In our previous post on ACM/IEEE Human Robot Interaction (HRI) 2022, we discussed a couple of interesting papers. This post discusses a couple of other interesting topics from late breaking sessions at HRI 2022.

How people associate with Cozmo’s emotions?

This work from Professor Emily Cross’s research group at SOBA lab, University of Glasgow, discusses how people react to videos displaying Cozmo’s emotions, and how their emotional state gets altered after seeing these videos. The study surveyed 103 participants, who were asked to look at 10 seconds long videos of Cozmo showing an emotion, and then asked to answer two questions: (i) Pick the emotion they believed Cozmo was showing and (ii) Pick how their emotional state changed after seeing Cozmo’s emotion. The displayed video was chosen from a collection of 348 videos in which Cozmo’s emotional state was determined by 3 experts who agreed on the emotional state. The findings of the study can be summarized from the following two results which are documented in the paper.

From Figure A, we can see the videos with emotional content of anger, happiness, and sadness were identified more correctly (Correct identification ratios of 79%, 62%, and 75% respectively) than the videos with no emotional content, or the emotional content of surprise. From Figure B. we can see that the videos that displayed Cozmo showing happiness or sadness had the most effect on emotions of the viewers. This study is promising, because it shows that a happy personal robot can indeed bring cheer in the life of people around it. Furthermore, since the main motivation from Anki to build Vector was to create a emotional robot, it appears that this intention indeed succeeded even though Vector was a commercial failure.

As the authors admit, the experiment findings are limited in that the people surveyed were shown videos of Cozmo, not the actual Cozmo robot… so another similar study with people actually interacting with Cozmo would shed more light into the validity of the conclusions.

How can robots be used in a family’s playtime?

Researchers from University of Tsukuba, Japan explored how family playtime with a Nao robot could be beneficial for family members in this paper. The researchers ran a study with three children who played with the Nao robot along with their family members for a duration of 25 minutes which was video recorded. They observed that activities changed when family members played together compared
to the child playing alone with the robot. As an example, for all the three children, the percentage of time doing verbal activities such as talking to the robot, or doing physical activities such as playing with balls with the robot, increased when family members were around. Similarly, all children spent lesser time playing Rock-Paper-Scissors with the robot when their family members are around.

The researchers make few important conclusions that will be important for manufacturers of home robots.

Home robots need to identify different humans in the family and interact with them differently.
Robots need to have a wide range of activities to interact with humans. As an example, it was shown that while 2 children enjoyed playing with the robot, one child enjoyed verbal activities with the robot.
Robots should be able to divide their attention span across different humans in the room evenly.

HRI 2022 certainly has a wide range of fascinating research topics. We will aim to bring discussions on interesting topics and ideas in future posts.

Learn With A Robot

HRI 2022 Late Breaking papers