A Self-Organizing Interaction and Synchronization Method between a Wearable Device and Mobile Robot

Min-Su Kim, Jae Geun Lee, Soon-Ju Kang2016 · DOI

TL;DR

This paper introduces "Pet-Bot," a mobile robot system that autonomously follows or leads a human without complex interfaces. It achieves this through a self-organizing interaction between a wearable device (smart belt and watch) and the robot, using low-frequency (LF) signals for master identification and Bluetooth Low Energy (BLE) for motion data exchange. The system demonstrates high accuracy in identifying users and synchronizing with their movements, paving the way for more intuitive human-robot collaboration.

What Problem Does This Paper Address?

The increasing demand for human-support robots, particularly those that can accompany users, is often hindered by the need for specific, complex control interfaces. Existing solutions like image processing require high computational power and struggle in crowded environments, while sensor network methods are spatially constrained. This paper addresses the challenge of creating a mobile robot that can seamlessly follow or lead a human, offering carriage services without requiring the user to constantly manage its operation. The goal is to enable robots to act more like intuitive companions, similar to pets, by self-organizing their interaction and synchronization with human movements.

Methods

The core of the Pet-Bot system lies in its self-organizing interaction and synchronization method, which combines a wearable device (smart watch and belt) with a mobile robot.

  • Master Identification: The robot uses low-frequency (LF) signals with a unique "wake-up pattern" to identify its authorized master within a 1-meter range. This is a key differentiator, as it avoids the need for complex image processing or pre-configuration.
  • Location Tracking: The robot determines the user's location using the Received Signal Strength Indication (RSSI) of the LF signal. By analyzing the RSSI from two LF transmitter antennas, the system calculates the user's distance and direction relative to the robot.
  • Motion Recognition: The wearable device's Inertial Measurement Unit (IMU), specifically a 3D accelerometer and gyro sensor, captures the user's motion. The system classifies movements into states like "Run," "Walk," "Rest," and "Stop" based on accelerometer data periodicity and Signal Magnitude Area (SMA). Rotation information is also captured to mimic the user's turns.
  • Communication: Two wireless communication protocols are employed:
  • LF (125 kHz): Used for initial master identification and proximity sensing.
  • Bluetooth Low Energy (BLE): Used for real-time exchange of motion and location data between the wearable device and the mobile robot.
  • Control System: The robot's movement is controlled by a system that considers user location, motion (movement and rotation), and desired operation mode (leading or following). It utilizes adaptive cruise control (ACC) principles for distance management and specific controllers for direction correction and synchronization. Mecanum wheels allow for omnidirectional movement, enabling smooth turns and lateral shifts.
  • Exception Handling: A gesture recognition function on the smart watch allows the user to provide commands in uncertain situations or to override the automatic modes.

Datasets

Not discussed in the paper. The paper focuses on experimental validation using a prototype system rather than training on large datasets.

Key Insights and Ideas

  • LF for Intuitive Master Identification: Using LF signals with unique wake-up patterns for master identification is a clever way to achieve quick and accurate authorization without complex vision systems, making the robot accessible even in crowded or low-light conditions.
  • RSSI for Proximity-Based Localization: Leveraging the RSSI of LF signals for location estimation, especially in conjunction with a simple triangle condition, provides a cost-effective and computationally light method for tracking the user's position.
  • Wearable IMU for Rich Motion Understanding: The integration of IMU data from a smart belt allows for a nuanced understanding of user movement (walking, running, stopping) and rotation, enabling the robot to synchronize its actions effectively.
  • Hybrid Communication for Efficiency: The combination of LF for initial contact and BLE for continuous data exchange strikes a good balance between low-power, robust identification and high-bandwidth, real-time motion data transfer.
  • "Pet-Bot" Analogy: The framing of the robot as a "Pet-Bot" effectively communicates its intended behavior – an intuitive, companion-like assistant that anticipates and responds to user needs without explicit commands.

Key Results

The paper reports promising results from experiments conducted with a prototype Pet-Bot and wearable devices:

  • Motion Recognition Accuracy: The system achieved a high level of precision in recognizing user movements and synchronization, with accuracy rates over 94% in user-motion recognition tests.
  • Location Identification Accuracy: The location identification test showed a root mean square error (RMSE) of 5.10 cm for distance and 6.52° for direction.
  • Synchronization Performance: The user-motion recognition and synchronization test demonstrated an average accuracy of over 94% across various target rotation angles.
  • Trajectory Tests: The trajectory tests confirmed the robot's ability to effectively follow and lead humans, adapting to straight movements, turns, and even mode changes (from following to leading) with stable control.

Limitations and Future Work

  • LF Signal Distortion: The authors acknowledge that LF signal distortion can worsen outside the defined "reliable region" (approximately 1 meter distance and ±25° angle). This limitation can lead to a loss of synchronization or missed user detection, especially during sudden movements.
  • Reliability Against Failures: Improving the Pet-Bot's reliability against failures caused by distorted LF signals is identified as a future research direction.
  • Swarm Robotics: The paper suggests that a swarm or combination of robots could provide more advanced carriage services, hinting at future work in multi-robot coordination.
  • Advanced Carriage Services: The system's applicability to regions requiring more sophisticated carriage services is envisioned, implying potential for expanded functionalities.

Summary updated

BibTeX
@misc{kim2016self,
  author = {Min-Su Kim and Jae Geun Lee and Soon-Ju Kang},
  title = {{A Self-Organizing Interaction and Synchronization Method between a Wearable Device and Mobile Robot}},
  year = {2016},
  doi = {10.3390/s16060842},
}
Figure