takayama rss2009 workshop

57
Toward a Science of Robotics: Goals and Standards for Experimental Research Leila Takayama Human-Robot Interaction Research Scientist

Upload: willow-garage

Post on 15-Nov-2014

122 views

Category:

Documents


0 download

DESCRIPTION

Leila Takayama's presentation for "Toward a Science of Robotics: Goals and Standards for Experimental Research." This presentation was delivered at the Workshop on Good Experimental Methodology in Robotics, RSS 2009.

TRANSCRIPT

Page 1: Takayama RSS2009 Workshop

Toward a Science of Robotics:Goals and Standards forExperimental Research

Leila TakayamaHuman-Robot InteractionResearch Scientist

Page 2: Takayama RSS2009 Workshop

Scientific Principles

•Hypothesis testing•Observable, empirical, and

measurable evidence•Reliable•Reproducible and falsifiable

Page 3: Takayama RSS2009 Workshop

Science & Technology

Page 4: Takayama RSS2009 Workshop

Variable-based Research

X → Y

“analog, keyboard and receiver input, high fidelity,two-way processing technologies”

Page 5: Takayama RSS2009 Workshop

Examples of Variables

•System complexity•Interactivity•Similarity to humans•Task types: collaborative, competitive•Synchronicity of interaction•Collocation

Page 6: Takayama RSS2009 Workshop

Study 1 Self Extension into Robots

Page 7: Takayama RSS2009 Workshop

Research Questions– What aspects of human-robot interface

design affect feelings of attachment, trust,control, responsibility, and agency inhuman-robot interaction?

– Does one’s sense of self-extension increasewhen a robot is built by the user?

– Does one’s sense of self-extension differbetween more or less anthropomorphicrobots?

Page 8: Takayama RSS2009 Workshop

Manipulations

Factor 1 (between): car vs. robot

Factor 2 (between): Use of their assembledrobot/car or a preassembled robot/car

Factor 3 (between): assembly vs. noassembly

Page 9: Takayama RSS2009 Workshop

HypothesesRobot form

Humanoid form is a strong cue ofidentity– H1. People will self extend more into the

car robot than the humanoid.

People extend their positive selfconcept into self-extended objects

– H2. People will prefer the personality ofthe car robot over the humanoid.

9

Page 10: Takayama RSS2009 Workshop

HypothesesRobot assembler

Building an object promotes self extension– H3. People will self extend more into a

robot they assemble than a robotassembled by another.

People extend their positive self concept intoself-extended objects

– H4. People will prefer the personality of arobot they assemble over a robotassembled by another.

10

Page 11: Takayama RSS2009 Workshop

Procedure• Participants fill in pre-questionnaire

• Participants given instructions anddiagrams on how to assemble robot

• Participants assemble robot– (M=9 min 49 sec, SD=3 min, 34 sec)

• Participants turn on robot and test it– Tethered control with on/off button

11

Page 12: Takayama RSS2009 Workshop

Procedure

12

Task game

• Goal is to collect most points in 10 minutes• Bombs sometimes explode when touched• Bomb detonations deduct 30 seconds• Bomb number and time controlled• Questionnaire

Page 13: Takayama RSS2009 Workshop

Assembler Manipulation

•Manipulating assembler– Self: Built a robot, operated same robot– Other: Built a robot, participants told they

needed to operate a different, identicalrobot

•In fact, all participants operated therobot they built

13

Page 14: Takayama RSS2009 Workshop

Robot Form Manipulation

Page 15: Takayama RSS2009 Workshop

Measures:Self extension

Trait overlap– Personality similarity of self and other

•Galinsky and Moskowitz– Overlap in concepts of self and human other

•Kiesler and Kiesler– Self extension into objects

15

Self Other

Page 16: Takayama RSS2009 Workshop

Measures:Self extension

Determining trait overlap– Thirty item modified Wiggin’s personality test

•Completed by participants about themselves before task•Completed by participants about robot after task

– Delta of items calculated, summed to index•(Cronbach’s α=.86)

– Smaller scores indicate greater overlap ofconcepts of self and robot

16

Page 17: Takayama RSS2009 Workshop

Measures:Self extension

Self reports• 10 point scales asking about “the device you

guided through the minefield”• Robot control (α=.83)

– Who was more responsible for your generalperformance on this task?

– Who had more control over your generalperformance on this task?

• Sense of team– “I felt that the robot and I were a team.”

17

Page 18: Takayama RSS2009 Workshop

Measures:Robot personality

•Robot friendliness– Nine item index (α=.90)– cheerful, enthusiastic, extroverted

•Robot integrity– Five item index (α=.73)– Honest, reliable, trustworthy

•Robot malice– Five item index (α=.74)– Dishonest, unkind, harsh

18

Page 19: Takayama RSS2009 Workshop

Results:Self extension

Trait overlap

19

F(1, 52)=4.04, p<.05, partial η²=.13

Greater trait overlap with car robots than humanoids

Page 20: Takayama RSS2009 Workshop

Results:Self extension

Robot control

20

Greater relative control attributed to humanoids thanto cars

F(1, 52)=5.47, p<.05, partial η²=.10

Page 21: Takayama RSS2009 Workshop

Results:Self extension

Sense of team

21

F(1, 52)=8.34, p<.01, partial η²=.14

Self-assembly participants felt more like a team with the robotthan did other-assembly participants

Page 22: Takayama RSS2009 Workshop

Measures:Robot Personality

Robot friendliness

22

F(1, 52)=4.25, p<.05, partial η²=.08 F(1, 52)=4.23, p<.05, partial η²=.08F(1, 52)=4.23, p<.05, partial η²=.08η²=.08η²

Car robots were friendlier thanhumanoids

Self-assembled robots were friendlierthan robots assembled by others

F(1, 52)=4.25, p<.05, partial η²=.08η²=.08η²

Page 23: Takayama RSS2009 Workshop

Measures:Robot Personality

Robot integrity

23

Car robots were rated as having more integrity than humanoids

F(1, 52)=4.20, p<.05, partial η²=.08

Page 24: Takayama RSS2009 Workshop

Measures:Robot Personality

Robot malice

24

F(1, 52)=8.94, p<.01, partial η²=.15 F(1, 52)=4.78, p<.05, partial η²=.08

Humanoid robots were moremalicious than cars

Robots assembled by otherswere more malicious than

self-assembled robots

F(1, 52)=8.94, <.01, partial η²=.15η²=.15η² F(1, 52)=4.78, p<.05, partial η²=.08η²=.08η²

Page 25: Takayama RSS2009 Workshop

Summary of Results

H1. People will self extend more into the carrobot than the humanoid.

H2. People will prefer the personality of thecar robot over the humanoid.

25

Page 26: Takayama RSS2009 Workshop

Summary of Results

H3. People will self extend more into a robotthey assemble than a robot assembled byanother.

H4. People will prefer the personality of arobot they assemble over a robot assembledby another.

26

Page 27: Takayama RSS2009 Workshop

Design Implications

Goal-specific guidelines

– No form, assembly experience is uniquelyoptimal

– Desirability of self extension informs design

27

Page 28: Takayama RSS2009 Workshop

Design Implications

•When self extension is desired– Tele-operated robots as media, human

representations•Medical care, remote therapy

– Non-humanoid form– Promote pre-mission interaction

•Assembly, customization

28

Page 29: Takayama RSS2009 Workshop

Design Implications

•When self extension is undesirable– Robots in hostile environments, likely

failures•Search and rescue

– Humanoid form– Minimize pre-mission interaction

•Identical but different robots– Change robot’s name

•Altered robots– Change voice, appearance

29

Page 30: Takayama RSS2009 Workshop

Limitations and Next Steps

• Broader population• Outside the lab• Using other robots

• Long-term interactions• Long-term effects• Balancing needs of people operating and

encountering robot

30

Page 31: Takayama RSS2009 Workshop

Study 2Disagreeing Robots

Page 32: Takayama RSS2009 Workshop

Why would a robot everdisagree with a person?

Page 33: Takayama RSS2009 Workshop

Research Questions

•What influences an interface’s point ofinteraction? Body location? Voicelocation?

•(How) do politeness strategies fromhuman-human interaction informhuman-computer interaction?

Page 34: Takayama RSS2009 Workshop

Design Questions

•What influences a robot’s point ofinteraction?

•Where should speakers be placed?•(How) can computer agents influence

human decisions, using effectivepoliteness strategies?

Page 35: Takayama RSS2009 Workshop

HypothesesH1. People will change their decisions more often

when the robot disagrees with them than when italways agrees with them, even with identicalsubstantive content.

H2. People will feel more similar to (H2a) and morepositively toward (H2b) the agreeing robot than thedisagreeing one.

H3. A disagreeing voice coming from a separatecontrol box will be more acceptable than adisagreeing voice that came from the robotic body.

Page 36: Takayama RSS2009 Workshop

Study Design (N=40)

20 men and 20 women, balanced across conditions

Disagree 60%

Disagree 0%

Voice location:in box

Voice location:on robot

Between-participants

Page 37: Takayama RSS2009 Workshop

Procedure

• Write down decisions about desert survival items• For each item

– Tell robot which item to retrieve– Robot responds with survival item information

and judgment about decision– Tell robot which item to retrieve

• Write down final decisions about survival items– These final ratings will be “evaluated”

• Fill out paper questionnaire

Page 38: Takayama RSS2009 Workshop

Desert survival task

You are one of the members of a geology clubthat is on a field trip to study unusual formationsin the New Mexico desert. It is the last week inJuly. You have been driving over old trails, farfrom any road, in order to see out-of-the-wayformations. At about 10:30 A.M. the speciallyequipped minibus in which your club is ridingoverturns, rolls into a 20-foot ravine, and burns.The driver and professional advisor to the clubare killed. Both of you are relatively uninjured…

Page 39: Takayama RSS2009 Workshop

Desert survival task

Rank the following items according to their importance to your survival, startingwith 1 for the most important one and proceeding to 12 for the least important one.

______ magnetic compass______ 20-ft by 20-ft piece of heavy-duty, light-blue canvas______ book, Plants in the Desert______ rearview mirror______ large knife______ flashlight (four-battery size)______ one jacket per person______ one transparent, plastic ground cloth (6-ft by 4-ft) per person______ .38-caliber loaded pistol______ one 2-quart plastic canteen per person, full of water______ accurate map of the area______ large box of kitchen matches

Page 40: Takayama RSS2009 Workshop

Experiment Set-up

Page 41: Takayama RSS2009 Workshop

Manipulation: Robotdisagreement

Statement Examples

1.

Description

of selected

item

The knife could be helpful in

cutting down stakes to build a

solar still or to build shelter. It

could also assist in cutting down

firewood for a fire.

2.

Judgment:

disagreeing

or agreeing

That is not as

good as…

That is a better

choice than…

3.

Description

of

alternative

item

The pistol, which could be good

for signaling for help. It could

provide an alternative noise

source if your voice is weak due

to dehydration.

4. Request

for final

selection

Which do you choose?

or

Page 42: Takayama RSS2009 Workshop

MeasuresBehavior• Number of decisions changed

Attitudes• Perceived agreeableness of robot

(2 items, Cronbach’s α=.69)

• Perceived similarity of robot to self(4 items, Cronbach’s α=.94)

• Liking of the robot(8 items, Cronbach’s α=.75)

Page 43: Takayama RSS2009 Workshop

Perceived robotagreeablenessagreeableness

Page 44: Takayama RSS2009 Workshop

Perceived similarity to robot

Page 45: Takayama RSS2009 Workshop

People changed their minds

Page 46: Takayama RSS2009 Workshop

People like disagreement tocome from elsewhere

Page 47: Takayama RSS2009 Workshop

Checking against hypothesesH1. People will change their decisions more often

when the robot disagrees with them than when italways agrees with them, even with identicalsubstantive content.

H2. People will feel more similar to (H2a) and morepositively toward (H2b) the agreeing robot than thedisagreeing one.

H3. A disagreeing voice coming from a separatecontrol box will be more acceptable than adisagreeing voice that came from the robotic body.

Page 48: Takayama RSS2009 Workshop

Theory-orientedInterpretations

•Politeness: distancing•Disembodiment•Perceived source

– Two separate agents: Thinker and doer– Single distributed agent

Page 49: Takayama RSS2009 Workshop

Design-oriented Implications

• Voices can be more evocative than robot bodies• Agents can be sources of judgment and opinions• People are sensitive to disagreements• Disagreement undermines feelings of similarity• When agreeing 100% of the time, put the voice on the

robot body• When disagreeing (sometimes), put the robot voice

elsewhere

Page 50: Takayama RSS2009 Workshop

Validity

• face validity: how reasonable a measureseems to be for its concept

• content validity: how thoroughly a measureaddresses the breadth of a concept

• construct validity: how much a measurecausally relates to other variables withinone’s theory

• external validity: how generalizable theresults will be to other systems and contexts

Page 51: Takayama RSS2009 Workshop

Experiment Designs

•Keep it simple•Ceteris paribus•Random assignment to conditions•Balancing•Standardized tasks and measures•Behavioral and attitudinal measures•Sample representativeness

Page 52: Takayama RSS2009 Workshop

Stats

•Use with caution!•Especially with

statistical modeling

Page 53: Takayama RSS2009 Workshop

Reporting Studies

• Research questions andhypotheses

• Statistical significance• Reproducible methods• Discuss limitations• Thoroughly review

related work• Reduce bias in language• Clear labeling• Define terms

Page 54: Takayama RSS2009 Workshop

Experiment Work Practices

•Pilot stimuli, measures, procedures withmultiple types of pilot participants

•Identifying the important variables andtheir relationships (grounded theory)

Page 55: Takayama RSS2009 Workshop

Sharing artifacts and code

www.willowgarage.com

Page 56: Takayama RSS2009 Workshop

Thanks!

Victoria Groom, Clifford Nass

Claudia Jimenez, Alison King, Morgan Ames, CourtneySchultz, Paloma Ochi, Jessica Yuan

Contact: Leila [email protected]

Page 57: Takayama RSS2009 Workshop

Data Frame Model