SOCIAL ARTIFACTS Mustafa Suphi Erden Cognitive Robotics Laboratory ENSTA ParisTech

Transcription

1 SOCIAL ARTIFACTS Mustafa Suphi Erden Cognitive Robotics Laboratory ENSTA ParisTech

2 What are we doing in (S)HRI? Utility + Aesthetics = (S)HRI 2 65

3 Content 1. Automatons 2. First Calculators 3. Cybernetics and Artificial intelligence Classical Background 4. Robotics 1. Social Human-Robot Interaction 2. Utility + Aesthetics 3. Best Examples 4. Work at ENSTA Emotional postures Developing games Emergent Field 3 65

4 Antique Art 4 65

5 Arabic Automatons (8th 13th C) Greek designs --> Aesthetics Dramatic illusion Arabic automatons --> Utility Practical usage Changing the environment Serving for humans 5 65

6 Al-Jazari ( ) Ebû l İz İbni İsmail İbni Rezzaz El Cezerî Arabic mathematician-mechanician Lived in Diyarbakir Kitáb fí ma'rifat al-hiyal al-handasiyya (Book of Knowledge of Ingenious Mechanical Devices, 126) More an engineer than an inventor Trial and error, rather than theoretical approach 6 65

7 Al-Jazari Hand washing machine with flushing mechanism Elephant clock Automatic music band 7 65

8 Al-Jazari Crank shaft Saqiya Machines provided water to Damascus since 13th C. Water pump 8 65

9 Automatons (13th 19th C) Self working mechanisms No electronics Imitating human or animals 19th century, Japan, Tea Serving Lady 9 65

10 Renaissance and afterwards Leonardo da Vinci (1495) designed an automaton Moving the arms Shaking head Able to sit 16.th C Automatic Clocks Descartes (17th C): Animal body is just a complex machine Bones, muscles, and organs Pistons, gears, pipes (--> Robot Inspired Biology!?) : Golden age of automatons 1 65

11 Jacques de Vaucanson ( ) French mechanician 1737: Flute player 12 different pieces Tambur player --> Pure Aesthetics 1745: Automatic weaving machine --> Pure Utility Punched card Digesting Duck --> Pure Imitation 11 65

12 Digesting Duck Jacques de Vaucanson (1739) 4 moving parts Flapping Drinking water Eating and digesting crop Defecating (No real model) 12 65

13 Digesting Duck Seems to digest (!) The crops are accumulated in a box Stored feces is thrown away No real digestion! (No real model) 13 65

14 First calculators --> Imitating the intellect Blaise (1642) Pascal Pascal calculator Automatic calculation of carries 14 65

15 Leibniz calculator G.W. Leibniz (1673) Calculus Ratiocinator Addition and subtraction multiplication and division The mechanization of human thinking 15 65

16 Mind Numbers Gottfried Wilhelm Leibniz ( ) Express the human thinking with numbers human intelligence can be mechanized Thinking machine can be built Binary counting system 16 65

17 Industrial Revolution Before Automatons Machines imitating humans One machine multiple human functions Industrial Revolution (18s) Machines No more looking like humans Realizing functions One machine function of a lot of people 17 65

18 Cybernetics (196s) The physics of Thinking and Mind Philosophers No technology Cyberneticists open to all technologies Neural networks and mind Biological considerations: Cybernetic - Neurophysiology 18 65

19 Artificial Intelligence (AI) Human thinking Social sciences no application to the world of numbers AI aims at projecting thinking to the world of numbers Input: Symbolic representations Program Output: Symbolic representations 19 65

20 AI 1956, Herbert Simon ve Alien Newell Logic Theorist Rule based symbolic operations 198s Neural Networks Imitating the self organization of the brain Intelligence is bound to the brain 2 65

21 Deep Blue Chess player On 11 May 1997, did beat Garry Kasparov Written in C 2 million positions in a second Today s chess playing computers are more intelligent Deep Fritz 8 million positions per second, but in average it sees turns ahead 21 65

22 Robotics Czech language: robota forced labor Karel Čapek s play Rossum s Universal Robots, 1921 Isaac Asimov (1941) Robotics Robotics: A branch of technology 22 65

23 Brooks (22) (Criticism of AI) It is too ambitious to imitate the human intellect by artificial bodies Chauvinism of computation Ignores the biological processes, neurotransmitters, hormones 23 65

24 Intelligence and Environment Is intelligence internal to the brain? What is its relation with the environment? An anthropoid can develop more intelligence with a bone or cutting tool in its hand Intelligence of human lags the technology related to that 24 65

25 Top-down vs. Bottom-up Top-down Programming Transferable Imitating human thinking Bottom-up Adaptation-learning Intelligence Emerging and evolving process In relation to the environment Evolving, becoming intelligence in machines 25 65

26 Language Nobody knows how it evolved Enlargement of the brain, development of the throat, social interactions, communication technologies The interaction of robots can develop a language like structure No analytical clue of how it is developed 26 65

27 Robotics Today Behavior + Environment = Intelligence Where is AESTHETICS? 27 65

28 Social Human-Robot Interaction Social robots are embodied agents that are part of a heterogeneous group: a society of robot or humans. They are able to recognize each other and engage in social interactions. They possess histories, perceive and interpret the world in terms of their own experience. They explicitly communicate with and learn from each other. (Dautenhahn and Billard, 1999) 28 65

29 Interdisciplinary (Fong et al., 23) 29 65

30 HRI: New Application Domain Problem Domain Behavioral and social sciences Psychology Techniques Engineering Robotics, control, image processing, speech processing 3 65

31 Utility + Aesthetics Imitating human appearance Imitating human actions Functional for achieving tasks Functional for attracting humans 31 65

32 Best Examples 32 65

33 Hiroshi Ishiguro DER 1 Actroid Hiroshi Ishiguro, Osaka University 33 65

34 34 65

35 Perspective taking video_pers_tak (Brazeal et al., 29) 35 65

36 36 65

37 Attractiveness (Kanda et al., 28) 37 65

38 Recognition of Friendship (Kanda et al., 28) 38 65

39 39 65

40 Affect modeling of the children (Liu et al., 28) 4 65

41 Adaptation to the affect of the child (Liu et al., 28) 41 65

42 Our Work at ENSTA 42 65

43 Goal Interaction with children with autism through games 43 65

44 Utility + Aesthetics Functions: Engaging in the game (perception, decision making) Helping the child to concentrate think attend Imitations: Humanoid Head tracking Emotional postures 44 65

45 Architecture of Interaction 45 65

46 NAO Robot 46 65

47 Coulson's work (24) 47 65

48 48 65

49 Adaptation of Human Postures Problems: 1) Kinematic constraints less degrees of freedom, less range of joint angles 2) Inertial constraints, mass distribution the head of Nao is heavy, unstable postures 49 65

50 Adaptation of Human Postures Problems: 3) Reference framing different than that of the Careograph 4) Size of the robot different than size of an adult human 5 65

51 Transformation of joint Angles Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Shoulder swing Elbow bend HeadYaw PP Abdomen Shoulder Chest bend Head bend t wist ad/abduct (- 2 --> - 1 ) Happiness RhipPit ch HA Weight T ransfer Nao Robot IN ES S Human Weight Transfer Postures Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards Forwards -Chest Bend Abdomen + RHipPitch_ twist increment Nao Joints LShoulderLShoulder- RshoulderPit ch / LhipPit ch HeadPit ch RelbowYaw LelbowYaw RelbowRoll LelbowRoll Roll Roll RshoulderPit ch Transformation equations based on Coulson's data -Chest Bend max{elbow min{-elbow Constant Constant + LHipPitch_ Head Bend -Shoulder +Shoulder +Shoulder bend, 95} bend, -95} increment ad/abduct ad/abduct swing Nao Joint Values

52 Videos video_happiness video_sadness video_anger 52 65

53 Results of Interview Without the context information No consistent identification! Implemented emotions Anger Anger Happiness Sadness Ekman's six emotions (Ekman and Friesen, 1984) Happiness Sadness Fear Surprise Disgust With context information Participants agreed about the associated emotional content of the postures Participants identified the postures strongly conveying the emotion 53 65

54 Anger 54 65

55 Happiness 55 65

56 Sadness 56 65

57 Head Tracking video_head_tracking 57 65

58 How to use? Games Robot wins --> Happy Robot looses --> Sad Robot is unlucky --> Angry 58 65

59 Scenario_1: Recognition Game Whether the child recognizes the difference between the sad and happy behaviors of the robot 59 65

60 Scenario_2: Association Game Whether the child associates the three emotional postures with the three feelings of the robot 6 65

61 Scenario_3: Inverse Imitation Game The imitation capability of the child is testified. Head Movements 61 65

62 Scenario_4: Teaching Game Robot is pointed to the objects and taught about their names by the child BALL BOX BAG 62 65

63 Human-Robot Interaction For ASSISTING HUMANs Not for replacing parents Not for promoting a lonely elderly Not for developing guns for military 63 65

64 Thank you for your attention

65 References 1) Erden, M.S and Tapus, A, (21), Postural expressions of emotions in a humanoid robot for assistive applications, Poster paper in Workshop on Learning for Human-Robot Interaction Modeling under the Conference Robotics Science and Systems 21. 2) Coulson, M., (24), Attributing emotion to static body postures: recognition accuracy, confusions, and viewpoint dependence, Journal of Nonverbal Behavior, 28 (2): ) Niran B. Abbas (26), Thinking Machines: Discourses of Artificial Intelligence, Hamburg: Lit Verlag. 4) 5) Fong, T., Nourbakhsh, I., and Dautenhahn, K., (23), A survey of socially interactive robots, Robotics and Autonomous Systems, 42: ) Breazeal, C., (23), Emotion and sociable humanoid robot, Int. J. of Human-Computer Studies, 59: ) Liu, C., Con, K., Sarkar, N., and Stone W., (28), Online affect detection and robot behavior adaptation for intervention of children with autism, IEEE Trans. On Robotics, 24 (4): ) Kanda, T., Nabe, S., Hiraki, K., Ishiguro, H., and Hagita, N., (28), Human friendship estimation model for communication robots, Autonomous Robots, 24: