Download to read offline
Uploaded byIllai J. Gescheit
Sociable Robotics
The document discusses social robotics and focuses on the robot Kismet. It describes Kismet as an early social robot created to interact with humans using social cues like turn-taking, eye contact and facial expressions. Experiments showed people responded intuitively to Kismet's social cues and engaged with it naturally through turn-taking exchanges, though Kismet had limitations in its perception, mechanics and responses. The document outlines the goals of evaluating social robots and improving their ability to naturally interact with humans.
Sociable Robotics
- 1. Sociable Robotics Illai J.Gescheit1
- 2. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Why did I chose Social Robotics? You’re Number 2
- 3. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Understanding Humans Sometimes it’s difficult to understand and live with others You’re Number 3
- 4. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Understanding Robots So what about living with robots? You’re Number 4
- 5. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics You’re Number 5
- 6. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics What is a social robot? Robots that people apply a social model to in order to interact with and to understand A question: “Does the robot only appear to be socially intelligent or is it genuinely so?” You’re Number 6
- 7. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Social Models Constrained Full You’re Number 7
- 8. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Robotic Social Development Leverage from playful, infant like interaction to foster its social Development Kismet You’re Number 8
- 9. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics What it should look like? Dogs are socially Responsive and socially intelligent in a genuine sense Robot doesn’t have to look like a human To be genuinely social intelligent You’re Number 9
- 10. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Social Robot Success What matters is how the robot interacts face to face with people, and how people interact with it in a human environment You’re Number 10
- 11. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Known Types Encourage people to anthropomorphize the technology in order To interact with it (but goes no further) Uses human-like social cues and communication modalities To facilitate interactions with humans, the social model Tends to be shallow and its social behavior is pre-canned. Benefit from interaction with people (training model). These Interactions their internal structure in a deeper level. People Shape the robot’s behavior. They are socially passive. Socially participative with their own internal goals and Motivations. Engage people in a social manner to benefit Both the person and the robot. The robot’s social behavior Stems from its computational social psychology. You’re Number 11
- 12. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Meet Kismet You’re Number 12
- 13. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Kismet Technicalities You’re Number 13
- 14. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Speaking Turns The ability to exchange turns during face-to-face interactions is the cornerstone of human style communication and instruction You’re Number 14
- 15. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Speaking turns Limitations Respond After 0.1 sec Respond After 0.25 sec Respond After 0.50 sec Respond After 0.25 sec You’re Number 15
- 16. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Envelope Displays Regulating Speaking turns Regulating Speaking turns Facial DisplaysFacial Displays gesturesgestures Shifts in gazeShifts in gaze Eye blinksEye blinks Eye contactEye contact Raise of eyebrowsRaise of eyebrows Paralinguistic Social Cues Paralinguistic Social Cues You’re Number 16
- 17. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Experiments You’re Number 17
- 18. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Kismet’s Envelope Displays You’re Number 18 No. Action Cue 1 Acquire the Floor Break eye contact Lean back 2 Start a speaking turn Vocalize a Kismet-esque babble 3 Stop a speaking turn Stop vocalization and re- establish eye contact 4 Hold the floor Look to the side 5 Relinquish the floor Raise brows Lean forward a bit Blinking occurs at most at the end of vocalization.
- 19. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Social Robot Evaluation • Do People intuitively read and naturally respond to Kismet’s social cues? • Can Kismet perceive and appropriately respond to theses naturally offered cues? • Does the human adapt to the robot and the robot adapt to the human, in a way that benefits the interaction • Specifically, is the resulting interaction natural, intuitive and enjoyable for the human • Can Kismet perform well despite its perceptual, mechanical, behavioral and computational limitations? You’re Number 19
- 20. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Kismet’s Turn-Taking Performance You’re Number 20
- 21. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Kismet’s Turn-Taking Performance You’re Number 21
- 22. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Kismet’s Turn-Taking Performance You’re Number 22
- 23. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics Future? You’re Number 24
- 24. Illai J. Gescheit– Mechanical Engineering Dept. – Tel Aviv University Social Robotics – Semester Project – Introduction to Robotics You’re Number 25
Editor's Notes
- #8 It is important to recognize that humans are a profoundly social species. Our social-emotional intelligence is a useful and powerful means for understanding the behavior of’ and for interacting with some of the most complex entities in our world – people and other living creatures . Faced with non-living things of sufficient complexity (i.e. when the observable behavior is not easily understood in terms of its underlying mechanisms)’ we oftern apply a social model to explain ‘ understand and predict their behabior as well. For instance’ we are all familiar that people anthoropomotrphize all sorts of technologies( e.g. cars’ computers ect.) . The studies of Premack and Premack show that people attribute mental state (i.e. intents’ beliefs ‘ feelings desires ect( to describe the bejavior of his ingenious vehicles. Right or wrong’ people rely on social modeels (* or fluidly seitch between using a social model with other mental models( to make complex behavior more familiar and understandable and more intuitive with which to interact. We do this because it is enjoyable forr us ‘ and it is often surprisingly wuite useful.
- #10 We take the position that dogs are another socially responsive species- socially intelligent in a genuine sense, althought their social sophistication is less than that of a human. Hence this criterion for success should not be confused with trying to build a robot that is indistinguishable from a human inside and out. – the appearance of the robot and its internal design details can be quite different from the human couterpart, what mattes is how it interacts face to face with people and how people interact with it in a human environment.
- #14 3 DOF – directs the robot’s gaze 3DOF – Controls the head orientation 15 DOF more – move the facial features (e.g eyelids, eyebrows, lips and ears) 4 CCD Cameras A lavalier mic. Is worn on the person interacting with kismet to process the vocalization
- #16 Given that a robotic implementation is limited by perceptual, motor and computational resources, we have found that theses social cues are usful in regulating the turn –taking of humans and robots. These limitations force the robot to exchange speaking turns at a slower rate than a regular human. The times specified are the interval time of response after pause after the speech has ended. Humans seem to intuitively read Kismet’s cues and use them to regulate the rate of exchange at a pace where both partners perform well. The robot uses a Kismet-esque babble for its speaking turn. Hence, the envelope displays are used to regulate the dynamics of interaction during the exchange of speaking turns rather than focus on the content of what is said. The same holds true for proto dialogs between human care-givers and their prelinguistic infants.
- #17 Well studied by discourse theorists, human employ avariety of para-linguistic social cues, called envelope displays’ to regulate the exchange of speaking turns. They involve facial displays, gestures, shifts in gaze and eye blinks – such as raising one’s brows and establishing eye contact to relinquish one’s speaking turn’ or looking aside to hold one’s speaking turn even when speech is paused. Given that a robotic implementation is limited by perceptual ‘ motor and computational resources’ we have found that these social cues are useful in regulating the turn-taking of humans and robots. This has proven particularly important for Kismet because its speech processing limitations force the robot to exchange speaking turn s at a slower rate thatn is typical for human adults.
- #18 Invitation of 4 naïve subjects ranged in age from 12 to 28 years old. 2 male and 2 female They were asked to carry on a “play” conversation with the robot. All was video recorded for later analysis. They were told that the robot does not speak nor understands English, but babbles in a characteristic manner. Here I should explain what does it mean to hold the floor ect…
- #21 This table shows that the robot is engaged in a smooth proto-dialog with the human partner the majority of the time (approx. 82.5% of the time) .
- #22 At first they use only the vocals of the robot to gauge their speaking turn They expect the robot to start talking immediately after they finish talking. And the robot is most likely to interrupt them (common at the first phrases the people perform) They call it “Hiccups” in the flow – they appear in short clusters of mutual interruptions and pauses (over 2-4 utterances of the speaker) before the turn phases become coordinated and the flow of the exchange of speaking turns smoothes out. As the subject seem to adjust their behavior according to Kismet’s envelope displays these “hiccups” within speaking turns become less frequent. As we can see in this table, for each subject there are progressively onlger runs of cleanly exchanged turns as time progresses. The flow of communication becomes smoother (e.g. fewer interuuptions, pauses, and significant flow disturbances), as people read and entrain to Kismet’s envelope displays. Then the rate of vocal exchange is well matched to the robot’s perceptual limitations.
- #23 This table shows that the robot is engaged in a smooth proto-dialog with the human partner the majority of the time (approx. 82.5% of the time) .