Papers
Topics
Authors
Recent
Search
2000 character limit reached

Cybernetic Android Avatar "Yui": System Integration, Field Deployment, and Evaluation

Published 6 Jun 2026 in cs.RO | (2606.08099v1)

Abstract: Remote communication technologies have become widely used; however, supporting a sense of shared physical space and conveying rich non-verbal cues remain challenging in many social interaction scenarios. This study presents "Yui," a full-body cybernetic android avatar designed to integrate operator-side immersive teleoperation with interlocutor-side human-like social signaling. Yui combines a 55-degrees of freedom full-body mechanism with a previously developed android head, facial expression and gaze control, upper-body and arm motion, hand actuation, and a mobile platform. It can be operated through either the immersive mode using a head mounted display-based interface or desktop mode using a webcam-based interface. We evaluated the system through three real-world deployments: a long-term public exhibition at Expo 2025 in Osaka, Kansai, Japan; a remote educational exchange between elementary school students; and a public interaction study with general participants. During the Expo deployment, two units accumulated approximately 1131 h of operation, demonstrating both operational feasibility and maintenance challenges. In the public study, both operators and interlocutors reported positive impressions of co-presence and willingness to use the system. Interlocutors also rated the avatar positively in terms of human likeness and the transmission of emotions and intentions. The results indicate usability for general operators while suggesting room for improvement in precise controllability. These findings provide field-derived evidence and design implications for socially deployable full-body android avatars.

Summary

  • The paper presents a fully integrated android avatar with 55-DOF that enables immersive, bi-directional telepresence via advanced hardware and low-latency interfaces.
  • The paper validates Yui's technical robustness through extensive field deployments, achieving over 1100 public exhibition hours with high social acceptance.
  • The paper outlines design guidelines for modular hardware, hybrid autonomy, and versatile operator interfaces that inform future telepresence and social robot applications.

Cybernetic Android Avatar "Yui": System Integration, Field Deployment, and Evaluation

Introduction

The paper "Cybernetic Android Avatar 'Yui': System Integration, Field Deployment, and Evaluation" (2606.08099) details the design, real-world deployment, and multifaceted evaluation of a full-body android avatar system, Yui, aimed at bridging the embodiment gap in telepresence by enabling bi-directional immersive interaction. Yui's architecture delivers both high-fidelity social signaling to interlocutors and first-person embodiment to operators, integrating advances in hardware and teleoperation interfaces. The evaluation utilizes extensive field deployments, from long-term public exhibitions to educational and public interaction studies, providing comprehensive evidence for both technical robustness and social acceptance in unconstrained, ecological settings.

System Design and Integration

Hardware Architecture

Yui incorporates a 55-DOF kinematic structure, integrating a modular android head capable of high-resolution facial expression synthesis (22 actuation points with 15 motors), articulated torso, arms, hands, wrists, and a mobile platform with vertical posture adjustment. Key subsystems employ DC motors or BLDC actuators with cycloidal and planetary reduction mechanisms to minimize acoustic emission and enhance motion fidelity. The hand design utilizes tendon-driven mechanisms for fine-grained finger control, and passive spring-based extensions. The electrical system employs a single LiFePO4 battery (24V, 60Ah) for simplified field operation, supporting uninterrupted exhibition with hot-swappable units and in-place charging.

Sensory and Communication Subsystems

Yui is equipped with embedded stereo microphones, wide-angle eye cameras for first-person visual streaming, and distributed joint sensing for closed-loop motion control. Audio output utilizes both upper-body and base-mounted speakers for environmental adaptability. Major actuators are governed via custom CAN or RS485-based control boards, managed by an onboard high-performance PC networked via Wi-Fi.

Teleoperation Interface

The system supports two operation modes:

  • Immersive (HMD-based): Employs commercial HMDs with integrated eye, face, and hand tracking, mapping high-dimensional operator kinematics (facial blendshapes, gaze, hand pose, head/body orientation) directly onto the avatar.
  • Desktop Mode: Uses webcam-based tracking (MediaPipe), allowing operation by children or users unable to utilize HMDs, with head and facial motion mapped but body motion fixed.

Low-latency bidirectional communication is achieved using WebRTC-based SDKs with custom mapping logic for operator-to-avatar synchronization, real-time voice transformation, and vision-audio streaming. Measured end-to-end latencies are within acceptable regimes for real-time social interaction (motion lag 0.38–0.87s, AV latency 0.34–1.12s).

Field Deployments and Empirical Evaluation

Long-term Public Exposure: Expo 2025 (Study 1)

Yui was deployed for over six months (1131 cumulative hours) at Expo 2025 in Osaka, alternating two units for field robustness. The exhibition setting imposed environmental stressors (noise, weather) and high interaction throughput. Operational modes alternated between autonomous playback and live teleoperation. The system demonstrated continuous technical viability, with rapid maintenance cycles managing failures (motor, battery), and high social engagement. Visitors exhibited naturalistic interaction patterns, photograph collection, and spontaneous conversational engagement even in unscripted interactions.

Remote Educational Exchange (Study 2)

Yui mediated remote interactions between 8–10-year-old students across Tokyo and Kanazawa, with desktop mode facilitating operation by children. Participants quickly achieved engagement, executing structured and free-form social dialogue. Teachers reported high perceived embodiment and novelty—however, user confusion was present for features unavailable in desktop mode (e.g., body motion), and immersion was recognized as suboptimal relative to HMD-based interface. Despite these, the system facilitated robust two-way remote communication in a genuine educational context.

Public Interaction Study (Study 3)

A quantitative, in situ evaluation was conducted with 108 adult operators/interlocutors and a follow-up with 16 interlocutors for operator proficiency effects. Experiment 1 paired random participants for two-minute interactions, collecting subjective impressions using a structured questionnaire (seven-point Likert and VAS scales).

Key results include:

  • 69.2–87.8% positive ratings for sense of co-presence, surpassing typical scores for videoconferencing.
  • 90.4% (operators) and 83.7% (interlocutors) would use the avatar-based system in lieu of traditional remote communication.
  • 91.8% of interlocutors rated the avatar as human-like, and 89.8% found emotion/intention transmission effective.
  • Operability was positively rated, though with caveats: only 73.1% felt complete control; median physical workload was low (VAS=17.5/100).
  • Interactions with previously unfamiliar partners yielded higher peak co-presence and affect transmission scores, contradicting expected in-group bias.
  • In Experiment 2 (operation by trained experts): all items (co-presence, willingness to use, human-likeness, emotion transfer) had ≥75% positive responses, and zero negative-outlier reports.

No significant correlation was observed between participants’ prior video call frequency and subjective comparative ratings, suggesting the observed effects are independent of digital communication familiarity.

Implications and Theoretical Significance

Yui empirically validates that full-body android avatars, when endowed with high-DOF, human-like actuation and immersive teleoperation interfaces, are technically robust for extended public deployment and effective at restoring non-verbal social cues. Subjective satisfaction in co-presence, emulation of affective signaling, and system acceptance was consistently high across diverse participant pools and contexts. The strong performance for first-time pairings and among child operators highlights potential for avatars in educational, interpersonal, and cross-cultural settings.

Contrary to many task-centric teleoperated platforms (iCub3, NimbRo-Avatar), Yui foregrounds social signaling and mutual embodiment. These results suggest that physical embodiment, even when decoupled from task-efficiency considerations, enhances affect transmission, social grounding, and trust, consistent with the tenets of media richness and telepresence theory.

Further, the study identifies architectural and operational guidelines:

  • Subsystem modularity is essential for field maintainability.
  • Hybrid autonomy (scripted/teleop switching) is required for sustainable real-world deployment.
  • Interface versatility broadens accessibility (e.g., for minors or non-experts), but immersion and controllability scale with operator proficiency and interface fidelity.

Limitations include the absence of direct comparative trials with alternative embodied/virtual agents and the lack of fine-grained, objective behavioral analytics. Potential bias in public participant recruitment (technology enthusiasts) is acknowledged, though broad age/gender ranges mitigate this risk.

Future Directions

Scaling Yui to other application domains will necessitate:

  • Comparative studies isolating the effect size of full-body embodiment versus partial or screen-based avatars.
  • Further reduction and harmonization of cross-modal latency to optimize synchrony of facial/body expressions and dialogue prosody.
  • Establishing operational and privacy frameworks for public deployments, particularly in mixed-age or sensitive environments.
  • Expansion of semi-autonomous behaviors to further reduce operator load without compromising social presence.

Conclusion

The Yui system advances the state of socially deployable android avatars by establishing empirical viability, technical robustness, and social acceptance in unconstrained environments. The outcomes underscore the critical importance of full-body, high-fidelity physical embodiment and immersive teleoperation for achieving effective non-verbal communication and co-presence remotely. This foundational work enables rigorous future investigations into the comparative utility of android telepresence, informs best practices for social robot design, and sets the stage for the broad deployment of teleoperational avatars across public, educational, and interpersonal domains.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Collections

Sign up for free to add this paper to one or more collections.