HR1 Robot: An Assistant for Healthcare Applications

Valentina Vasco, Alexandre G. P. Antunes, Vadim Tikhanoff, Ugo Pattacini, Lorenzo Natale, Valerio Gower, Marco MaggialiFrontiers Robotics AI 2022 · DOI · OpenReview

Introduction

As global populations age, healthcare systems face increasing pressure to provide efficient mobility and balance assessments. The Timed Up and Go (TUG) test is a widely used clinical tool for evaluating functional mobility. In this test, a patient is required to stand from a chair, walk 3 meters, turn 180 degrees, walk back, and sit down.

The classical Timed Up and Go (TUG) test procedure.

Currently, the TUG test relies on manual timing and subjective observation by clinicians, which can lead to variability and human error. Researchers from the Istituto Italiano di Tecnologia (IIT) have developed an autonomous robotic architecture for the R1 humanoid robot to standardize this assessment, providing precise kinematic data while acting as a social assistant.

Methodology

The R1 humanoid robot serves as the platform for this autonomous "test administrator." Its software architecture, built on the YARP (Yet Another Robot Platform) framework, integrates several key modules:

  • 3D Skeleton Acquisition: Utilizing an Intel RealSense RGB-D camera and the yarpOpenPose library, the robot detects 2D keypoints and projects them into 3D space to track patient movement.
  • Reactive Navigation: To ensure the patient remains within the camera's field of view, the robot employs a "bang-bang" controller that maintains a fixed distance, moving dynamically as the patient walks.
  • Motion Analysis: A specialized module calculates clinical parameters such as step length, walking speed, and specific TUG events (e.g., standing up or crossing the line).
  • Speech Interaction: The robot uses Google’s Speech-to-Text and Text-to-Speech APIs to explain the test, answer questions, and provide feedback.

The system was validated through 3,000 trials in the Gazebo simulation environment, using a programmed "Actor" to simulate various walking speeds and distances.

The TUG test environment implemented in the Gazebo simulator.

Key Insights

A significant technical challenge addressed in this work is the accurate detection of the 3-meter finish line. The researchers found that tracking the "hip center" was insufficient due to the patient's physical volume. By implementing a bounding box around the torso, they significantly reduced false negatives. Additionally, the study reinforces the value of Socially Assistive Robots (SARs), noting that patients show higher engagement with a physical robot than with virtual agents.

Results and Performance

The simulation results demonstrated that the robotic architecture is highly reliable for clinical use:

  • Timing Accuracy: The robot measured completion time with an average error of 0.47 ± 0.06 seconds.
  • Step Counting: Step estimation errors ranged from 0.64 to 1.11 steps, with most errors occurring during the complex 180-degree turn phase.
  • Pass/Fail Detection: The system achieved a 92.3% accuracy in determining if the patient successfully crossed the 3m line, with a 100% recall rate.

Limitations and Future Outlook

While the simulation results are promising, several challenges remain for real-world deployment. At walking speeds above 1.1 m/s, navigation lag can cause the patient to exit the camera's field of view. Furthermore, the 180-degree turn remains a difficult phase for skeleton tracking due to occlusion. Future work will involve testing the R1 platform in hospital settings with human subjects to evaluate performance against real-world variables like lighting and clothing.

Summary updated

Abstract

p>According to the World Health Organization<xref ref-type="fn" rid="fn1"><sup>1,</sup></xref><xref ref-type="fn" rid="fn2"><sup>2</sup></xref> the percentage of healthcare dependent population, such as elderly and people with disabilities, among others, will increase over the next years. This trend will put a strain on the health and social systems of most countries. The adoption of robots could assist these health systems in responding to this increased demand, particularly in high intensity and repetitive tasks. In a previous work, we compared a Socially Assistive Robot (SAR) with a Virtual Agent (VA) during the execution of a rehabilitation task. The SAR consisted of a humanoid R1 robot, while the Virtual Agent represented its simulated counter-part. In both cases, the agents evaluated the participants’ motions and provided verbal feedback. Participants reported higher levels of engagement when training with the SAR. Given that the architecture has been proven to be successful for a rehabilitation task, other sets of repetitive tasks could also take advantage of the platform, such as clinical tests. A commonly performed clinical trial is the Timed Up and Go (TUG), where the patient has to stand up, walk 3 m to a goal line and back, and sit down. To handle this test, we extended the architecture to evaluate lower limbs’ motions, follow the participants while continuously interacting with them, and verify that the test is completed successfully. We implemented the scenario in Gazebo, by simulating both participants and the interaction with the robot<xref ref-type="fn" rid="fn3"><sup>3</sup></xref>. A full interactive report is created when the test is over, providing the extracted information to the specialist. We validate the architecture in three different experiments, each with 1,000 trials, using the Gazebo simulation. These experiments evaluate the ability of this architecture to analyse the patient, verify if they are able to complete the TUG test, and the accuracy of the measurements obtained during the test. This work provides the foundations towards more thorough clinical experiments with a large number of participants with a physical platform in the future. The software is publicly available in the assistive-rehab repository<xref ref-type="fn" rid="fn4"><sup>4</sup></xref> and fully documented.</p>
BibTeX
@article{vasco2022hr,
  author = {Valentina Vasco and Alexandre G. P. Antunes and Vadim Tikhanoff and Ugo Pattacini and Lorenzo Natale and Valerio Gower and Marco Maggiali},
  title = {{HR1 Robot: An Assistant for Healthcare Applications}},
  journal = {Frontiers Robotics AI},
  year = {2022},
  doi = {10.3389/frobt.2022.813843},
}
Figure