SelfIE: Self-Initiated Explorable Instructions Towards Enhanced User Experience

TL;DR

SelfIE introduces a design paradigm for procedural instructions that blends linear and non-linear access to information during tasks. Through Wizard-of-Oz studies in toy-block assembly, the authors show a 71% increase in user preference for SelfIE over linear instructions and identify three flexible-access strategies, plus the potential of wearables. They then realize working prototypes on tablet and optical head-mounted displays to assess usability and trade-offs, highlighting the benefits and challenges of each configuration. The work demonstrates that allowing situational information retrieval can enhance user experience and motivates future research on cognitive-process support and hybrid wearable-tabletop implementations.

Abstract

Given the widespread use of procedural instructions with non-linear access (situational information retrieval), there has been a proposal to accommodate both linear and non-linear usage in instructional design. However, it has received inadequate scholarly attention, leading to limited exploration. This paper introduces Self-Initiated Explorable (SelfIE) instructions, a new design concept aiming at enabling users to navigate instructions flexibly by blending linear and non-linear access according to individual needs and situations during tasks. Using a Wizard-of-Oz protocol, we initially embodied SelfIE instructions within a toy-block assembly context and compared it with baseline instructions offering linear-only access (N=21). Results show a 71% increase in user preferences due to its ease of reflecting individual differences, empirically supporting the prior proposal. Besides, our observations identify three strategies for flexible access and suggest the potential of enhancing the user experience by considering cognitive processes and implementing flexible access in a wearable configuration. Following the design phase, we translated the WoZ-based design embodiment as working prototypes on the tablet and OHMD to assess usability and compare user experience between the two configurations (N=8). Our data yields valuable insights into managing the trade-offs between the two configurations, thereby facilitating more effective flexible access development.

Figures (6)

• Figure 1: In the context of the toy-block assembly task, with SelfIE assembly instructions, assemblers can flexibly navigate instructions either linearly (by uttering "next" or "previous") or non-linearly (by uttering "this one" and showing toy block(s) towards the front-located camera). After each utterance, the corresponding step would be displayed on the screen as requested by the assembler.
• Figure 2: For our design iterations, we prepared three LEGO® vehicle models of different complexity (from left to right: the training model consisting of 24 pieces; the simple and complex models consisting of 48 and 84 pieces, respectively, for test trials to test our design in a different context robustly). To vary the complexity in the three models, we considered three factors introduced in richardson2004identifying: the same structure among the three models but different numbers of assembly components and sub-parts.
• Figure 3: The provided table displays the distribution of participants across SelfIE and conventional instructions based on their total voice command utterances while assembling models of similar complexity, each consisting of 28 instructional steps. The following stepped area chart visualises this distribution, highlighting the variation across the two types of instructions. If participants carried out their assembly tasks almost equally across both types of instructions, the stepped area chart would appear almost overlapped.
• Figure 4: This chart visualises participants' diverse instructional navigation paths by mapping between voice commands uttered by each participant and access operations presenting in either linear in blue/circle or nonlinear. The operations for situational nonlinear access were further decomposed into the three identified strategies: selective-skipping (in orange/square), block-scanning (in red/plus), and debugging (in cyan/ex).
• Figure 5: This figure shows the time difference between using conventional and SelfIE instructions for each participant. Note that positive values (in blue) indicate that SelfiE instructions took longer, while negative values (in red) indicate that conventional instructions took longer. The colour saturation of each bar corresponds to the magnitude of the time difference.