200 likes | 219 Views
The Benefits Of Interactions With Physically Present Robots Over Video-displayed Agents.
E N D
The Benefits Of Interactions With Physically Present Robots Over Video-displayed Agents. Bainbridge, W. A., Hart, J. W., Kim, E. S., & Scassellati, B. (2011). The benefits of interactions with physically present robots over video-displayed agents. International Journal of Social Robotics, 3(1), 41-52. Presented by Prerna Chikersal
Goal Greetings. Cooperation. Trust. Personal Space. VS.
Previous Work • Hypothesis: “Participants will afford greater trust and personal space to physically present robots.” • ➕ Physical >> Virtual • Cooperative block-stacking task with a talking agent (engaging, enjoyable, informative, credible). • Cooperative game (watchfulness, helpfulness). • ➖ Physical ~ Virtual • Verbal, desert survival, role-playing task (social perceptions). • Interacting with Sony AiboIf lonely: physical >> virtual. Not lonely: physical ~ virtual.
Study Conditions • Condition 1 (20 participants)Physically present robot • Condition 2 (20 participants)Robot in live video display (frontal view) • Condition 3 (19 participants)Robot in live video display (frontal view) ➕ 3D information in another live video display (overhead view)
Study Tasks and Outcomes • Task 1a: GreetingReciprocation • Task 1b: DistractionAcclimation • Task 2: Simple Task CooperationResponse time • Task 3: Unusual Task CooperationTrust • Task 4: Proximity Task CooperationPersonal Space Trust!
Methods • Behavioral observations • E.g. Greeting, response time, obeying instructions • Questionnaires • General impressions, characteristics of interactions, overall impressions, and biographical information.
Results • Greeting • 10 in Physical, 10 in Video, 10 in Video+3D. • No significant difference. • Physical ~ Video ~ Video+3D. • Response Time in Simple Tasks • 20.5 secs in Physical, 27.09 secs in Video, 19.73 secs in Video+3D. • Significant difference. • Video >> Physical and Video+3D • Difficulty interpreting 3D gestures in 2D video.
Results contd. • Trust • 12 in Physical, 2 in Video, 3 in Video+3D. • Significant difference. • Physical >> Video and Video+3D. • Trust OR difficulty interpreting 3D gestures? • Interaction with the garbage can • 19 in Physical, 9 in Video, 11 in Video+3D. • Significant difference. • Physical >> Video and Video+3D. • So, could be difficulty interpreting 3D gestures? • Excluded? Probably not?
Results contd. • “Trust” response rate (for unusual task) • 17.8 secs for Physical, 42.18 secs for Video, 19.2 secs for Video+3D. • Significant difference • Video >> Physical and Video+3D • Difficulty interpreting 3D gestures in 2D video. • Personal space • 17 in Physical, 5 in Video, 8 in Video+3D. • Significant difference. • Physical >> Video and Video+3D • Physical space OR risk of damage/ space occupied by the hardware?
Results contd. • Questionnaires • Most Questions • No significant difference. • Usually physical >> Video and Video+3D. • Four questions • Significant differences. • Natural: physical >> Video and Video+3D. • Homogenous, negative, and varied: Video and Video+3D >> Physical. • Varied qualitative responses for the trust task.
Discussion Questions Q1. How convinced are you of the paper's findings? Choices? Confounding variables?
Discussion Questions Q1. How convinced are you of the paper's findings? Choices? Confounding variables? Trust • “I struggle with the concept of trust as presented like this. The assumption that the task is completed because of an unconscious trust metric. What if they threw the books, just because they were told to throw the books? Sometimes people just do the random stuff they are told to do?” – Lynn • “Neutral task” – Nur Yildirim • Participant demographics?
Discussion Questions Q1 contd. How convinced are you of the paper's findings? Choices? Confounding variables? Unrealistic study conditions • “I agree with Steven here. If the study was done online the results might be different. Also, instead of virtually showing the robot, we know that people are not bothered by the embodiment of the robot in virtual space - like a virtual assistant like Google or Siri. So I do not totally get the point of skyping with a physical robot. Is it to fill in the literature, or maybe it is to understand the interaction effects, but I am confused - would such an interaction ever exist?” – Karan
Discussion Questions Q2. Human agent instead of a robot?
Discussion Questions Q2. Human agent instead of a robot? • “So I get that people feel more engaged when the robot has physical proximity, but isn't that true of everything? For instance, I would feel more engaged and socially interactive with my dogs being there rather than just a video of my dogs: similarly with humans. I appreciate this work but would have liked a comparison with humans. Like what makes robots a "special" category of creature that has different interaction modalities than other interactive agents.” – Yasha
Discussion Questions Q3. What if the task did not require interaction with the physical environment?
Discussion Questions Q3. What if the task did not require interaction with the physical environment? • “I bet since people are use to software (walk through wizards or other “help” agents) walking them through online tasks, that they might have more trust in that realm with a robot who is also “in that realm” of online too.” – Steven
Discussion Questions Q4. Does Task 4 truly measure personal space?
Discussion Questions Q4. Does Task 4 truly measure personal space? • “I believe personal space is a culturally subjective thing. I think there are those who will always reach above the monitor and then apologize to whoever is on the other end of the screen, others will move the monitor aside because it’s a monitor, not the robot/human.” – Lynn • ”I also think the participants give the physical robot more space just because they are afraid of damaging it, or afraid of getting in its way as it may do something unexpected (especially after the "throw this book into the garbage can" task).” – Anh