evaluating motion constraints for 3d wayfinding in immersive and desktop virtual environments

Evaluating Motion Constraints for 3D

Wayfinding in Immersive and Desktop Virtual

Environments

Niklas Elmqvistelm@lri.fr

M. Eduard Tudoreanumetudoreanu@ualr.edu

Philippas Tsigastsigas@chalmers.se

CHI 2008 – Florence, Italy

2

Where am I? Where am I going?

3

How to Aid Wayfinding?

• Wayfinding: navigation to solve specific task– Performed on cognitive map– Poor map leads to poor performance

• Problem: aiding wayfinding by supporting cognitive map building– Motion constraints and guides– Example: sightseeing tour of new city

4

Cognitive Maps

Hotel

Tour Eiffel

Montmartre

Notre DameLouvre

Airport (CDG)

?

5

Supporting Cognitive Maps

• GC: Global coverage– Expose viewer to whole environment

• CM: Continuous motion– Support spatial relations

• LC: Local control– Learning by doing

Hotel

Tour Eiffel

Montmartre

Notre DameLouvre

Airport (CDG)

?

6

Virtual vs. Physical Worlds

• Why is wayfinding more difficult in virtual worlds?– Low visual fidelity– Mouse and keyboard poorly mapped to

3D navigation– Lack of sensorial cues

• High cognitive load on users

7

Reducing Cognitive Load

• Method: Immersive Virtual Reality– Full 3D input– Full 3D output

• But: No widespread use, expensive (?)

• Mouse and keyboard are standard

8

Guiding using 3D Motion Constraints

• Tour-based motion constraints (GC)• Smooth animation (CM)• Spring-based control (LC)

9

User Study

• Hypotheses– H1: Guiding navigation helps wayfinding– H2: Local control improves familiarization– H3: Navigation guidance has higher impact for

desktop than for CAVE

• Controlled experiment (mixed design)• Two experiment sites• 35 participants

– 16 (4 female) on desktop computer– 19 (2 female) on CAVE system

10

Factors

• Platform (BS): desktop or CAVE• Navigation (BS/WS): free, follow,

spring• Scenario (WS): outdoor, indoor,

infoscape, conetree• Collect distance, error, and time

11

Procedure

• Phase I: Familiarization– Create cognitive map (5 minutes)– Supported by guidance technique– Three target object types

• Phase II: Recall– Locate two targets on overhead map

• Phase III: Evaluation– Collect third target in world – No navigation guidance

12

Results – Summary

• Navigation method:– Free navigation: CAVE better– Motion constraints: desktop significantly

better (p < 0.05) [H1]

• Desktop platform:– Spring-based guidance gave better

accuracy than other methods [H2]– Navigation guidance more efficient than

free flight [H3]

13

evaluation error

0

0,05

0,1

0,15

0,2

0,25

0,3

follow free spring

no

rma

lize

d e

rro

r

CAVE

Desktop

average time per target

0

5

10

15

20

25

30

35

40

45

50

follow free spring

tim

e p

er

targ

et

(se

co

nd

s)

CAVE

Desktop

recall distance

0

0,05

0,1

0,15

0,2

0,25

0,3

follow free spring

no

rma

lize

d e

rro

r

CAVE

Desktop

Results – Details

14

Conclusions

• Navigation guidance based on tours– Improve cognitive map building

• Evaluation on desktop and CAVE– Navigation guidance on desktop

outperforms CAVE– Less focus on interaction mechanics

• Take-away message: removing (some) freedom in 3D navigation may actually help wayfinding!

15

Questions?

• Main findings:– Free-flight best on

immersive platforms– 3D guidance: desktop

outperform CAVE– Allowing local control

helped desktop

– Removing freedom improves performance

• Contact information– Niklas Elmqvist

(elm@lri.fr)

– Edi Tudoreanu (metudoreanu@ualr.edu)

– Philippas Tsigas (tsigas@chalmers.se)