Adapting Nielsen's Usability Heuristics to the Context of Mobile Augmented Reality
AAdapting Usability Heuristics to the Context of MobileAugmented Reality
Audrey Labrie
Polytechnique MontréalMontréal, Québec, [email protected]
Jinghui Cheng
Polytechnique MontréalMontréal, Québec, [email protected]
ABSTRACT
Augmented reality (AR) is an emerging technology in mobileapp design during recent years. However, usability challengesin these apps are prominent. There are currently no establishedguidelines for designing and evaluating interactions in AR asthere are in traditional user interfaces. In this work, we aimedto examine the usability of current mobile AR applicationsand interpreting classic usability heuristics in the context ofmobile AR. Particularly, we focused on AR home design appsbecause of their popularity and ability to incorporate importantmobile AR interaction schemas. Our findings indicated thatit is important for the designers to consider the unfamiliarityof AR technology to the vast users and to take technologicallimitations into consideration when designing mobile AR apps.Our work serves as a first step for establishing more generalheuristics and guidelines for mobile AR.
INTRODUCTION
Mobile augmented reality (AR) is gaining an increasing at-tention and started to break into the mass market due to theavailability of the technology on major mobile platforms (e.g.iOS and Android). Although an increasing number of usersare discovering mobile AR, there are no commonly acceptedguidelines or heuristics to support the designers of this type ofapplications to create an optimized user experience [1]. Weaim to address this challenge by identifying these guidelinesand heuristics from the evaluation of the existing commercial-off-the-shelf applications. As a first step towards this goal, weexamine how Jakob Nielsen’s Usability Heuristics [2], a setof widely-known heuristics for general applications, can beapplied to the context of mobile AR design.Particularly, we conducted a case study on AR home designapps. We chose this application domain because, from ourpreliminary investigation of mobile AR applications, homedesign is a domain where a lot of commercial-off-the-shelfapps are available and are created by professional companies.Additionally, this type of applications has common patterns ofinteractions that are important in most mobile AR applications(i.e. placing and manipulating virtual objects in a scene) [5].
This is the pre-print for a poster paper accepted to UIST ’20
Finally, the design and evaluation of AR home design apps isan active research direction. For example, Park et al. [4, 3]developed a questionnaire, based on usability analysis guide-lines [7] and the usability principles [1], for evaluating an ARhome design application. Viyanon et al. [6] also created aquestionnaire based on Nielsen’s definition of usability.This paper situates in these related studies and focuses oninterpreting and adapting Nielsen’s heuristics in the AR homedesign context. Through this effort, we provide a first steptowards the interpretation of Nielsen’s usability heuristics formobile AR applications, as well as the creation of specializedguidelines and heuristics for mobile AR application design.
METHODS
To select a set of apps for analysis, we first used the searchstring "augmented reality home design" in the Apple AppStore and found 14 apps. To narrow down our selection, weexcluded those that did not update within a month to focus onthe active projects. Additionally, we added one other app (i.e.Sayduck) because it had unique features. The apps selectedthis way were: IKEA Place, Houzz Home Design Renovation,Interior Define AR, Stresslesshome, Graham Brown DesignRenovation, and Sayduck. The common goal of the AR homedesign apps is to enable the user to view virtual objects (e.g.furniture, wallpapers, and tiles) in a physical environment toget a sense of how it would look in real life.After selection of the mobile applications, we analyzed eachapps by answering thoroughly the following questions: (Q1)How can users interact with the AR objects? (Q2) What kindof information is presented on the mobile screen and how?(Q3) What aspects of the app are well designed? (Q4) Whatare the prominent usability issues? The two authors first inde-pendently reviewed the two most popular apps (IKEA Placeand Houzz Home Design) and discussed their findings. Then,we collaboratively mapped each feature and issue identified inthose questions to the Nielsen’s 10 Usability Heuristics. Thisserved as a groundwork to the first draft of interpretation ofNielsen’s heuristics in AR home design apps. The other appswere reviewed using the same process and the interpretationof the heuristics were modified iteratively.
RESULTS
The following interpretation of Nielsen’s heuristics is basedon both the usability issues and the well-designed featuresidentified in the analyzed AR home design apps.
Visibility of system status.
The system should keep the userinformed during the interaction. When scanning surfaces, a r X i v : . [ c s . H C ] A ug or example, the user should be informed when the physicalsurfaces are difficult to detect (e.g., too reflective or too dark).After successful surface detection, the system should notifythe user that the object can be placed. If it takes time to loadan object, the system should inform the user. Match between system and the real world.
Virtual objectsshould appear and behave as realistically as possible. Theirsizes should be proportional to the physical environment andthey should be fixed to surfaces, as it is in the real world.Objects should not overlap with each other, i.e., collisionsbetween objects should be detected by the system.
User control and freedom.
It is possible that the user placesa virtual object in an unwanted location, or selects, deletes orrotates the object by mistake. Therefore, the system shouldsupport undo and redo and it should confirm with the userwhen deletion is selected.
Consistency and standards.
Gestures used to translate androtate an object should be intuitive. It is particularly importantto indicate to the user how the object can be manipulated. Ifadjusting the height is required for an object, it should bedone by an extra UI component (e.g. a vertical slider) becausedirectly moving the object on the vertical dimension can beeasily confused with object translation.
Error prevention.
Possible errors include placing or deletionof an object by mistake. For these error-prone actions that canlead to consequences and inconveniences, the system shouldask for confirmation. If the user does not confirm, it wouldreturn the object to its original location.
Recognition rather than recall.
Possible actions regardinginteraction with virtual objects should be always visible. Whenadding an object to the scene, to avoid recall (i.e. rememberingwhat object is to be added), the system should display an imageof the selected object. When an object is selected, all possibleactions should be shown on the screen (e.g. undo/redo, delete,change color); useful gestures for translating and rotating theobject should also be indicated. When no object is selected,the system should show the possible actions related to theentire scene (e.g. take a photo, favorites, reset scene).
Flexibility and efficiency of use.
Regarding frequent actions,the system should allow the user to place several objects onthe same scene at the same time. Regarding accelerator, thepresence of additional information about objects such as mea-surements, physical front, material, and price would be helpful.If it is technologically possible, objects should be automati-cally snapped to a physical corner, which would reduce thenumber of actions needed by the user.
Aesthetic and minimalist design.
The system should not showirrelevant or rarely used information to the user. Extra informa-tion could be shown when the user makes the correspondingselection. Examples of the common relevant information dur-ing the different stages of interaction are: (1) when placingan object, there should be a placement shape indicating theposition and the size of the object; and (2) when selectingan object, an indicator of selection and options for objectoperations should be displayed.
Help users recognize, diagnose, and recover from errors.
Helpfor error recovery is important. For example, the momentthe system has trouble detecting surfaces, the user should beinformed of the reason and help the user correct the error. Ifthe user is moving a virtual object on an undetected surface,the user should have the option to re-activate surface detection.
Help and documentation.
Such a system usually should nothave explicit documentation because it should be intuitive touse. However, they should have a brief tutorial for first timeusers. Particularly, surface detection is an action that mostusers are not familiar with. So there should be a brief tutorialand explicit indication telling the user how to detect surfaces.The possibility of translating and rotating the objects shouldalso be clearly communicated to the user.
DISCUSSION
One of our prominent findings is that many usability issuesin AR home design apps are associated with the current lim-itations of AR technology. For example, one of the mostproblematic aspects of these apps is the incorrectness or in-ability to detect surfaces. These technological limitations canaffect usability. For example, if a surface is too dark to bedetected and the system does not tell the user, the user willonly be left wondering why surface detection is taking a longtime and not working. Therefore, when developing an ARmobile app, identifying technological limitations and takingthem into consideration during the design process is critical.Further, because mobile AR is still an emerging technology,most AR app users should be considered as novices. Forthese users, several actions such as surface detection or objectmanipulation (i.e. interactions with virtual 3D objects on a2D space) can be confusing. Therefore, it is important toshow the users how to use the technology and how to interactwith the objects with gestures. One solution could be to showanimation gestures during the first time of use. Also, confusinginteractions need to be avoided. For example, there should betwo different types of interactions for adjusting the height ofan object and vertical translation (moves the object closer orfurther from the user) because intuitively, the gesture for bothactions would be swiping up and down.As previously mentioned, the mapping to Nielsen’s heuristicsis specific to AR mobile home design apps and they could beused to establish more general heuristics and guidelines aboutmobile AR design. To evaluate their validity, more analysison other applications is needed and a user study where UXdesigners use the heuristics to design or evaluate AR homedesign interfaces could be done in future work.
CONCLUSION
In this study, we focused on interpreting Nielsen’s 10 Us-ability Heuristics in the context of AR home design mobileapplications. To do this, we evaluated six AR home designapps available on the Apple App Store to identify their well-designed features and usability issues to map them to Nielsen’sheuristics. This effort serves as a first step in establishing morefull-fledged heuristics and guidelines for the design of mobileAR apps.
EFERENCES [1] Andreas Dünser, Raphaël Grasset, Hartmut Seichter,and Mark Billinghurst. 2007. Applying HCI principlesto AR systems design. (2007).[2] Jakob Nielsen and Rolf Molich. 1990. HeuristicEvaluation of User Interfaces. In
Proceedings of theSIGCHI Conference on Human Factors in ComputingSystems (CHI â ˘A ´Z90) . Association for ComputingMachinery, New York, NY, USA, 249â ˘A ¸S256.
DOI: http://dx.doi.org/10.1145/97243.97281 [3] Jinwoo Park, Sung Sil Kim, Hyerim Park, and WoontackWoo. 2016. â ˘AIJDreamHouseâ ˘A˙I: NUI-BasedPhoto-Realistic AR Authoring System for InteriorDesign. In
Proceedings of the 7th Augmented HumanInternational Conference 2016 (AH â ˘A ´Z16) . Associationfor Computing Machinery, New York, NY, USA, ArticleArticle 14, 7 pages.
DOI: http://dx.doi.org/10.1145/2875194.2875221 [4] Y. J. Park, Y. Yang, S. Chae, I. Kim, and T. Han. 2017.DesignAR: Portable projection-based AR systemspecialized in interior design. In . 2879–2884.
DOI: http://dx.doi.org/10.1109/SMC.2017.8123064 [5] S. Sharma, Y. Kaikini, P. Bhodia, and S. Vaidya. 2018.Markerless Augmented Reality based Interior DesigningSystem. In . 1–5.
DOI: http://dx.doi.org/10.1109/ICSCET.2018.8537349 [6] Waraporn Viyanon, Thanadon Songsuittipong,Phattarika Piyapaisarn, and Suwanun Sudchid. 2017.AR Furniture: Integrating Augmented RealityTechnology to Enhance Interior Design Using Markerand Markerless Tracking. In
Proceedings of the 2ndInternational Conference on Intelligent InformationProcessing (IIPâ ˘A ´Z17) . Association for ComputingMachinery, New York, NY, USA, Article Article 32, 7pages.
DOI: http://dx.doi.org/10.1145/3144789.3144825 [7] Beth F. Wheeler Atkinson, Troy O. Bennett, G. SusanneBahr, and Melissa M. Walwanis Nelson. 2007.Development of a Multiple Heuristics Evaluation Table(MHET) to Support Software Development andUsability Analysis. In