CHI 2008 – Florence, Italy E E valuating Motion Constr valuating Motion Constr aints aints for 3D Wayfinding in Imme r sive and De sktop Vir tual E E nvir nvir onme nts onme nts Niklas Elm qvist M. Eduard Tudoreanu Philippas Tsigas elm@lri.fr metudoreanu@ualr.edu tsigas@chalmers.se g
2 he Challe nge he Challe nge • 3D motion constraints! vs T T
Outline Outline • Problem • Problem • Design space: Wayfinding in 3D • Solution: Motion constraints S l ti M ti t i t • User study • Results and discussion • Conclusions and future work • Conclusions and future work 3
Pr Pr oble m oble m • W ayfinding : navigation to solve • W ayfinding : navigation to solve specific task – Performed on cognitive map – Performed on cognitive map – Poor map leads to poor performance • Objective : support wayfinding by Obje ti e t fi di b aiding cognitive map building – Motion constraints and guides d d – Exam ple : sightseeing tour of new city 4
Vir Vir tual vs Physic al Wor tual vs. Physic al Wor lds lds • Why is wayfinding more difficult in • Why is wayfinding more difficult in virtual worlds? – Low visual fidelity – Low visual fidelity – Mouse and keyboard poorly mapped to 3D navigation 3D navigation – Lack of sensorial cues • High cognitive load on users • High cognitive load on users 5
R R e duc ing Cognitive L e duc ing Cognitive L oad oad • Method : Immersive et od e s e Virtual Reality – Full 3D input – Full 3D output F ll 3D • But : No widespread use expensive (?) use, expensive (?) • Mouse and keyboard are standard – Even for 3D games! 6
Cognitive Maps Cognitive Maps Montmar Montmar tr tr e e Air por t (CDG) ? L ouvr e Notr e Dame T our E iffe l Hote l Hote l 7
Suppor Suppor ting Cognitive Maps ting Cognitive Maps • Global coverage • Global coverage – Expose viewer to whole environment • Continuous motion Continuous motion – Support spatial relations • Local control Montmar tr e Air por t (CDG) – Learning by doing ? L ouvr e Notr e Dame T our E iffe l Hote l 8
3D Motion Constr 3D Motion Constr aints aints • Tour-based motion constraints • Tour based motion constraints • Spring-based control • Smooth animation S th i ti 9
Use r Use r Study Study • Predictions Predictions – P1 : Guiding navigation helps wayfinding – P2 : User control will improve familiarization p – P3 : More improvement for desktop • Controlled experiment • Two experiment sites • 35 participants p p – 16 (4 female) on desktop computer – 19 (2 female) on CAVE system 10
E E xpe r xpe r ime ntal Conditions ime ntal Conditions • Platform ( BS) : desktop or CAVE • Platform ( BS) : desktop or CAVE • Navigation ( BS/ W S) : free, follow, spring spring • Scenario ( W S) : outdoor, indoor, i f infoscape, conetree t • Collect distance, error, and time 11
Pr Pr oc e dur oc e dur e e • Phase I : Familiarization Phase I : Familiarization – Create cognitive map (5 minutes) – Supported by guidance technique Supported by guidance technique – Three target object types • Phase I I : Recall Phase I I : Recall – Locate two targets on overhead map • Phase I I I : Evaluation Phase I I I : Evaluation – Collect target in world – No navigation guidance No navigation guidance 12
R R e sults e sults • Navigation m ethod : • Navigation m ethod : – Free navigation: CAVE better – Motion constraints: desktop significantly Motion constraints: desktop significantly better ( p < 0.05) recall distance evaluation error 0,3 0,3 0,25 0,25 or or 0,2 0 2 0 2 0,2 normalized erro normalized erro CAVE CAVE 0,15 0,15 Desktop Desktop 0,1 0,1 0,05 0,05 0 0 free follow spring free follow spring 13
R R e sults (c ont’d) e sults (c ont d) • Desktop platform : • Desktop platform : – Spring-based guidance gave better accuracy than other methods accuracy than other methods – Navigation guidance more efficient than none none average time per target 50 45 40 econds) 35 time per target (se 30 CAVE 25 Desktop 20 15 10 5 0 free follow spring 14
Disc ussion Disc ussion • Unaided navigation easier in CAVE • Unaided navigation easier in CAVE • Guidance improved performance (P1) – Guidance reduces cognitive load G id d iti l d • Local control improved accuracy (P2) – Learning by doing works for desktops • CAVE performed worse with guidance p g – Motion constraints work against – Partial confirmation of P3 15
Conc lusions and F Conc lusions and F utur utur e Wor e Wor k k • Navigation guidance based on tours • Navigation guidance based on tours – Improve cognitive map building – Improve visual search Improve visual search • Evaluation on desktop and CAVE – Navigation guidance on desktop d d k outperforms CAVE – Less focus on interaction mechanics L f i t ti h i 16
Que stions? Que stions? • Contact inform ation • Main findings: Main findings: – Niklas Elmqvist – Free-flight best on (elm@lri.fr) immersive platforms – Motion guidance – Edi Tudoreanu (metudoreanu@ualr.fr) helped desktop users outperform CAVE outperform CAVE users – Philippas Tsigas (tsigas@chalmers.se) – Allowing local deviations improved correctness for desktop desktop 17
Recommend
More recommend
