1 of 36

Human Perception on Displayed Expressions and Movement

By: The Ford F-150 Group

Cara Blashill, Madel Humbert, Ishrat Khan, Carol Wang

2 of 36

Agenda

  • Project Introduction
  • Related Work
  • Methods
  • Results & Discussion
  • Conclusions & Future Work
  • Q & A

3 of 36

Introduction

Goals and Research Questions

01

4 of 36

Context

As robotics get more integrated into daily life, humans are more bound to interact with these systems. Therefore, it is important to understand how a robot’s non-verbal communication methods will be perceived by the humans during an interaction.

5 of 36

Goals

Displayed emotions & their effect on perception of a robot’s movement and intention

How understanding a robot’s intentions can improve comfortability and trust in HRI

6 of 36

Research Questions

How does displayed emotion affect the perception of movement?

How does the perceived movement of the robot affect comfortability and trust in the robot to clean a space?

7 of 36

Constructs

Perception of movement

Comfortability

Trust

8 of 36

Hypotheses

The perceived movement of a robot, influenced by its facial expressions, will affect human comfort and trust in its ability to complete a task

A robot’s displayed facial expressions will influence human perception of its movement, such that the same movement path will be interpreted differently depending on the associated emotion

9 of 36

Related Work

Work from Dietz, Xie, Mavrogiannis and others

02

10 of 36

Human Perception of Swarm Robot Motion [1]

​

​

G. Dietz, J. L. E, P. Washington, L. H. Kim, S. Follmer

Takeaway

Results

Hypothesis

Study

Measure swarm speed, synchronization and smoothness to vary valence, arousal and dominance

​

​

Speed correlates with arousal, and smoothness correlates with valence

​

High speed and smoothness produce more positive emotional response

Learn how to vary robot paths to product a specific response

11 of 36

How Do Humans Recognize the Motion Arousal of Non-Humanoid Robots [2]

​

Q. Xie, Z. Chen, D. Luh

Takeaway

Results

Hypothesis

Study

Tested different robot jump heights in relation to perceived arousal levels

Different robot altitudes will correlate to higher arousal

Hypothesis is true, and position affects perception

Learn more about perception of non-humanoid robots

12 of 36

Influencing Behavioral Attributions to Robot Motion During Task Execution [3]

N. Walker, C. Mavrogiannis, S. Srinivasa, M. Cakmak

Takeaway

Results

Hypothesis

Study

Apply curiosity and competence attributes to a robot with planned trajectories

Path generator will successfully apply curiosity and competence attributes

Participants were able to select the coded attribute

Learn more about trajectory implying emotions/attributes

13 of 36

Methods

Study background and protocol

03

14 of 36

Planned Methods

College students

Lab environment

Between-participants

Pre & Post study interviews,

During-study survey

Population

Setting

Study Design

Data Collection

15 of 36

Protocol

  1. Welcome participant & inform context of study
  2. Sign consent form
  3. Interview participant with pre-study questions
  4. Introduction to the task
  5. Run trials of six conditions and complete survey at the end of each
  6. Interview participants with post-study questions
  7. Debrief & thank the participant for their participation

16 of 36

Interview & Survey Questions

  • Pre-study interview to assess experience with robots
  • Survey at the end of each trial to assess perception of movement, comfortability and trust in completing task
  • Post-study interview for general feedback

17 of 36

Robot embodiment

Alvik Arduino Robot

18 of 36

Facial Expressions

Russell Circumplex Model of Affect (1980)

19 of 36

Facial Expressions

Sad

Neutral

Happy

20 of 36

Excited Path

Demo played in x4 speed

21 of 36

Neutral Path

Demo played in x4 speed

22 of 36

Lazy Path

Demo played in x4 speed

23 of 36

Results & Discussion

Study results and their implications

04

24 of 36

Survey Results

25 of 36

Survey Results

26 of 36

  1. Alvik’s expression influenced most participants level of comfort and trust

Overall: Happy > Neutral > Sad

Comfort

  • No connection with neutral face
  • With sad face seemed like Alvik didn’t want to complete the task

​

Trust

  • Expressions caused bias
  • With sad face would not complete task
    • “Didn’t like seeing it suffering”
  • Happy made task more enjoyable to watch and do better job

27 of 36

  1. Alvik’s movement influenced most participants level of comfort and trust

Overall: Neutral > Excited > Lazy

Comfort

  • Liked smoother movements, pauses made uncomfortable
  • Didn’t like when it would go to trash in a roundabout way

​

Trust

  • Liked when Alvik moved efficiently to the trash
    • Some didn’t - might miss other possible trash in the room
  • Roundabout movements and pauses decreased trust
  • Scanning motions - trying to clean entire room rather than one piece of trash

28 of 36

  1. Alvik’s expression had an impact on how people perceived its movement

29 of 36

  1. Alvik’s expression had an impact on how people perceived its movement
  • Happy made it seem more friendly
  • Lazy movement was looked at more positively with happy face despite pauses
  • Neutral movement was looked at more negatively with sad face
    • “Felt very contradictory”
  • Movements were analyzed in a more technical manner

30 of 36

Discussion

Reasoning for participant variations in perception:

  • Most participants noticed that some paths were repeated
    • Answers involving movement were similar between conditions they thought had repeated paths
  • Different levels of experience with cleaning robots
  • Different levels of technical/robotics experience

31 of 36

Conclusions & Future Work

What this means and what we can further explore

05

32 of 36

Conclusions

H2: The perceived movement of a robot, influenced by its facial expressions, will affect human comfort and trust in its ability to complete a task

H1: A robot’s displayed facial expressions will influence human perception of its movement, such that the same movement path will be interpreted differently depending on the associated emotion

Supported!

Supported!

33 of 36

Conclusions

  • Direct paths and positive expressions = higher comfortability
    • Higher awareness, joy for task
  • Longer/indirect paths with many pauses = lower trust
    • Perceived as confused or hesitant for next steps
  • Scanning motion provided higher trust in system’s ability to thoroughly clean the environment
  • The importance of expressive non-verbal communication for robot design to improve user experience

34 of 36

Future Work

  • Have participant assign a specific emotion to each combination
  • Within-subjects study design, field study setting
  • Increase interaction between participant and robot in the task
  • Recruit more diverse population of participants
  • Have more specific questions to guide answers to be more specific

35 of 36

Thank you!

Questions?

36 of 36

Citations

[1] G. Dietz, J. L. E, P. Washington, L. H. Kim, and S. Follmer, “Human Perception of Swarm Robot Motion,” May 2017, doi: https://doi.org/10.1145/3027063.3053220.

[2] Q. Xie, Z. Chen, and D. Luh, “How Do Humans Recognize the Motion Arousal of

Non-Humanoid Robots?,” Applied Sciences, vol. 15, no. 4, p. 1887, Feb. 2025, doi:

https://doi.org/10.3390/app15041887.

​

[3] N. Walker, C.Mavrogiannis, S. Srinivasa, M. Cakmak, “Influencing Behavioral Attributions to Robot Motion During Task Execution,” PMLR 164:169-179, 2022, https://proceedings.mlr.press/v164/walker22a.html

​

​

​