cdiLect1 - The University of Edinburgh · 2015. 9. 21. · 9/21/15 4 Introduction to Social...
Transcript of cdiLect1 - The University of Edinburgh · 2015. 9. 21. · 9/21/15 4 Introduction to Social...
9/21/15
1
Kaneda71 2008
Case Studies in Design Informatics 1 & 2 Jon Oberlander
Lecture 1: Overview and Introduction to Social Robotics
Slides quote or paraphrase cited papers
http://www.inf.ed.ac.uk/teaching/courses/cdi1/
2
Structure of lecture
1. Overview of Case Studies Course – Goal – Structure – Assessment
2. Introduction to Social Robotics by Design – Bartnek and Forlizzi 2004 – Fong, Nourbakhsh and Dautenhahn 2003 – Summary
3 Kaneda71 2008
9/21/15
2
Courtesy Boris Lau
9/21/15
3
Course Goal
! How would you do it differently?
! Every time a design decision is made to pursue one course of action, other routes are closed off.
! The goal is: – to work in groups to see why specific project design decisions
were taken, and – to envisage a different service or product that could be built
from the same components.
9
Course Structure
! There are two main linked case studies: – Human robot interaction in the JAMES project – Personal data obfuscation in the SOCIAM project
! The link is this: – How can design elicit the “correct” human behaviour?
! Assessment is by term paper only; there is no final exam. – A1 30%: first term paper on HRI (Group). – A2 40%: second term paper on PDO (Group). – A3 30%: third term paper on implementation & learning (Group).
! Feedback: – Summative:
• Term papers will be marked, and written feedback given
– Formative: • All tutorials provide formative verbal feedback; • 1 tutorial provides formative written feedback on draft A2 reports
10
Course Assessment
! Assignment 1: – Start: Week 2, Monday, 16:00: 28th September. Available from course page. – Submit: Week 4, Thursday, 16:00: 15th October
• (30% of overall coursework grade). – Use Informatics submit, or DVD/thumbdrive to Informatics Teaching Office. – Return: Week 5, Friday, 16:00: 23rd October.
! Assignment 2: – Start: Week 5, Monday, 16:00: 19th October. Available from course page. – Submit: Week 8, Thursday, 16:00: 12th November
• (40% of overall coursework grade).
– Use Informatics submit, or DVD/thumbdrive to Informatics Teaching Office. – Return: Week 9, Friday, 16:00: 20th November.
! Assignment 3: – Start: Week 9, Monday, 16:00: 16th November. Available from course page. – Submit: Week 11, Thursday, 16:00: 3rd December
• (30% of overall coursework grade).
– Use Informatics submit, or DVD/thumbdrive to Informatics Teaching Office. – Return: Week 12, Friday, 16:00: 11th December.
11
Course Timetable
Week Topic Mon Mon Thu Submit 16:00 Thu
1 HRI Intro (JO) <No class>
2 HRI Designing a robot (JO) Tutorial State-of-the-art (JO)
3 HRI Towards JAMES (JO) Tutorial JAMES (JO)
4 HRI HRI vs HDI (JO) Tutorial Animals A1
5 Eval Usability evaluation (JO) Tutorial Usability evaluation (JO)
6 Data Personal data (DMR) Tutorial Personal data (DMR) A2-draft
7 Data Personal data (DMR) Tutorial* Personal data (DMR)
8 Data Personal data (DMR) Tutorial Personal data (DMR) A2
9 New Reflection (JO) Tutorial ADI 1
10 New ADI 2 Tutorial ADI 3
11 New ADI 4 (Tutorial) ADI 5 A3
12
9/21/15
4
Introduction to Social Robotics by Design
1. Bartnek and Forlizzi 2004 2. Fong, Nourbakhsh and Dautenhahn 2003 3. Summary
13
Bartnek and Forlizzi 2004
! A Design-‐Centred Framework for Social Human-‐Robot InteracBon
! Proceedings of the 2004 IEEE InternaBonal Workshop on Robot and Human InteracBve CommunicaBon, Kurashiki, Japan, September 20-‐22, 2004 – Defines social robots – Classifies properBes of social robots
14 Quoting Bartneck and Forlizzi 2004
Robots
! Industrial (e.g. automoBve)
! Professional service (e.g. mining, nuclear)
! Personal service (e.g. home, hospital) – Robust enough to be deployed, but … – “how these robots should behave and interact with humans -‐ act socially -‐
remains largely unclear”
– “Researchers and designers have only just begun to understand these criBcal issues”
15 Quoting Bartneck and Forlizzi 2004
Defining social robots
! The lnternaBonal FederaBon of RoboBcs (IFR) – personal service – A robot which operates semi or fully autonomously to perform services
useful to the well being of humans and equipment, excluding manufacturing operaBons.
! Engelhardt and Edwards 1992 – personal service – Systems that funcBon as smart, programmable tools, that can sense, think,
and act to benefit or enable humans or extend/enhance human producBvity.
! Bartnek and Forlizzi 2004 – social – A social robot is an autonomous or semi-‐autonomous robot that interacts
and communicates with humans by following the behavioral norms expected by the people with whom the robot is intended to interact.
16 Quoting Bartneck and Forlizzi 2004
9/21/15
5
Social robots
! Exclude: – Purely virtual – Teleoperated – Robots that interact only with other robots
! Include: – CooperaBve robots – CompeBBve robots (e.g. in a game)
– Silent partners (as well as talking partners) ! Being social:
– understanding (and in some cases, mimicking) human acBvity, and
– (understanding) the surrounding society and culture, which shapes social values, norms and standards.
– e.g. roboBc butler – follow established rules of good service: • an#cipate, be reliable, be discreet.
17 Quoting Bartneck and Forlizzi 2004
A design-‐centred perspecBve on robot properBes
! Social robots = – products that facilitate co-‐experience and social interacBon.
! Designed form = – the total expression of the product. – not just appearance, but whole experience of interacBng with the product.
18 Quoting Bartneck and Forlizzi 2004
Bartnek and Forlizzi’s Framework
19
robot as “A reprogrammable, multifunctiional manipulator designed to move material, parts, tools, or specialized devices through various programmed motions for the performance of a variety of tasks” and the International Standard Organization (ISO) in IS0 8373 defines a robot as “An automatically controlled, reprogrammable, multipurpose, manipulator programmable in three or more axes, which may be either fixed in place or mobile for use in industrial automation applications.”
These definitions are clearly targeted to autonomous or semi-autonomous industrial robots aind do not take interaction with humans into account aside from the attached safety regulations, such as ANSIIRIA R15.06-1999 and IS012100. The growth in personal service robots necessitates a definition for the kinds of robots that work with people. The lnternational Federation of Robotics (IFR) has adopted a preliminary definition of Service Robots as “A robot which operates semi or fully autono~nously to perform services useful to the well being of humans and equipment, excluding manufacturing operations.”
What is not explicitly mentioned in this definition is the interaction between people and robots, which is mentioned by Engelhardt (Engelhardt & Edwards, 1992) in his definition of service robots as “Systems that function as smart, programmable tools, that can sense, think, and act to benefit or enable humans or extendlenhance human productivity.” This definition speaks more about human productivity and less about social interaction, the goals of which are not always productivity. For example, entertainment robots including products such as the Sony Aibo (Sony, 1999) are not exceedingly productive but are still very valuable to their owners. We would like to propose the following definition of a social robot:
A social robot is an autonomous or semi-autonomous robot that interacts and communicates with hwnans by following the behavioral norms expected by the people with whom the robot is intended to interact.
This definition implies that a social robot has a physical embodiment. Screen characters or any kiind of virtual agent would be excluded by this definition. Rlecently, a class of robots have been developed that use a screen to display the robot’s head (RoboticPerformanceCompany, 2004). Because the screen-based head sets an expectation for and a locus of interaction, it can be considered to be a social robot.
Autonomy is a requirement for a social robot. A semi- autonomous robot can be defined as social if it communicates an acceptable set of social norms. A completely remote controlled robot cannot be considered to be social since it does not make decisions by itself. It is merely an extension of another human.
Communication and interaction with humans is a critical point in this definition. Therefore, robots that only interact and communicate with other robots would not be considered to be social robots. The interaction is likely to be
cooperative, but is not limited to it. Also uncooperative behavior can be considered social in certain situations. The robot could, for example, exhibit competitive behavior within the framework of a game. The robot could also interact with a minimum or no communication. It could, for example, hand tools to an astronaut working on a space station (Goza, Ambrose, Diftler, & Spain, 2004).
In our definition, being social is bound to understanding and in some cases, mimicking human activity and the surrounding society and culture, which shapes social values, norms and standards. For example, a robotic butler should comply with established rules of good service. It should anticipate, be reliable, and most of all, be discreet. However, the precise activities are likely to vary between cultures since social values, norms and standards differ between cultures (Hofstede, 1984). With this definition in place we can now turn to a framework that classifies properties of social robots.
4. The Framework Our framework takes a design-centered perspective, viewing social robots as products that facilitate co-experience and social interaction (Forlizzi & Battarbee, 2004). The framework also focuses on the notion of designed form. Design approaches form as the total expression of the product - not just how something appears, but the whole experience of the interacting with the product. Form includes a product’s physical manifestation, materials, and behavioral qualities (DiSalvo, Gemperle, Forlizzi, & Montgomery, 2003). Designers use form to balance the needs of people, the capabilities of technology, and the context of use into a single product.
Form 1 I I ebsbect womorphii ao”pMnomhrc
Modaltty
Untmodal
Social norma
munimadsl
nohowledge minimal knowledge iull knowladge ot m a l “ 5 ofsocis]” of social I”
Autonomous
Figure 1. Framework for classifying social robots.
Our framework (Figure 1) contains the following properties:
a. form For the purposes of this investigation, we group form
(shape, materials, and behavioral qualities) into three categories that suggest social behavior: abstract, biomorphic
- 592 -
Quoting Bartneck and Forlizzi 2004
Bartnek and Forlizzi’s Framework
! Form – Shape, materials, behavioural qualiBes – Abstract – Biomorphic (lifelike) – Anthropomorphic (humanlike)
! Modality – Number of communicaBon channels
• Varying over one, few, many – Channels include:
• Visual, auditory, hap#c, olfactory, gustatory ! Social norms
– Behaviour is influenced by that of other members of the group – Varying over no norms (e.g. Furby) to reciprocal norms
! Autonomy – Capability to act on behalf of humans, without direct input. – Varying over non, semi, full
! InteracBvity – PotenBal to show causal behaviour (respond to human acBon) – Varying over non, semi, full
20 Quoting Bartneck and Forlizzi 2004
9/21/15
6
Design Guideline 1. The form of the social robot should match its abiliBes.
! Shape, size, and material qualiBes of a social robot should match the task it is designed for to avoid false expectaBons.
! A humanoid robot, for example, is usually expected to have robust speech recogniBon capabiliBes, and users are confused when their expectaBons are not met.
! A biomorphic form, such as a dog or cat, may be more appropriate in sedng expectaBons about the robot's capabiliBes.
21 Quoting Bartneck and Forlizzi 2004
Design Guideline 2. The social robot should mimic human-‐human dialogue in human-‐robot dialogue and be able to manage communicaBon failures.
! Social robots should recognize, respond to, and employ where possible all modaliBes that humans naturally use to communicate.
! These include verbal cues such as speech, intonaBon, and tone of voice, and non-‐verbal cues such as gesture, posture, and stance, among others.
! However, the robot should only communicate states it actually has.
! It should not fake emoBons if it does not genuine[ly] use them for its own benefit.
! Such a fake would be detected and eventually it would be perceived negaBvely.
22 Quoting Bartneck and Forlizzi 2004
Design Guideline 3. The robot should mimic human social norms and be able to provide a consistent set of behaviors.
! Social robots should be aware of human social rules and norms, and grant privilege to them at all Bmes.
! When possible, the robot should be aware of its own social role, its world knowledge, and what it does not know.
! It must be able to deal with uncertainty, and adhere to the ethical principle of least harm.
23 Quoting Bartneck and Forlizzi 2004
Bartnek and Forlizzi’s design guidelines: summary
1. Form should match ability
2. Use human dialogue modaliBes – Do not fake internal states
3. Use human social norms – Respect humans – Be consistent
24 Quoting Bartneck and Forlizzi 2004
9/21/15
7
Fong, Nourbakhsh and Dautenhahn 2003
! A survey of socially interacBve robots
! RoboBcs and Autonomous Systems, 42, 143-‐166. – Reviews “socially interacBve robots” – The forms of social robots
– Design methods, components
– Impact of robots on humans
25 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Another definiBon of social robots
! Dautenhahn and Billard 1999 -‐ Social robots: – Embodied agents that are part of a heterogeneous group: a society of
robots or humans.
– They are able to recognize each other and engage in social interacBons, – they possess histories (perceive and interpret the world in terms of their
own experience), and
– they explicitly communicate with and learn from each other.
26 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Concepts
! Breazeal defines four classes of social robots in terms of: 1. how well the robot can support the social model that is ascribed to it and
2. the complexity of the interacBon scenario that can be supported
27 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Breazeal’s four classes
! Socially evocaBve. – Robots that rely on the human tendency to anthropomorphize and capitalize on
feelings evoked when humans nurture, care, or [are] involved with their “creaBon”.
! Social interface. – Robots that provide a “natural” interface by employing human-‐like social cues and
communicaBon modaliBes. – Social behavior is only modeled at the interface, which usually results in shallow
models of social cogniBon.
! Socially recepBve. – Robots that are socially passive but that can benefit from interacBon (e.g. learning
skills by imitaBon). – Deeper models of human social competencies are required than with social interface
robots.
! Sociable. – Robots that pro-‐acBvely engage with humans in order to saBsfy internal social aims
(drives, emoBons, etc.). – These robots require deep models of social cogniBon.
28 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
9/21/15
8
FND add three classes:
! Socially situated. – Robots that are surrounded by a social environment that they perceive and
react to.
– Socially situated robots must be able to disBnguish between other social agents and various objects in the environment.
! Socially embedded. – Robots that are:
• (a) situated in a social environment and interact with other agents and humans;
• (b) structurally coupled with their social environment; and
• (c) at least par#ally aware of human interac#onal structures (e.g., turn-‐taking).
! Socially intelligent. – Robots that show aspects of human style social intelligence, based on deep
models of human cogniBon and social competence.
29 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
FND also define …
! Socially interacBve robots. – Robots for which social interacBon plays a key role.
! CharacterisBcs: – express and/or perceive emoBons;
– communicate with high-‐level dialogue;
– learn/recognize models of other agents; – establish/maintain social relaBonships;
– use natural cues (gaze, gestures, etc.); – exhibit disBncBve personality and character; – may learn/develop social competencies.
30 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Fong, Nourbakhsh and Dautenhahn’s Landscape
31
T. Fong et al. / Robotics and Autonomous Systems 42 (2003) 143–166 145
Fig. 6. Fields of major impact. Note that “collective robots” and“social robots” overlap where individuality plays a lesser role.
social” collective robots (Fig. 6). In particular, sociallearning and imitation, gesture and natural languagecommunication, emotion, and recognition of interac-tion partners are all important factors. Moreover, mostresearch in this area has focused on the applicationof “benign” social behavior. Thus, social robots areusually designed as assistants, companions, or pets, inaddition to the more traditional role of servants.
1.2. Social robots and social embeddedness:concepts and definitions
Robots in individualized societies exhibit a widerange of social behavior, regardless if the society con-tains other social robots, humans, or both. In [19],Breazeal defines four classes of social robots in termsof: (1) how well the robot can support the social modelthat is ascribed to it and (2) the complexity of the in-teraction scenario that can be supported as follows.Socially evocative. Robots that rely on the human
tendency to anthropomorphize and capitalize on feel-ings evoked when humans nurture, care, or involvedwith their “creation”.Social interface. Robots that provide a “natural”
interface by employing human-like social cues andcommunication modalities. Social behavior is onlymodeled at the interface, which usually results in shal-low models of social cognition.Socially receptive. Robots that are socially passive
but that can benefit from interaction (e.g. learningskills by imitation). Deeper models of human social
competencies are required than with social interfacerobots.Sociable. Robots that pro-actively engage with hu-
mans in order to satisfy internal social aims (drives,emotions, etc.). These robots require deep models ofsocial cognition.Complementary to this list we can add the following
three classes:Socially situated. Robots that are surrounded by a
social environment that they perceive and react to [48].Socially situated robots must be able to distinguishbetween other social agents and various objects in theenvironment.1Socially embedded. Robots that are: (a) situated in
a social environment and interact with other agentsand humans; (b) structurally coupled with their socialenvironment; and (c) at least partially aware of humaninteractional structures (e.g., turn-taking) [48].Socially intelligent. Robots that show aspects of hu-
man style social intelligence, based on deep modelsof human cognition and social competence [38,40].
1.3. Socially interactive robots
For the purposes of this paper, we use the term “so-cially interactive robots” to describe robots for whichsocial interaction plays a key role. We do this, not tointroduce another class of social robot, but rather todistinguish these robots from other robots that involve“conventional” human–robot interaction, such as thoseused in teleoperation scenarios.In this paper, we focus on peer-to-peer human–robot
interaction. Specifically, we describe robots that ex-hibit the following “human social” characteristics:
• express and/or perceive emotions;• communicate with high-level dialogue;• learn/recognize models of other agents;• establish/maintain social relationships;• use natural cues (gaze, gestures, etc.);• exhibit distinctive personality and character;• may learn/develop social competencies.
Socially interactive robots can be used for a vari-ety of purposes: as research platforms, as toys, as ed-ucational tools, or as therapeutic aids. The common,
1 Other researchers place different emphasis on what sociallysituated implies (e.g., [97]).
Quoting Fong, Nourbakhsh & Dautenhahn 2003
Uses of socially interacBve robots
! Socially interacBve robots are important for domains in which robots must exhibit peer-‐to-‐peer interacBon skills, either – because such skills are required for solving specific tasks, or – because the primary funcBon of the robot is to interact socially with people.
! Robot as persuasive machine. – used to change the behavior, feelings or adtudes of humans.
– e.g. when robots mediate human–human interacBon, as in auBsm therapy
! Robot as avatar. – a representaBon of, or representaBve for, the human.
– e.g. a remote communicaBon robot may need to act socially in order to effecBvely convey informaBon.
32 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
9/21/15
9
Design methods: Biologically inspired
! Robots that internally simulate social intelligence – To be understandable (by users), must have naturalisBc embodiment,
interact with environment, and perceive salience as humans do.
– To allow tesBng of scienBfic theories • e.g. ethology, dialogue, joint aAen#on, developmental psychology
33 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Design methods: FuncBonally designed
! Robots that merely appear socially intelligent – Only need superficial social competence.
• e.g. only for short-‐term interac#on
– Limited embodiment, or capability for interacBon. • Or may be constrained by the environment.
– Even limited social expression improves usability. • e.g. Recorded or scripted speech sufficient.
– ArBficial designs can provide compelling interacBon. • e.g. Many video games engage, even no real-‐world counterparts.
34 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
FuncBonal design methods
! HCI design. – e.g. cogniBve modelling, heurisBc evaluaBon
! Systems engineering. – e.g. criBcal path design
! IteraBve design. – e.g. sequenBal improvement via formaBve evaluaBon
35 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Design issues
! All robots need mulBple design problems to be solved: – CogniBon, percepBon, acBon, interacBon, architecture
! Socially interacBve robots also need to address social interacBon: – Human-‐oriented percepBon
• Perceive, interpret human ac#vity, including gestures, and other social cues to intent and human state
! Natural human-‐robot interacBon – For peer interacBon, robot must express believable behaviour, sedng and
meeBng social expectaBons, regulaBng interacBon
! Readable social cues – Express signals to
• Provide feedback on internal state • Allow transparent interac#on
– Channels for [emoBonal] expression include face, body, gesture, voice.
! RealBme performance – Solve all of these problems (and task behaviour) at human interacBon rates.
36 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
9/21/15
10
Embodiment
! The more a robot can perturb an environment, and be perturbed by it, the more it is embodied.
! This also means that social robots do not necessarily need a physical body.
! For example, conversaBonal agents might be embodied to the same extent as robots with limited actuaBon.
! e.g. Aibo is more embodied than Khepera: – more actuators, more sensors
– so, more perturbatory channels and bandwidth
37 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Embodiment: morphology
! Form and structure help set (iniBal) social expectaBons – e.g. dog versus humanoid
! RelaBve familiarity -‐> accessibility, apracBveness, expressiveness
! Form may constrain interacBon opBons – e.g. head vs torso vs mobile plaqorm
! To date, liple work has focussed on industrial design.
38 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Embodiment: design
! Morphology must match funcBonal category – Robot as Tool -‐> productness – Robot as Partner -‐> humanness
! Morphology must control expectaBons – Both -‐> robotness
! Morphology must maintain Familiarity via Similarity – While avoiding the “uncanny valley” (Mori)
39 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
Embodiment design
! Embodiment design 1: anthropomorphic – [Humans establish a peer-‐peer or carer-‐charge relaBonship]
– To interact as humans do, need structural and funcBonal similarity.
– To learn (with feedback), must behave as humans do.
– Requires balance of illusory sophisBcaBon and real funcBonality
! Embodiment design 2: zoomorphic – Animals establish an owner-‐pet relaBonship
• e.g. dogs, cats, companions
– Avoids the uncanny valley
! Embodiment design 3: caricatured – Disney showed that realism not needed for believability
– Caricature biases user apenBon towards some features (and away from others) • e.g. cartoon face as “focal point” for apenBon
! Embodiment design 4: funcBonal – Alternately, robot form follows only physical funcBon (task)
• e.g. service robots in health care; toy robots in play
40 Quoting Fong, Nourbakhsh &
Dautenhahn 2003
9/21/15
11
Summary: definitions
! Bartnek and Forlizzi 2004 – social – A social robot is an autonomous or semi-‐autonomous robot that interacts
and communicates with humans by following the behavioral norms expected by the people with whom the robot is intended to interact.
! Breazeal – sociable – Robots that pro-‐acBvely engage with humans in order to saBsfy internal
social aims (drives, emoBons, etc.) [and require deep cogniBve models]
! Fong et al. 2003 – socially intelligent – Robots that show aspects of human style social intelligence, based on deep
models of human cogniBon and social competence.
! Fong et al. 2003 – socially interacBve – Robots for which social interacBon plays a key role.
41
Summary: guidelines
! Bartnek and Forlizzi 2004 – guidelines 1. Form should match ability
2. Use human dialogue modaliBes • Do not fake internal states
3. Use human social norms • Respect humans
• Be consistent
! Fong et al. 2003 -‐ guidelines 1. Morphology must match funcBonal category
• Robot as Tool -‐> productness
• Robot as Partner -‐> humanness
2. Morphology must control expectaBons • Both -‐> robotness
3. Morphology must maintain Familiarity via Similarity • While avoiding the “uncanny valley” (Mori)
42
43