Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Remote telepresence over large distances via robot avatars: case studies

Mohamed Elobaid, Stefano Dafarra, Ehsan Ranjbari, Giulio Romualdi, Tomohiro Chaki, Tomohiro Kawakami, Takahide Yoshiike, Daniele Pucci

TL;DR

The paper investigates enabling remote telepresence via robot avatars across intercontinental distances by adapting a flexible icub3_avatar-based architecture to diverse robot morphologies, including ErgoCub and a Honda-based wheeled avatar. It combines middleware bridging (YARP-ROS) with bandwidth-optimization techniques—such as glove retargeting at lower frequencies and hardware-accelerated video encoding—to sustain telepresence under constrained networks, demonstrated through ICRA 2023 remote participation and a Europe-wide EU Parliament visit, as well as an intercontinental Honda test between Italy and Japan. Key contributions include practical strategies for reducing bandwidth without sacrificing task execution, a versatile locomotion interface for varied robot bases, and insights into the trade-offs between latency, immersion, and social interaction in telexistence scenarios. The findings show that substantial bandwidth reductions ($>30$ Mbps to about $15.5$ Mbps, and even lower in some tests) can enable near-latency-free telepresence, guiding future work on social telepresence and latency-aware control enhancements.

Abstract

This paper discusses the necessary considerations and adjustments that allow a recently proposed avatar system architecture to be used with different robotic avatar morphologies (both wheeled and legged robots with various types of hands and kinematic structures) for the purpose of enabling remote (intercontinental) telepresence under communication bandwidth restrictions. The case studies reported involve robots using both position and torque control modes, independently of their software middleware.

Remote telepresence over large distances via robot avatars: case studies

TL;DR

The paper investigates enabling remote telepresence via robot avatars across intercontinental distances by adapting a flexible icub3_avatar-based architecture to diverse robot morphologies, including ErgoCub and a Honda-based wheeled avatar. It combines middleware bridging (YARP-ROS) with bandwidth-optimization techniques—such as glove retargeting at lower frequencies and hardware-accelerated video encoding—to sustain telepresence under constrained networks, demonstrated through ICRA 2023 remote participation and a Europe-wide EU Parliament visit, as well as an intercontinental Honda test between Italy and Japan. Key contributions include practical strategies for reducing bandwidth without sacrificing task execution, a versatile locomotion interface for varied robot bases, and insights into the trade-offs between latency, immersion, and social interaction in telexistence scenarios. The findings show that substantial bandwidth reductions ( Mbps to about Mbps, and even lower in some tests) can enable near-latency-free telepresence, guiding future work on social telepresence and latency-aware control enhancements.

Abstract

This paper discusses the necessary considerations and adjustments that allow a recently proposed avatar system architecture to be used with different robotic avatar morphologies (both wheeled and legged robots with various types of hands and kinematic structures) for the purpose of enabling remote (intercontinental) telepresence under communication bandwidth restrictions. The case studies reported involve robots using both position and torque control modes, independently of their software middleware.
Paper Structure (15 sections, 7 figures)

This paper contains 15 sections, 7 figures.

Table of Contents

  1. Introduction
  2. Participating in an International conference and visiting the EU parliament via a humanoid avatar
  3. Remote telepresence bandwidth requirements
  4. Participating in ICRA 2023 remotely
  5. Visiting the EU Parliament remotely
  6. On the teleoperation of a torque-controlled robot between Italy and Japan
  7. The Honda avatar robot
  8. The dexterous Honda hand
  9. The software architecture
  10. The telepresence scenario
  11. Results
  12. General considerations and discussion
  13. Comments on the locomotion interface
  14. Comments on the visual feedback
  15. Conclusion

Figures (7)

  • Figure 1: An example of the telepresence avatar system in action with a torque-controlled wheeled robot with a humanoid upper body and a kinematic structure different to that of iCub
  • Figure 2: The ergoCub robot interacting with other robots in London during ICRA 2023 while being teleoperated from Genoa.
  • Figure 3: During a talk in the European Parliament in Brussels, a remote operator pictured in the large screen is listening attentively through the ergoCub avatar.
  • Figure 4: The Honda avatar robot
  • Figure 5: Overview of the architecture enabling the teleoperation of the Honda avatar robot in Japan from Genoa
  • ...and 2 more figures