l arge d isplays i mprove 2 d and 3 d n avigation by john bell and tom peterka university of...

30
LARGE DISPLAYS IMPROVE 2D AND 3D NAVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor Moher, Spring 2005 10 February 2005

Upload: denis-ferguson

Post on 15-Jan-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

LARGE DISPLAYS IMPROVE

2D AND 3D NAVIGATION

ByJOHN BELL and TOM PETERKA

University of Illinois at Chicago

CS 522Human Computer InteractionProfessor Moher, Spring 2005

10 February 2005

Page 2: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

This presentation is based on two major papers:

Desney Tan

School of CS, Carnegie Mellon

Peter ScupelliDarren Gergle Randy Pausch

“With Similar Visual Angles, Larger Displays Improve Spatial Performance”, CHI 03

“Physically Large Displays Improve Path Integration in 3D Virtual Navigation Tasks”, CHI 04

Page 3: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The authors make 3 key points:

Large displays work better for 2D tasks.

Large displays work better for 3D tasks.

The observed effects are cognitive, not locomotive.

But first, some background on large displays . . .

Page 4: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Tan et al. provide location and place to aid human memory in the Infocockpit.

Page 5: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

MacIntyre et al. support multitasking and background awareness using interactive

peripheral displays.

Page 6: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Baudisch et al. compare focus + context vs. overviews vs. zooming and panning.

Page 7: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Robertson et al. manage windows in 3D space in The Task Gallery.

Page 8: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Key Concepts and Terminology

• Wide field of view ( FOV ) increases presence.Do large displays provide similar benefits?

• Egocentric perspective perceives the user to be moving in a stationary universe.

• Exocentric perspective perceives the universe to be moving around a stationary user.

• Path integration is a wayfinding method that requires the user to “integrate” their motion,as opposed to piloting by navigational landmarks.

• Active navigation has the user controlling the joystick; Passive allows him / her to ride along.

Page 9: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Large displays are better in 2D.

Tan et al., “With similar visual angles, larger displays improve spatial performance,” CHI 03.

Fig. 1

Page 10: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The experiment hardware consisted oflarge and small displays at same FOV;

the input device was a USB keypad.

24 college students participated, 12 male.A primary study and follow up study were conducted.

Fig. 2

Page 11: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The primary study consisted of 2 tasks: a reading comprehension test and the Guilford-Zimmerman

(G-Z) spatial orientation test.

Imagine a camera fixed to the bow of the boat.Select the proper change in orientation symbols.

Use exocentric or egocentric point of view.

Fig. 3

Page 12: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The large display produced better testscores on the spatial test and no

difference on the reading test.

2(size) x 2(position) x 2(gender) RM-ANOVA.F(1,20) = 9.470, p = .006

Fig. 4

55.4%

43.8%

Page 13: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The follow-up study was a modified shape test.

24 college studentsmodified shape test x G-Z test x display size

Exocentric view for shape testEgocentric view for G-Z test

Fig. 5

Page 14: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The large display produced better testscores on the G-Z test and little

difference on the shape test.

2(size) x 2(position) x 2(task type) RM-ANOVAF(1,21) = 5.512, p = .0288 large display improves

the G-Z test more than the shape test.

Fig. 6

53.2%42.9%

40.8%39.5%

Page 15: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Three conclusions can be drawnfrom the 2D tests.

1. No benefit is gained from a larger display for reading comprehension.

2. A larger display is better for 2D orientation tasks such as the G-Z test.

3. A larger display benefits egocentric tasks such as the G-Z test more than it does exocentric tasks such as the shape test.

Page 16: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Large displays are better in 3D.

Tan et al., “Physically Large Displays Improve Path Integration in 3D Virtual Navigation Tasks,” CHI 04.

Equipment is the same as the 03 paper.

Page 17: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The new task involves navigation in an immersive 3D ( VR ) environment, with

motion cues but no navigational landmarks.

Page 18: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Users must path-integrate two sides,and then wayfind back to the origin

without the posts displayed.

Page 19: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Other Experimental Notes

• Users could only turn at the vertices, BUTthey could back up and change their angle if they did not like the results.

• Users received practice trials with unique triangles and an overhead map, with the warning that the map would not be available during the actual tests.

Page 20: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The authors postulate 3 hypotheses:

1. Users will perform better in path integration using large displays, due to increased likelihood of egocentric strategies.

2. Users will perform better in path integration when using active navigation.

3. Display size and interactivity mode are independent effects.

Page 21: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Experimental design was 2 x 2 x 6 trials.

• 3 dependent variables• Two display sizes – large & small• Two navigation modes for the first 2 legs – active & passive• Six triangles – 60°, 90°, 120°,

and 5 or 3m second leg• 24 trials / user; 384 total trials.• Participants were 16 college students ( 8 male ), 19 to 29,

with normal eyesight, average to experienced computer users, who played less than 1 hr / week 3D games.

• Display size / interactivity mode balanced for confounding effects. Triangles were fully randomized.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3

Page 22: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Larger displays weresignificantly better.0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3

display size:F(1,339)=11.24p<.0012.88m vs. 3.48m

interactivity:F(1,339)=12.38p<.0012.87m vs. 3.49m

Fig. 6

Mixed model ANOVA (display size x interactivity x gender)

Page 23: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Unpacking component errors helpsclarify significant and

insignificant correlations.0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3

displaysize interactivity gender

distanceerror

angleerror

trianglesize

triangleshape

displaysize 0 0 - 0 0 0

interactivity 0 0 0 0 0

gender 0 0 0 0

distanceerror 0 + 0

angleerror +

trianglesize

triangleshape

+ positive correlation (significant)- negative correlation (significant)0 no correlation (not significant)

Page 24: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

A follow-up study explored whethererrors were cognitive wayfinding

or physical locomotion errors.

• Users were guided a specified distance, and told the specific angle to turn. ( 3 to 8 m. and 60, 90, 120, or 150 degrees. )

• Only the small display and active navigation were investigated. ( The combination with the largest errors in the original experiments. )

• Eight ( 4 male ) students participated, who had not participated in the first study, but with similar demographic characteristics.

Page 25: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

Results indicate that earlier errorswere cognitive, not locomotive.

• With 95% confidence, distance to origin error was 0.31 to 0.39 m., distance moved error was 0.18 to 0.22 m., and angle turned error was 2.31 to 2.75 degrees.

• Mean errors in original experiment were 3.78 m., 1.71 m., and 31.52 degrees respectively.

• Conclusion: Locomotive errors were a small portion of the overall error observed.

Page 26: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

The authors arrived at 3 conclusions.

Bigger is better: given similar FOV, larger screen size promotes an egocentric point of view resulting in improved wayfinding.

Interactivity does not necessarily improve wayfinding; in fact it may prove to be a distraction to wayfinding.

Wayfinding errors are cognitive rather than mechanical.

Page 27: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

We have a few unanswered questions.

If interactivity was not statistically correlated to the component errors, should it be statistically correlated to the resultant total error?

Head mounted displays are the other extreme of display size with similar visual angle. Could or should this display medium also be included in the experiment?

What effect do the borders around the LCD monitor have on ego/exocentric viewpoint? ( Assuming projection screen is borderless. )

Page 28: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

We have more unanswered questions. ( continued )

The LCD is in the user’s foreground ( ie., there is space behind it ) while the projection wall is in the user’s background ( ie., at the horizon ). What effect does this have on viewpoint?

If the distance traveled was consistently short, is there some underlying phenomenon that is causing the underestimation of distance? E.g., Is the undershoot in distance actually perceptually constant, but scaled longer in the small display because the same perceptual error represents a longer distance in world space?

Page 29: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

John Bell and Tom Peterka 10 Feb 2005 CS 522

References

• Desney S. Tan, Darren Gergle, Peter G. Scupelli, Randy Pausch, “Physically Large Displays Improve Path Integration in 3D Virtual Navigation Tasks”, CHI 2004.

• Desney S. Tan, Darren Gergle, Peter G. Scupelli, Randy Pausch, “With Similar Visual Angles, Larger Displays Improve Spatial Performance”, CHI 2003.

• Desney S. Tan, Jeanine K. Stefanucci, Dennis R. Proffitt, Randy Pausch, “The Infocockpit: Providing Location and Place to Aid Human Memory”, PUI 2001.

• Blair MacIntyre, Elizabeth D. Mynatt, Stephen Voida, Klaus M. Hansen, Jope Tullio, Gregory M. Corso, “Support for Multitasking and Background Awareness Using Interactive Peripheral Displays”, UIST 2001.

• Patrick Baudisch, Nathanial Good, Victoria Bellotti, Pamela Scraedley, “Keeping Things in Context: A comparative Evaluation of Focus Plus Context Screens, Overviews, and Zooming”, CHI 2002.

• George Robertson, Maarten van Dantzich, Daniel Robbins, Mary Czerwinski, Ken Hinckley, Kirsten Risden, David Thiel, Vadim Gorokhovsky, “The Task Gallery: A 3D Window Manager”, CHI 2000.

Page 30: L ARGE D ISPLAYS I MPROVE 2 D AND 3 D N AVIGATION By JOHN BELL and TOM PETERKA University of Illinois at Chicago CS 522 Human Computer Interaction Professor

LARGE DISPLAYS IMPROVE

2D AND 3D NAVIGATION

ByJOHN BELL and TOM PETERKA

University of Illinois at Chicago

CS 522Human Computer InteractionProfessor Moher, Spring 2005

10 February 2005