基於使用者行為之擴增實境動態調整顯示技術

A User Behavior Driven Display Technique for Augmented Reality System

高嘉駿(Jia-Chung Kao)

指導教授 ： 黃俊堯

國立臺北大學/電機資訊學院/資訊工程學系/碩士(2014年)

摘要

隨著近幾年擴增實境技術的迅速發展，其載具不僅是手持式裝置，也包括笨重的頭戴式裝置，但隨著科技進步，這些裝置正朝向輕薄化發展，穿戴式擴增實境系統也如雨後春筍般被提出來，例如Google Glass，然而穿戴式裝置的應用面臨一些問題與挑戰，包含在有限大小的螢幕顯示擴增資訊所造成的使用者感知負擔與不方便的人機介面造成使用者不易與擴增資訊互動。
本論文提出透過使用者當前的行為狀態來動態調整擴增實境的顯示模式，以解決上述感知負擔與人機互動問題。本研究所提出之方法，首先探討使用者在使用擴增實境系統時的行為，以建立使用者的行為模型，接著由擴增實境載具的感測器資料萃取出能夠判別使用者行為的特徵資料，運用此特徵資料歸類使用者行為，但是純粹依靠歸類特徵資料並不能準確判斷使用者行為，因為感測器常會因不可預期的環境因素影響而產生雜訊(Noise)，進而影響到歸類的準確度，因此本研究提出以使用者過去使用擴增實境系統的行為習性，來推測使用者當前的行為狀態，最後將此兩種不同來源的狀態資料進行交叉檢測，以推論出最符合當下使用者情境之使用者狀態。
為驗證本論文所提出的方法，本研究以現今最常見的擴增實境應用為例－擴增實境瀏覽器(AR Browser)，建構基於使用者行為來動態調整擴增資訊顯示模式的雛型系統，該系統將AR Browser使用者行為分為五種運動型態，包括：1.原地靜止；2.原地搜尋；3.轉彎移動；4.移動時觀看興趣點；5.移動搜尋。並且根據這些運動行為來切換不同顯示模式，提高使用者透過AR Browser系統搜尋周遭資訊的效率，讓整體擴增資訊瀏覽融入人類日常生活，未來更可能加入使用者的個人資訊延伸推論的結果，使得其顯示資訊能夠更加符合使用者的當下需求。

關鍵字

馬可夫模型 ； 行為感知 ； 人機介面 ； 擴增實境 ； 互補式濾波器

並列摘要

With the evolvement of mobile augmented reality(MAR) technology, its hardware platform range from handheld device to head mounted display in the past. Along with the advancement of embedded technology into lighter and thinner in the recent years, the applications of augmented reality on various wearable devices are massively proposed. The Google glass is one good example of such development. However, the wearable MAR applications always face the issues of limited screen display size that increase the cognitive load to the user and unhandy interface that obstruct the interaction with augmented information.
To solve above cognitive load and Human-Computer Interaction(HCI) problems, this research proposes an adaptive augmented information display method that are based on the user's behavior. The approach started from building the user behavior model when they are using the mobile augmented reality system. The behavior features that are accessable from wearable device are then identified. These behavior features become an important clue to identify the behavior states of the user. That is, the feature data are used to classify the possible current state of the user. However, due to the unpredictable environment interference to the sensors, the feature data along can’t accuratly determine the user’s behavior. Therefore, the study further applied the past behavior of the user, when he was using MAR stsem, to predict his current possible behavior state. The predicted state and clssified state are then cross-referenced to infer the most likely user’s motion state.
In order to verify the proposed method, the study takes the commonly available augmented reality application, Augmented Reality Browser, as an example to develop a prototype system. Based on user’s behavior when using a MAR system, the prototype system categorize user motion into five states: stationary state, stationary but look around state, turning while moving state, moving and staring state, and searching while moving state. The display pattern of each state is then designed. The system is then adaptive switch among these display patterns according to perceived five states. The research goal is to improve the efficiency of searching surrounding augmented information when the user is using AR Browser so that the application of MAR can seamlessly integrate into our daily life. In the future, the user's personal profile can be further integrated into perception process to provide more adaptive augmented service based upon user's current state.

並列關鍵字

Markov Model ； Context-aware ； User Interface ； Augmented Reailty ； Complementary Filtering

參考文獻

[1] Azuma, R.T. 1997. A survey of augmented reality. Presence. 6, 4 (1997), 355–385.

[3] Furmanski, C., Azuma, R. and Daily, M. 2002. Augmented-reality visualizations guided by cognition: perceptual heuristics for combining visible and obscured information. Proceedings. International Symposium on Mixed and Augmented Reality (2002), 215–320.

[6] Seungjun Kim, A.K.D. 2009. Simulated Augmented Reality Windshield Display as a Cognitive Mapping Aid for Elder Driver Navigation. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (2009), 133–142.

[7] Wickens, C.D. Engineering psychology and human performance (2nd ed.).

[8] Chandler, P. & Sweller, J. (1992). The split-attention effect as a factor in the design of instruction. British Journal of Educational Psychology. 62, 2 (Jun. 1992), 233–246.