Human-Robot Interaction Elective in Artificial Intelligence - - PDF document

Human-Robot Interaction

Elective in Artificial Intelligence Lecture 1 - Introduction

Luca Iocchi DIAG, Sapienza University of Rome, Italy

[Goodrich&Schultz 2007]

• M. A. Goodrich and A. C. Schultz.

Human-Robot Interaction: A Survey. Foundations and Trends in Human-Computer Interaction, 1(3), 2007, pp. 203-275. https://faculty.cs.byu.edu/~mike/mikeg/papers/HRISurvey.pdf [Fong et al. 2003] A Survey of Socially Interactive Robots. Robotics and Autonomous Systems, 42, 2003, pp. 143-166. http://www.cs.cmu.edu/~illah/PAPERS/socialroboticssurvey.pdf

• L. Iocchi - Human-Robot Interaction

2

Readings

• What is HRI
• History
• Design features
• Applications
• Benchmarking
• Social Robots
• RoboCup@Home
• L. Iocchi - Human-Robot Interaction

3

Outline

• L. Iocchi - Human-Robot Interaction

4

What is HRI ?

• Interaction: the process of working together to

accomplish a goal

• HRI: “Understand and shape the interactions between
• ne or more humans and one or more robots”

[Goodrich&Schultz 2007]

• Analysis, design, modeling, implementation and

evaluation of robots used in applications involving humans Have you been involved in a human-robot interaction task?

• L. Iocchi - Human-Robot Interaction

5

What is HRI ?

• L. Iocchi - Human-Robot Interaction

6

What is HRI ?

• Remote interaction: humans and robots

are separated spatially or temporally

• Proximate interaction: humans and robots

share the same physical space

• Increased interest in applications involving HRI

(natural interactions with non-expert users).

• Deployment of robots in public spaces (houses, offices,

schools, roads, hotels, airports, shopping malls, hospitals, …)

• Social robotics/social intelligence.
• L. Iocchi - Human-Robot Interaction

7

Motivation for HRI

• Engineering (mechanical, electrical, …)
• Computer science (artificial intelligence, robotics,

human-computer interaction, natural language understanding, computer vision, …)

• Design (product design, interaction design, …)
• Social sciences (psychology, cognitive science,

communications, human factors, …)

• Humanities (ethics, philosophy, …)
• L. Iocchi - Human-Robot Interaction

8

Multi-disciplinary HRI

• Automata and mechanical creatures in ancient Egypt, Greece,

and China

• Golem – artificial beings
• 1495 Leonardo’s Mechanical man
• 1800 Radio controlled machines (e.g., Tesla’s boat)
• 1920 Karel Chapek’s play Rossum’s Universal Robots
• L. Iocchi - Human-Robot Interaction

9

History

• 1969 – Shakey project
• 1992 – First IEEE International

Symposium on Robot & Human Interactive Communication (RoMan)

• 1999 IEEE RAS TC on Human-

Robot Interaction & Coordination

• 2006 ACM International Conference
• n Human-Robot Interaction
• 2006 RoboCup@Home
• 2009 Intern. Conf. on Social Robotics
• L. Iocchi - Human-Robot Interaction

10

History

[Goodrich&Schultz 2007]

• Autonomy
• Information Exchange (Communication media and

format)

• Teams (human-robot teams)
• Adaptation, Learning, and Training
• Task-Shaping
• L. Iocchi - Human-Robot Interaction

11

Design features

• 1. R offers no assistance; human does it all.
• 2. R offers a complete set of action alternatives.
• 3. R narrows the selection down to a few choices.
• 4. R suggests a single action.
• 5. R executes that action if human approves.
• 6. R allows the human limited time to veto before automatic execution.
• 7. R executes automatically then necessarily informs the human.
• 8. R informs human after automatic execution only if human asks.
• 9. R informs human after automatic execution only if it decides too.
• 10. R decides everything and acts autonomously, ignoring the human.

[Sheridan, Verplank, 1978]

• L. Iocchi - Human-Robot Interaction

12

Autonomy

• Full-Autonomy (level 10)
• no human in the loop / no HRI
• Direct Tele-operation (levels 1-3)
• high workload / not scalable / for expert users
• High-level interaction (levels 4-9)
• Multi-modal communications
• Partial autonomy
• Suitable for non-expert users
• Adjustable autonomy (levels 4-9)
• Autonomy levels change and adapt during operation
• L. Iocchi - Human-Robot Interaction

13

Autonomy and HRI

• L. Iocchi - Human-Robot Interaction

14

Autonomy in soccer robots

Full tele-operation vs. full autonomy

• L. Iocchi - Human-Robot Interaction

15

Autonomy in soccer robots

• L. Iocchi - Human-Robot Interaction

16

Information Exchange

How information is exchanged

• Communication media
• Vision (visual displays, gestures, body movements)
• Audio (speech, natural language)
• Touch (physical contacts, haptics)
• Multi-modal (integrating many modalities)
• Communication format
• Scripts/Formal languages
• Full natural language
• Subset of natural language
• Social behaviors/interactions
• L. Iocchi - Human-Robot Interaction

17

Information Exchange

Measures of efficiency

• Interaction time
• Cognitive workload
• Situation awareness
• Shared understanding
• L. Iocchi - Human-Robot Interaction

18

Information Exchange

• L. Iocchi - Human-Robot Interaction

19

Vision Processing

• Face detection / tracking / recognition (including expressions)
• Person detection / tracking / recognition
• L. Iocchi - Human-Robot Interaction

20

Vision Processing

• Virtual faces

• L. Iocchi - Human-Robot Interaction

21

Audio Processing

• Speech recognition / synthesis
• L. Iocchi - Human-Robot Interaction

22

Physical interaction

• Physical human-robot interaction (pHRI)
• Wearable robots

• L. Iocchi - Human-Robot Interaction

23

Human-Robot Teams

HRI = multi-agent system (at least one robot and one person) Organization of the team

• Who takes decisions?
• Human (user or designer)
• Robot
• Interface software
• Which levels of robot commands/instructions are used
• strategic
• tactical
• operational
• L. Iocchi - Human-Robot Interaction

24

Human-Robot Teams

• Conflict resolution
• Role definition
• robot peer
• robot assistant
• robot slave
• Role assignment
• static
• dynamic (changes in responsibilities, authorities, roles)
• L. Iocchi - Human-Robot Interaction

25

Human-Robot Teams

• Multi-agent coordination
• Centralized
• Distributed
• Coordination protocol
• Coordination with other software agents
• Joint intentions and joint actions
• L. Iocchi - Human-Robot Interaction

26

Human-Robot Teams

• Robot Adaptation, Learning, and Training
• Human Adaptation, Learning, and Training
• Mutual adaptation for long-term interactions
• Learning method
• Supervised
• Semi-supervised
• Reinforcement
• L. Iocchi - Human-Robot Interaction

27

Adaptation, Learning, Training

• How the task should be performed
• Robots help users to perform a certain task =>

The way in which the task is performed changes.

• Analysis and design of HRI task execution
• Different from Human-Human Interaction (HHI)
• L. Iocchi - Human-Robot Interaction

28

Task Shaping

(Ordered by decreasing requirements for social skills)

• Home companion
• Assistive / health-care / elderly-care / rehabilitation
• Edutainment
• Tour guide, office assistant
• Agriculture, cleaning
• Industrial partners
• Search and rescue, firefighting, surveillance
• Space
• L. Iocchi - Human-Robot Interaction

29

Application fields

• Supervisor (monitor and control the overall situation)
• Operator (operator intervenes in sw when needed)
• Mechanic (operator intervenes in hw when needed)
• Peer (high-level interaction about intentions)
• Bystander (low-level interaction about a subset of actions)
• Mentor (robot has teaching leadership)
• Information consumer (human uses information coming

from the robot) [Scholtz et al., 2002]

• L. Iocchi - Human-Robot Interaction

30

Human/Robot roles in HRI

What to measure?

• Robot features
• Safety
• Scalability wrt complex situations in public environments
• Single functionalities
• Social interaction
• Autonomy
• Imitation
• Privacy
• L. Iocchi - Human-Robot Interaction

31

Benchmarking HRI

What to measure?

• Task-Oriented benchmarks
• Success
• Understanding of the domain by the robot
• Understanding robot abilities (and limitations) by human
• Assistive evaluation
• Usability
• Impact
• Satisfaction
• Quality of Live measures
• L. Iocchi - Human-Robot Interaction

32

Benchmarking HRI

A series of questionnaires to measure the user’s perception

• f robots. Rates user's impression of the robot about
• Anthropomorphism
• Animacy
• Likeability:
• Perceived Intelligence
• Perceived Safety

[Bartneck et al. 2009]

• L. Iocchi - Human-Robot Interaction

33

Godspeed Questionnaire

• Robotics
• Autonomous mobile robots
• Actuated faces / manipulators
• Artificial Intelligence
• Knowledge representation
• Planning
• Computer Vision
• Person / face

detection/recognition/tracking

• Gesture recognition
• Machine Learning
• Adaptive behaviors
• Imitation
• Natural Language

Processing

• Speech recognition and

synthesis

• Dialogues
• L. Iocchi - Human-Robot Interaction

34

Relation to other disciplines

[Fong et al. 2003]

• Robots embedded in a society of humans and robots that

are expected to behave by following social norms.

• Different levels of “sociality” (social skills)
• Express/recognize emotions
• Express/recognize personality
• Towards autonomous robot peers interacting with humans

in a natural way with high degree of flexibility and adaptability.

• L. Iocchi - Human-Robot Interaction

35

Social Robots

• Biological inspiration
• Simulate or mimic social behaviors of biological systems
• Based on theories from natural sciences
• Nature is the best model for artificial systems
• Functional design
• External resembling social intelligence
• Internal mechanisms not based on science and nature
• L. Iocchi - Human-Robot Interaction

36

Design of Social Robots

• Robot morphology must match its intended function
• Morphology has a high impact in the way humans will

interact with robots

• Familiarity/acceptance vs. human similarity

(uncanny valley)

• L. Iocchi - Human-Robot Interaction

37

Robot Morphology

• Uncanny valley
• L. Iocchi - Human-Robot Interaction

38

Robot Morphology

• Artificial emotions to
• facilitate human-robot interactions
• provide feedback to the user
• used as a control mechanism
• Facial expressions
• Actuated robotic faces
• 3D rendered virtual faces
• L. Iocchi - Human-Robot Interaction

39

Emotions/facial expressions

• Robotic faces
• L. Iocchi - Human-Robot Interaction

40

Emotions/facial expressions

• Robotic faces
• L. Iocchi - Human-Robot Interaction

41

Emotions/facial expressions

• Personality traits identifying a single human/robot
• Robot Personality increases willingness of interacting
• Recognition of human personality increases effectiveness
• f interaction
• L. Iocchi - Human-Robot Interaction

42

Personality

• HRI challenging and expanding research field for

Artificial Intelligence and Robotics

• Multi-disciplinary approaches
• Many applications
• A lot of work to be done!
• L. Iocchi - Human-Robot Interaction

43

Conclusions

[Bartneck et al. 2009] Bartneck, C., Croft, E., Kulic, D. & Zoghbi, S. (2009). Measurement instruments for the anthropomorphism, animacy, likeability, perceived intelligence, and perceived safety of robots. International Journal of Social Robotics, 1(1) 71-81. | DOI: 10.1007/s12369-008-0001-3 [Scholtz et al., 2002] J. Scholtz, M. Theofanos, and B. Antonishek, Theory and evaluation of human robot interactions, in 36th International Conference on Systems Sciences, Hawaii: IEEE, 2002. [Sheridan, Verplank, 1978] T. B. Sheridan and W. L. Verplank, Human and Computer Control for Under-sea Teleoperators. MIT Man-Machine Systems Laboratory, 1978.

• L. Iocchi - Human-Robot Interaction

44

References