What is the effect of interface complexity on risk perception tasks?

What is the effect of interface complexity on risk perception tasks?
of 2
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  What is the Effect of Interface Immersiveness on Risk Perception Tasks? Vitor Jorge ∗ Anderson Maciel † Luciana Nedel ‡ Jackson Oliveira § Frederico Faria  ¶ Universidade Federal do Rio Grande do Sul (UFRGS), AES Sul, Nexo Art A BSTRACT Most human computer interfaces impose a certain level of addi-tional cognitive load to the user task. In this project we study theuseofVRtoevaluatetheusercompetencyforsafebehaviorinwork environments. We propose an assessment of two interfaces withdifferent levels of cognitive load for locomotion in a risk percep-tion task. We describe user experiments with two similar groups of subjects, each using one of the interfaces. We identified importantopen questions on the design of this kind of interface. Index Terms:  H.1.2. [User/Machine Systems]: Human factors— 1 I NTRODUCTION Safety is a permanent concern in work environments. Besides hav-ing the skills for the job, workers have to acquire a number of com-petencies that allow them to behave safely. More than training, toacquire any competency depends on experience, human factors andenvironmental influence It is then very hard to measure the compe-tency matrix of a worker. Written exams, practice tests in simulatedphysical environments, etc. are often not effective. Accident rate,unfortunately, remains the most reliable assessment parameter.We believe that a VR simulator is an effective tool to measurethe competency matrix for safe behavior. It provides the ideal envi-ronment to expose people to danger with no chances to cause themany harm. However, while in a physical work environment workershave to perform tasks at the same time they follow a safe behavior,in a VR simulator they have the additional load of dealing with theuser interface. How the cognitive overload of the interface impactsthe risk perception? How this interface influences the behavior?How the immersion can help the interface to be transparent? Howcontrol overload (excessive number of DoFs) affects attention andleads to a less safe behavior?In this poster we address these questions in the context of a VRsimulator of an office environment. Our main contribution is in auser study to evaluate danger awareness in a virtual environment(VE) and the study of the impact of the user interface in this aware-ness. 2 R ELATED  W ORKS VR simulators researchers often claim that the cognitive overloadcaused by the interface increases the learning experience. For in-stance, Akioshi et al. [1] evaluated the learning experience in VRmaintenance system applied in construction. They conclude thatthe free exploration of the VE enhances the learning experience.However, in the literature of 3D interaction an increase in thenumber of DoFs is often associated with a decrease in user per-formance. Bowman et al. [3] recommend the use of constraints to ∗ e-mail: vamjorge@inf.ufrgs.br † e-mail: amaciel@inf.ufrgs.br ‡ e-mail: nedel@inf.ufrgs.br § e-mail: jackson.oliveira@aessul.com.br  ¶ e-mail: fred@nexo.art.br (a) (b) Figure 1: The user view of the VE and a selectable object (a) beforeand (b) after selection. Notice the highlight on the selected object. reduce the DoFs of the user interface. They argue that when weconstrain the VR locomotion to the ground plane, we are in fact re-ducing the problem from 3D to 2D. It is said that if the user has tothink about  how  to turn left or move forward, then they have beendistracted from the primary task. In a previous study [2], they per-form a comparison of traveling techniques using different numbersof DOFs. They could not show any difference in the task com-pletion times using a reduced DoF technique, but the authors stillpoint out that reducing the DoFs is important for many applicationswhere users do not need the extra DoFs. They recommend furtherstudy on the performance of constrained vs. unconstrained motionin a denser VE with the addition of distractor tasks. 3 E XPERIMENTS We designed an immersive VR simulator for the assessment or risk perception (see Figure 1). The user navigates through an officescenario, identifies risky situations, moves their head to place thecross-hair over the target and then selects the object using the gamepad button. In this context, we proposed an experiment to deter-mine how the number of DoFs in the locomotion strategy affectsthe user perception and performance.Experts in security inspection defined a set of 46 risks spread inthe VE. We designed a virtual path from which the user can see allrisks. In order to guide them, the path was marked with arrows onthe floor.We evaluated two interaction techniques. The first allows theuser to turn, strafing and walk in the virtual world using a joystick.The other technique permits the user to go only forward or back-ward on a constrained path using the same joystick. The secondtechnique is expected to reduce cognitive overload as it cuts downDoFs in locomotion interaction. In either cases the user can seethe VE in any direction through the HMD. Figure 2 shows the dif-ferences between the techniques. The users are asked to select theobjects that represent danger (potential risks) in the virtual world.In addition, all VE objects, with the exception of the walls, are se-lectable. Apparatus.  A zSight Sensics HMD provided immersive visu-alization. To improve the user immersion, presence and proprio-ception, a MS Kinect sensor was used to track the user movementsand generate a skeleton that allows them to see their own limbs.The control of locomotion and selection is made through a standardgamepad to prioritize accuracy. Population . The test subjects were divided in two groups: theConstrainedLocomotionGroup(CR),inwhichthenumberofDoFs  Figure 2: Illustration of the differences between two interaction ap-proaches we evaluated.  FR CR  M  s ( σ  s )  29.08 (10.56) 30.45 (11.02)  M  r  ( σ  r  )  20.23 (4.90) 22.18 (6.50)  M  t  ( σ  t  )  503.40s (132.19s) 527.39s (234.11s)  M  FP ( σ  FP )  25.7% (15.9%) 25.5% (12.2%) Table 1: Statistics collected is reduced to one, and the Free Locomotion Group (FR) in whichthe users could freely navigate. The CR and FR groups were com-posedinitiallyby11and13subjectsrespectively, mostlymale, with2 females per group, with a mean age of 26.7 years old. There isno significant difference between groups in gaming expertise andprevious training on safety at work. Procedure.  To prepare the test subjects, an instructor passedon written instructions on the basic definitions of risk and safetyin the work environment. The user was guided on how to use theVR equipment, and was invited to fill a subject characterizationform with basic identification and previous experiences in safetyperception, video games and VR devices. After that, the users wereplaced in a basic training VE where they practiced as much as theywanted. To avoid user fatigue or cybersickness from prolonged useof VR equipment, they were allowed to rest for 1 to 5 minutes.The user was placed in the experiment VE and asked to select themaximum number of potential risks they could identify, withoutany time constraint. After the end of the test, users filled anotherform related to the risks perceived, and interface usability. Collected data and metrics.  We computed for each group theaverage object selection count,  M  s , the average number of risks se-lected,  M  r  , and the average time to complete the task,  M  t  , alongwith the corresponding standard deviations σ  s , σ  r   and σ  t  . To assessaccuracy, i.e. a selected object is actually a risk, we computed thenumber of false positives,  FP = S  e / S  c  for each user, where  S  e  is thetotal number of risks missed by a user in the observed group and  S  c is the total number of selections. Then, we computed the averagefalse positives for each group,  M  FP . 4 T EST  R ESULTS Table 1 contains the statistics collected from the experiment. Re-garding the effects of DoF constraint on the number of correct se-lections, the mean results are slightly better using the constrainedlocomotion strategy.From the questionnaires, the CR group stated that the systemwas easy to use, while users of the FR group state that they coulddo the task but they think that they took too long. Even thoughmost users using CR felt they have a good or regular performanceusing the system, most of the users stated that the most difficultaction was not the interface but the selection of risks. The mostinterestingisthatbothgroupshaveasimilarpointofviewregardingthe difficulties of using the system. 5 O PEN  Q UESTIONS From our user studies we identified the following questions thatshould be addressed in the design of VR simulators for risk percep-tion analysis. Does limiting the number of DOFs help or harm the trans-parency of the interaction?  We could not detect any significantadvantage or disadvantage of using a constrained technique over afree locomotion strategy technique. A possible reason is that, in theVE, the HMD yaw DoF is somewhat overlapped with the need toturnrightorleft, minimizingthedemandforturningonlocomotion. Does the answer above apply for the risk perception task? There is a chance that the risk perception task is a special type of task. That is, risk perception might be something more instinctivethan other tasks not associated with individual sensation of safety.In this case, finding risks triggers a stronger commitment than othertasks. This would minimize the effects of changing the number of DoFs. Does previous training in safety and risk detection help inanyway?  Our test subjects sample was mainly composed of peoplewith no special training in risk perception. This might have resultedin a group searching for only those risks that were learned throughinstinct and personal experience and not training. We hypothesizethat a group with solid training in risk perception would see suchtask as something more cognitively complex, since not only ba-sic knowledge would be involved, but also all the cognitive logicassociated to the extensive training experience on risk perception.However, it is still an open question. Motion sickness: Immersion with HMD vs. immersion withCAVE.  Almost all subjects reported some level of motion sicknessafter the experiment. In an on-going work, we are deeply inves-tigating the reasons for that and how can we bypass this problem.An implementation of the simulator in a mini-cave setup is beingtested. How does the level of proprioception influence risk percep-tion?  Natural navigation in immersive VE is still an open problemin 3D interaction and, even if we have tested some known tech-niques, none was satisfactory for our purposes. We are still lookingfor a navigation technique that fits well with our simulator. 6 C ONCLUSIONS AND  F UTURE  W ORK In this paper we investigated the role of DoF constraints for a risk perception task. Through preliminar user experiments, we couldnot determine that a constraint on locomotion significantly affectsthe performance in this task, but we identified some still open ques-tions in the field. Continuing this work, we plan to perform similartests on users with solid training and experience in risk analysis.This would enable us to determine if the complexity seen by trainedpersonnel is different from that seen by untrained people. A CKNOWLEDGEMENTS This work was supported by the Brazilian Electricity RegulatoryAgency – ANEEL through subsides to AES Sul, and by CNPq-Brazil under the projects 483947/2010-5 and 302679/2009-0. R EFERENCES [1] M. Akiyoshi, S. Miwa, T. Ueda, and S. Nishida. A learning environ-ment for maintenance of power equipment using virtual reality. In  Im-age Processing and its Applications, 1995., Fifth International Confer-ence on , pages 331 –335, July 1995.[2] D.A.Bowman, D.Koller, andL.F.Hodges. Travelinimmersivevirtualenvironments: An evaluation of viewpoint motion control techniques.In  Proceedings of the Virtual Reality Annual International Symposium ,pages 45–52, 1997.[3] D. A. Bowman, E. Kruijff, J. J. LaViola, and I. Poupyrev.  3D User  Interfaces: Theory and Practice . Addison Wesley Longman PublishingCo., Inc., Redwood City, CA, USA, 2004.
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!