RO-MAN 2020 Keynote: The value of interaction in terms of commercial service robots

ROMAN 2020 Keynote
http://ro-man2020.unina.it/keynote-speakers.php

Transcript

1. The Value of Interaction in Terms of Commercial Service Robots Jun Baba CyberAgent, Inc. RO-MAN 2020 Keynote Talk
2. Brief Introduction of Our Projects
3. CyberAgent, Inc. Programming Education Cloud Funding Develops a variety of businesses centered on the Internet industry.
4. Interactive agents as a next solution PC growing phase Smartphone growing phase CyberAgent Internet Advertising Sales Interactive agents as marketing support solutions Next phase
5. Joint research project with Osaka Univ. Ishiguro Lab.
6. Joint research chair since 2017 Ishiguro lab staff Specially appointed position CyberAgent researchers Frontier Intelligent System Joint Research Chair Specially appointed position Research fund Advise/Discussion
7. Our Field Studies (1/2) Service robot in a hotel room [IROS 2020] Robot handing out coupons in a shopping mall [HRI 2020] Teleoperated service robot in a summer festival [CHI 2020]
8. Our Field Studies (2/2) Service robots in a hotel elevator hall [HAI 2018] Chatbot promoting products in an online shopping site (chatbot) Greeting robot in a office [Advanced Robotics 2020]
9. The value of interaction
10. Commercial Service Robots
11. Case: Reception / Information Counter Guide robots at an information counter in a japanese shopping mall
12. Other options that can be applied Self-reception system in offices Touchable digital signages Shopping mall guide apps
13. Sad scenes Matthew I. Beane. 2020. In Storage, Yet on Display: An Empirical Investigation of Robots' Value as Social Signals. In Proceedings of the 2020 ACM/IEEE International Conference on Human-Robot Interaction (HRI '20). from Twitter: https://twitter.com/takenao708/status/1003077139506159616
14. Difference between robots and other systems The main of differences is “eliciting people’s social/interpersonal reactions/behaviors”. • Impressions of the robots: Expertise / Pressure / Trust ... • Emotional reactions: Empathy / sympathy ... Social reactions/behaviors ⇒ Building an interpersonal relationship with a customer ⇒ What happens next?
15. Changing people’s actions Influencing customers to take actions that would not have happened without interacting with the robots. ⇒ The essential value of interaction in terms of business
16. 3 Steps to change people's actions Step1 Step2 Step3 Environmental improvement Build an environment that allows for conversation Emotional relationship Build an emotional relationship for increasing acceptability Persuation Encourage customers to take the desired actions
17. Step 1: Environmental improvement • “Please Listen to Me: How to Make Passersby Stop by a Humanoid Robot in a Shopping Mall” [HRI (Companion) 2020] • “Behavioral assessment of a humanoid robot beyond human performance when attracting pedestrians in a mall” [in-review]
18. Service robots are often ignored by customers Greeting and announcement to make customers stop in a hotel looby Customer stop rate (= # of stops / # of pedestrians )
19. Field study with a service robot in a shopping mall A service robot talks to passersby, introduces the store, and hands out coupons for free water. All customers are labeled as one of the above paths from recorded videos.
20. Robot behaviors to make passersby stop
21. Results • All stop rates are higher than the results of our previous experiments. • Passive-negative behavior (troubling) has the highest SR / DSR. ⇒ Are these results more competent than human staffs?
22. Field study with human staffs in the shopping mall A human participant talks to passersby and hands out free water coupons in the similar regulations with the robot conditions.
23. Results of human staffs • All stop rates are lower than the results of robot conditions. • DSRs are extremely high compared with robots
24. Human staffs vs Robots Humans Robots Robot with the troubling behavior won all staffs. Humans > Robots Humans < Robots
25. Discussion • Robots originally has an ability to create a more favorable environment that allows for conversation with customers than human staffs. ◦ The ability can be further enhanced by designing the behavior thoroughly. • We can also improve the effect of robot behaviors after building the environment: ◦ Conversational ability ◦ Attractive/persuasive scenario ◦ Social presence and pressure of robots
26. Step 2: Emotional relationship • “Can a Humanoid Robot Engage in Heartwarming Interaction Service at a Hotel?” [HAI 2018] • “Soliloquising Social Robot in a Hotel Room” [OZCHI 2019] • “Smart Speaker vs. Social Robot in a Case of Hotel Room” [IROS 2020]
27. Robot’s Service in a Hotel
28. Focus on gaps in service • Can a robot build a positive emotional relationship in a hotel room? • Can the recommendations of local stores by robots be accepted?
29. 部屋にロボットを起きました

30. Interaction scenario with robot behaviors from our paper in OZCHI 2019

31. V.S. smart speaker from our paper in IROS 2020

32. Result 1: Emotional relationship Dialogue count: Robot > Speaker Positive impression: Robot > Speaker
33. Result 2: Recommendation Accepted recommendations: Robot > Speaker (including “I visited” & “I intend to visit”) Robot Speaker
34. Step 3: For effective persuation • “Teleoperated Robot Acting Autonomous for Better Customer Satisfaction” [CHI EA 2020] • “Mind The Voice!: Effect of Robot Voice Pitch, Robot Voice Gender, and User Gender on User Perception of Teleoperated Robots” [CHI EA 2020]
35. Persuasive talk Persuasive talk is difficult for humans as well as robots. • The company still need to train their salespersons [Guo 2014] • A good talker ≠ a good robot scenario designer Can a human's communication skills work with a robot's appearance? [Guo 2014] Wenxia Guo and Kelley Main (2014) ,"Can Persuasion Knowledge Change Preferred Persuasion Based on Self-Construal?", in NA Advances in Consumer Research
36. Teleoperated robot for providing services Tele-working platform without restriction of time and location This remote-controlled robot allows you to provide services through the robot from anywhere with your PC or tablet connected to the Internet. We aim at creating a platform that makes the best use of the robot's acceptability and expressive power to easily serve customers from a remote location. Features Easy operation Robotic voice Extend the expression
38. Field test in a summer festival
39. Performance and finding • Many customers were satisfied, intended to repeat, and enjoyed the talks. • People who not realize the operation rated higher than those who realized it. ◦ A CG agent as a computer increased willingness to disclose than as an avatar [Lucas, 2014] Satifaction Intention to Repeat Enjoyed Satifaction Intention to Repeat Enjoyed [Lucas 2014] Lucas GM, Gratch J, King A, Morency L-P (2014) It’s only a computer: Virtual humans increase willingness to disclose. Computers in Human Behavior,
40. Next issues to be addressed
41. Observations and findings in the field studies • Robots can change customer behaviors/actions in real fields with thoroughly designed interaction. ◦ Their performances are equal to or higher than conventional methods (humans / devices). ◦ Human-Robot interaction has a powerful effect in terms of business • Appearance has a great power to encourage people to take desired actions. ◦ Customers imagine and expect the future of interaction from the appearance of robots. ◦ The positive gap between the expectation and the actual can change customer behaviors. ◦ This may be a distinctive finding that reflects real reactions in real fields.
42. Issue to be tackled for step 1 For step 1 Environmental improvement Build an environment that allows for conversation Interactions to build quickly an environment and a relationship for conversation • Verbal/non-verbal interaction to attract customers quickly • Making people set the expectations instantly
43. Issue to be tackled for step 2 A persuasive dialogue system with robot’s intentions/desires expression • Emotions are intentional structures of beliefs and desires. [Green 1992]. • Incorporating robot’s desires into dialogue strategies to perform a emotional conversation For step 2 Emotional relationship Build an emotional relationship for increasing acceptability [Green 1992] Green O.H. (1992) The Belief-Desire Theory of Emotions. In: The Emotions. Philosophical Studies Series, vol 53.
44. Summary • The essential value of interaction: ◦ Changing customer actions/behaviors ◦ 3 steps and 3 cases for showing the value: i. Environmental improvement - Robot handing out coupons in a shopping mall ii. Emotional relationship - Service robot in a hotel room iii. Persuation - Teleoperated robot for better customer satisfaction • Please contact us: ◦ baba_jun@cyberagent.co.jp CyberAgent AI Lab ◦ @baba5246 ◦ Jun Baba