Flashlight Robot

Tennis Club: Matthew Prince, Donovan Cece, Eric Chen, Matthew Greenbaum

Introduction

Robots as Helpers

• “Dull, Dangerous, Dirty”
• Robots tend to be utilized as replacements for humans in certain situations
• “Delicate”
• Robots can also be utilized as helpers, aiding humans in teamwork scenarios

Scenarios for Robot Helpers

• Care Robots (i.e. ROB 204)
• Medical Assistants
• Robotic Prosthetics

​

Scenarios for Robot Helpers

• Care Robots (i.e. ROB 204)
• Medical Assistants
• Robotic Prosthetics

​

Goal

Investigate the trust individuals have towards team- work with robots and gain a better understanding of the effectiveness of human-robot teamwork.

​

Related Works

Related Works

“I Need a Third Arm! Eliciting Body-Based Interactions with a Wearable Robotic Arm”

​

Related Works

• “I Need a Third Arm! Eliciting Body-Based Interactions with a Wearable Robotic Arm”
• Investigates control methods for wearable robotic arms [2]
• Uncovers which modalities of control were the most helpful
• Basic Control: Sounds
• One hand is free: Gestures
• Overall: Several means of autonomy work best

​

Related Works

“3HANDS Dataset: Learning from Humans for Generating Naturalistic Handovers with Supernumerary Robotic Limbs”

​

Related Works

• “3HANDS Dataset: Learning from Humans for Generating Naturalistic Handovers with Supernumerary Robotic Limbs”
• Develops a dataset and models for natural human-robot handovers [1]
• Evaluates key handover behaviors
• Where to hand the object
• When to begin the handover
• How to move naturally

​

Related Works

• Given Study RQ
• What is the best way to control robotic peripherals?
• Given Study Metrics
• Strategies for control used
• Occupied / Hands Free Scenarios
• Type of gesture / body part used

​

• Our Study RQ
• How do people perceive the help of robotic peripherals?
• Our Study Metrics
• Success of teamwork tasks (Timing)
• Testaments from participants (Trust shown)

Which control methods are the best for robotic peripherals?

What variables impact the comfort and usability of a human-robot interaction?

How do people perceive the help of robotic peripherals?

Task Completion

User Preference

Robot Helpfulness

Control Methods

Usability in Daily Tasks

Our Study

I need a Third Arm

3HANDS Dataset

Methods

Research Question

To what extent, if any, can a flashlight holding robot help users complete hands-on tasks in a dark environment, and make them feel less burdened by the task?

Hypothesis

• The participant will complete the tasks faster with the robot holding the flashlight for them, rather than if they needed to hold the flashlight themselves
• The participant will feel that the robot holding the flashlight for them was helpful, and would trust the robot to help them in the future

Protocol Overview

1. Introductions and and context
2. Consent forms
3. Introduction to tasks
4. Study start
1. Condition 1: No robot assistance
2. Condition 2: With robot assistance
5. Post-task metrics
6. Debriefing and goodbyes

Study Design

• Constructs
• Trust
• Effective Teamwork
• Sample: Engineering Students
• Easier to recruit
• Relevant background
• Environment: Dark classroom/workspace
• Low/no light conditions
• Task materials separated

​

Study Design

• Within-subjects design

​

​

​

​

​

• Task order is randomized between conditions

Condition 1

Condition 2

Planned Methods

Selected tasks based on the Common Atomic Manual Actions Library [4]

Inserting

Screwdriver

Screwing

Planned Methods

Opening/Closing Bottles

Outlet Plugging

Planned Methods

Screwdriver Task

Nuts and Bolts

Planned Methods

Buttoning

Metrics

• Success of the task
• Timing of task completion
• Task success
• Survey Results
• Robot perception and performance
• Trust in the robot
• Future use

• Behavioral Observations
• Flashlight use
• Task understanding
• Robot teamwork

Results & Discussion

Table of times

Survey Results

• 1-5 scale of options
• More natural responses

Survey

Robot Performance:

• Did the robot flash exactly where you wanted it to?
• Did you feel that the robot got in the way of completing your tasks?

Somewhat

Somewhat

Neutral

No

Not Really

Not Really

Survey

Human Perception of the robot:

• Did you feel like the robot arm assisted you in completing the tasks?
• Did you feel that using the robot arm was easier than just holding the flashlight yourself?

Somewhat

Somewhat

Somewhat

Yes

Yes

Yes

Survey

Questions on the future:

• Compared to managing the flashlight yourself, was the robot worth it?
• Would you want to use the robot arm again in the future?

Somewhat

Neutral

Not Really

Yes

Somewhat

Neutral

Behavioral Analysis - Flashlight Handling

Participant 1

Participant 2

Participant 3

Located task then left flashlight on table

​

Used his MOUTH during the shirt buttoning

Located task and then kept flashlight in one hand while performing tasks

Located task and then kept flashlight in one hand while performing tasks

Behavioral Analysis - Task Understanding

Participant 1

Participant 2

Participant 3

Took a little longer to understand each task on the first iteration

​

Relied on the paper with instructions

Took a little longer to understand each task on the first iteration

​

Almost didn’t use the paper instructions at all

Understood the tasks perfectly on both iterations

​

Used the paper instructions a little

Behavioral Analysis - Robot Teamwork

Participant 1

Participant 2

Participant 3

Had to wait a little bit for the robot to move

​

Noted how more light was directed onto the tasks then the previous condition

Noted how the light didn’t fully reach where they wanted it to

Very easily followed the robot

​

Noted how it was much easier to not have to hold the flashlight

Behavioral Analysis - Participant #1

• Located the tasks using the flashlight, left light on table, used his MOUTH!!
• Took a little longer to understand the tasks on the first iteration
• Relied a lot on the paper with written instructions
• Waited on robot a little
• There was much more light on the tasks during robot phase

Behavioral Analysis - Participant #2

• Very quick to understand all tasks, even from the first iteration
• Without the robot, utilized the flashlight to locate the objects and read certain orderings, but then used both hands to complete tasks
• Very easily followed the robot during the second condition
• Noted how it was much easier to not have to hold the robot during the study

Behavioral Analysis - Participant #3

​

• Held the flashlight in hand after locating task
• Took a little adjustment to understanding tasks on first iteration
• Was very quick upon second iteration
• Didn’t use the paper much
• Noted on how the robot didn’t fully reach where they wanted the light

Conclusions & Future Work

Conclusion & Interpretation

• Potential
• Time data
• Participant feedback
• Skewed data
• Worth it?
• Improvements

Ease

Trials

Future Work

• Future studies could include:
• Deception
• Person holding the flashlight
• More complex tasks
• Put into industry
• Navigation around a room
• More participants
• Familiarize participants with tasks

References

Works Cited

[1] “3HANDS Dataset: Learning from Humans for Generating Naturalistic Handovers with Supernumerary Robotic Limbs.” Arxiv.org, 2025, arxiv.org/html/2503.04635. Accessed 19 Mar. 2025.

[2] Muehlhaus, Marie, et al. I Need a Third Arm! Eliciting Body-Based Interactions with a Wearable Robotic Arm. 19 Apr. 2023, https://doi.org/10.1145/3544548.3581184. Accessed 22 Nov. 2023.

[3] Murthy, G. N. Keshava, et al. “Mind Control Robotic Arm: Augmentative and Alternative Communication in the Classroom Environment.” International Journal on Recent and Innovation Trends in Computing and Communication, vol. 11, no. 10s, 7 Oct. 2023, pp. 412–422, ijritcc.org/index.php/ijritcc/article/view/7649, https://doi.org/10.17762/ijritcc.v11i10s.7649. Accessed 19 Mar. 2025.

[4] H. Qi and J. Malik, “Common atomic manual actions for robots,” 2025, accessed: Apr. 03, 2025. [Online]. Available: https://robo-cama.github.io/

​

​

Work together. Create smart machines. Serve society.

@umrobotics

robotics.umich.edu