuning significantly improved model performance: on average, the error in testing decreased by 26.97% after fine-tuning. The system was well-received by older adult participants, who provided valuable feedback and suggestions for future work.
Paper Structure (25 sections, 9 figures, 1 table)

This paper contains 25 sections, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Technical Approach
  3. Problem Formulation
  4. Proxemics Model
  5. System Overview
  6. AR Application
  7. ATL Sampler
  8. Evaluation With User Studies
  9. Study 1: Effectiveness Study
  10. Study Setup
  11. Participants
  12. Study Procedure
  13. Study 2: Older Adult User Study
  14. Study Setup
  15. Participants
  16. ...and 10 more sections

Figures (9)

  • Figure 1: Older adults have a variety of proxemics preferences
  • Figure 2: Study participants used an AR interface to indicate when the approaching virtual robot was at a comfortable distance
  • Figure 3: Example of features considered in this work
  • Figure 4: System Architecture Diagram
  • Figure 5: Effectiveness study setup: left) a drawing of the environment and the robot approach lines, right) a view of the experiment setup
  • ...and 4 more figures