Online Investigation of Human-Robot Proxemics

Gerson Uriarte Author

08/03/2021 Added

4 Plays

Description

Presentation outlining the research done through NIMBUS lab. This study included 47 participants who interacted with small unmanned aerial vehicles through an online platform.

Searchable Transcript

Search:

[00:00:01.620]Hello. My name is Gerson Uriarte.
[00:00:03.930]I'm a senior computer engineering student at the University of Nebraska Lincoln.
[00:00:07.920]This summer I've been conducting research through NIMBUS labs.
[00:00:11.310]I conduct an online investigation of human robot proxemics with the help of my
[00:00:15.750]advisor, Brittany Duncan, and her graduate student, Siya.
[00:00:21.720]The motivation for this project is that the introduction of robots into our
[00:00:25.320]daily lives will cause some degree of discomfort. As a robot designer.
[00:00:30.090]You want to create something that will not only allow your invention to do its
[00:00:33.480]intended purpose, but to do it without causing any discomfort to others,
[00:00:38.550]the field of human-robot proxemics helps explain this comfort.
[00:00:42.870]The definition of human robot proxemics is the study of how humans distanced
[00:00:46.700]themselves from robots to explore this field.
[00:00:51.300]We want to break down the components of a small unmanned aerial vehicle or
[00:00:55.830]sUAV that contribute to or hinder comfort.
[00:01:03.280]The objectives of this online study is to provide data that will allow us to
[00:01:07.810]understand the role of appearance,
[00:01:10.450]primarily focusing on the structure and modifications to a sUAV and
[00:01:15.310]also finding a correlation between participant comfort and the degree of safety
[00:01:20.020]perceived from modifications. Finally,
[00:01:23.650]the study will be using an online platform to reduce costs and participant
[00:01:27.670]response times
[00:01:32.430]the materials used for the study include a DJI flame wheel F three-thirty
[00:01:37.410]kit. This kit can be seen here on the right,
[00:01:40.920]and it is the top picture.
[00:01:43.920]The prop guards that are seen on the flame wheel in the middle picture,
[00:01:49.290]um, were 3d printed.
[00:01:52.260]And the picture of the F330 at the bottom shows a cage modification.
[00:01:56.700]And this cage was built out of carbon fiber rods and aquarium netting.
[00:02:02.730]This F330 was also flown autonomously using ROS,
[00:02:07.530]the robot operating system,
[00:02:09.480]and the position was provided using a vicon system.
[00:02:16.260]The online study was run on a platform built by Amazon called mechanical Turk.
[00:02:21.540]The user's actions were recorded throughout the study and were later evaluated.
[00:02:26.970]The study included 47 participants who got randomly put into two groups that
[00:02:31.770]experience either an egocentric view or an existential view.
[00:02:36.120]Both groups viewed the same conditions,
[00:02:37.740]but the order that they viewed them in was randomized.
[00:02:41.160]All the videos were recorded using autonomous flight,
[00:02:43.860]alongside safety code that stopped the drone half a meter away from the
[00:02:47.640]participant or camera, depending on the group that the participant was in
[00:02:53.700]participants answered, uh, questionnaires prior to the beginning of the study.
[00:02:57.510]And then subsequently after every condition.
[00:03:03.760]This picture here is the view of the exocentric group's video.
[00:03:08.110]This shows all three conditions.
[00:03:10.870]This view shows a sUAV approaching a participant and various modifications.
[00:03:16.510]The modification on the far left is the prop guard modification.
[00:03:21.070]The middle modification is no modification.
[00:03:24.190]The right is a case modification. As a side note,
[00:03:28.330]the participants did not view all modifications approaching the participant at a
[00:03:32.500]single time.
[00:03:33.820]Instead they viewed the approach with a single modification in each condition.
[00:03:41.800]Now for the results of the study, overall,
[00:03:44.830]the average results showed that the participants distanced themselves farthest
[00:03:48.820]from the no guard and closest to the cage.
[00:03:52.690]The results still need to be broken down further.
[00:03:54.850]Since the study was run very recently,
[00:03:58.120]these tables at the bottom show the distance for each condition,
[00:04:02.680]the average results,
[00:04:03.580]which is the table on the left here show the combination of egocentric data and
[00:04:07.630]exocentric data.
[00:04:09.910]The two most interesting thing about the results though,
[00:04:13.630]show when we see the two groups independently
[00:04:19.960]and the two groups,
[00:04:20.710]we can see that the overall distancing of the note modifications versus the
[00:04:24.490]occasion modification shows higher comfort with the cage.
[00:04:29.320]The two groups had the same trend, but not the same distancing.
[00:04:33.190]The exocentric view allowed the different sUAV systems to approach closer
[00:04:38.050]than the egocentric group.
[00:04:40.330]This could be due to the appearance of the video and the quality of the camera
[00:04:44.020]when facing the drone prop guards also had an inconclusive effect on
[00:04:48.880]distancing, but again,
[00:04:50.170]this could be due to the difference in appearance between the two views
[00:04:59.080]To describe future work on the proximity study,
[00:05:01.480]using an in-person study could allow for more accurate results because the
[00:05:04.930]online experience differs from participants that are participant due to certain
[00:05:08.770]factors.
[00:05:10.210]This could be due to something as simple as the internet connection or the
[00:05:13.510]screen quality and could even be the headphone quality as sound is
[00:05:18.310]very important and distancing
[00:05:22.210]an in-person study could also help understand the difference between prop guards
[00:05:26.050]versus no modifications due to the ability to see the difference more clearly in
[00:05:30.370]an in-person setting for the research needs to be done.
[00:05:34.270]So identify the key components that allow closer distances, for example,
[00:05:37.900]some questions that were not answered by the study,
[00:05:39.820]where was the closer distancing proceed from protection from the
[00:05:44.260]prompts,
[00:05:45.700]or was it because the edges of the propellers and the sUAV body didn't
[00:05:50.620]appear as sharp because of the netting?
[00:05:57.080]Finally,
[00:05:57.560]I would like to acknowledge those who helped me make my research possible.
[00:06:01.820]Alisha Bevins was a graduate student who helped me in the setup process of my
[00:06:05.750]study.
[00:06:07.550]Ajaywas also another graduate student and Ajay helped me get autonomous flight
[00:06:12.500]working with the F330 frame wheel.
[00:06:17.180]And Jacob helped me gather all the hardware that I need to go perform my study.
[00:06:21.740]I would also like to thank the Nimbus lab for allowing me to work with them this
[00:06:24.770]summer. Lastly,
[00:06:28.900]I would like to thank all of you for being interested in my research and for
[00:06:32.080]taking the time to view my presentation.

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Online Investigation of Human-Robot Proxemics

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments