Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Human Stress Response and Perceived Safety during Encounters with Quadruped Robots

Ryan Gupta, Hyonyoung Shin, Emily Norman, Keri K. Stephens, Nanshu Lu, Luis Sentis

TL;DR

The study investigates perceived safety during encounters with mobile quadruped robots by decoding acute stress from multimodal biosignals ($EDA$ and $HRV$) collected as Spot and Go1 autonomously navigate a realistic apartment with human participants. Using a synchronized data pipeline and a factorial experimental design, the authors show that stress increases with robot presence, is higher for multiple robots, and is greater during navigation than searching, with seating having no significant effect. They perform robust feature selection and LOSO cross-validated classification to distinguish encounter types, and provide an open-access dataset to foster further research. The work advances understanding of HRI safety in real-world-like settings and points toward online, adaptive robot behaviors informed by physiological states.

Abstract

Despite the rise of mobile robot deployments in home and work settings, perceived safety of users and bystanders is understudied in the human-robot interaction (HRI) literature. To address this, we present a study designed to identify elements of a human-robot encounter that correlate with observed stress response. Stress is a key component of perceived safety and is strongly associated with human physiological response. In this study a Boston Dynamics Spot and a Unitree Go1 navigate autonomously through a shared environment occupied by human participants wearing multimodal physiological sensors to track their electrocardiography (ECG) and electrodermal activity (EDA). The encounters are varied through several trials and participants self-rate their stress levels after each encounter. The study resulted in a multidimensional dataset archiving various objective and subjective aspects of a human-robot encounter, containing insights for understanding perceived safety in such encounters. To this end, acute stress responses were decoded from the human participants' ECG and EDA and compared across different human-robot encounter conditions. Statistical analysis of data indicate that on average (1) participants feel more stress during encounters compared to baselines, (2) participants feel more stress encountering multiple robots compared to a single robot and (3) participants stress increases during navigation behavior compared with search behavior.

Human Stress Response and Perceived Safety during Encounters with Quadruped Robots

TL;DR

The study investigates perceived safety during encounters with mobile quadruped robots by decoding acute stress from multimodal biosignals ( and ) collected as Spot and Go1 autonomously navigate a realistic apartment with human participants. Using a synchronized data pipeline and a factorial experimental design, the authors show that stress increases with robot presence, is higher for multiple robots, and is greater during navigation than searching, with seating having no significant effect. They perform robust feature selection and LOSO cross-validated classification to distinguish encounter types, and provide an open-access dataset to foster further research. The work advances understanding of HRI safety in real-world-like settings and points toward online, adaptive robot behaviors informed by physiological states.

Abstract

Despite the rise of mobile robot deployments in home and work settings, perceived safety of users and bystanders is understudied in the human-robot interaction (HRI) literature. To address this, we present a study designed to identify elements of a human-robot encounter that correlate with observed stress response. Stress is a key component of perceived safety and is strongly associated with human physiological response. In this study a Boston Dynamics Spot and a Unitree Go1 navigate autonomously through a shared environment occupied by human participants wearing multimodal physiological sensors to track their electrocardiography (ECG) and electrodermal activity (EDA). The encounters are varied through several trials and participants self-rate their stress levels after each encounter. The study resulted in a multidimensional dataset archiving various objective and subjective aspects of a human-robot encounter, containing insights for understanding perceived safety in such encounters. To this end, acute stress responses were decoded from the human participants' ECG and EDA and compared across different human-robot encounter conditions. Statistical analysis of data indicate that on average (1) participants feel more stress during encounters compared to baselines, (2) participants feel more stress encountering multiple robots compared to a single robot and (3) participants stress increases during navigation behavior compared with search behavior.
Paper Structure (16 sections, 8 figures, 4 tables)

This paper contains 16 sections, 8 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Stress Response and Comfort in HRI
  4. Stress Decoding from Physiological Signals
  5. Experiment Design
  6. Data Acquisition
  7. Conditions
  8. Experiment Procedure
  9. Data Analysis Methods
  10. Signal Processing and Feature Extraction
  11. Exploratory Data Analysis
  12. Classification of Human-Robot Encounters
  13. Results
  14. Discussion
  15. Conclusions
  16. ...and 1 more sections

Figures (8)

  • Figure 1: An overview of the experiments. The Unitree Go1 and Boston Dynamics Spot perform various mobile behaviors in shared spaces with UT community members. Physiological measures are collected from the E4 wristwatch (green) e4 and chest e-tattoo (cyan) bhattacharya2023chest.
  • Figure 2: Overview of an example experiment session designed to gain insights into human responses to encounters with robots. On the left are the list of 4 robot behaviors and 2 participant seating options. The order of each are randomized and then inserted into the experiment session as shown on the right side. The first step is receiving consent from the participants. The nature video is repeated before and after the experimental session as a relaxed baseline phsyiological state. After sensors are removed, participants take part in an interview.
  • Figure 3: A figure demonstrating the syncronized data acquisition architecture. The Spot and Go1 robots send state information via Robofleet sikand2021robofleet to an Intel NUC base station. Two omnidirectional microphones connect to a MacBook Pro (MBP) and Ubuntu (knapsack) machines. Finally, the two sensors that was participants wears connect via Bluetooth to designated Android and Windows machines. LabStreamingLayer lsl is used to push timestamped sensor data from all different sources to the network. LabRecorder labrecorder is used to record all streams into a single synchronized XDF file.
  • Figure 4: An overview of the three control variables. Namely, four robot behaviors and two participant seating settings. In (a) the social and isolated participant seating locations are shown with the single robot navigation path. (b) is the two robot navigation paths. In (c) is the single robot search path and (d) shows the two robot search paths.
  • Figure 5: Normalized features HRVCMSEn and nsEDRfreq over trials i.e. experiment time. Data shows averaged data from 17 subjects and their standard error. The trends show negative correlation as expected ($r = -0.8478$, see also Section \ref{['ssec:explore']}) and show a clear U-shape as values representing the baseline relaxed state are perturbed by the introduction of robots into the environment in trials 1 through 8 before returning to normal (see further analysis in Fig. \ref{['fig:principal_features']}).
  • ...and 3 more figures