UCEID - The best of both worlds: Combining

0 downloads 0 Views 576KB Size Report
aims to make constraints of the system and environment explicit, so that the ..... Glaser, Barney G. and Strauss, Anselm L. (1967) The discovery of grounded ...
UCEID - The best of both worlds: Combining Ecological Interface Design with User Centred Design in a novel HF method applied to automated driving. Kirsten Revell1,*, Pat Langdon2, Mike Bradley2, Ioannis Politis2, James Brown1 Simon Thompson3, Lee Skrypchuk3, Alexandros Mouzakitis3 and Neville Stanton1 1

Human Factors Engineering, Transportation Research Group, Building 176, Boldrewood Innovation Campus, University of Southampton, Burgess Road, Southampton, SO16 7QF, UK. {K.M.Revell, J.W.Brown, Neville.Stanton}@Soton.ac.uk 2 Engineering Design Centre, Department of Engineering, University of Cambridge, Trumpington Street,Cambridge, CB2 1PZ, UK [email protected] {mdb54, ip332}@cam.ac.uk 3 Engineering Design Centre, Department of Engineering, University of Cambridge, Trumpington Street,Cambridge, CB2 1PZ, UK {sthom261, lskrypch, amouzak1}@jaguarlandrover.com

Abstract. Drawing together the strengths of Ecological Interface Design (EID) and inclusive User Centred Design (UCD), UCEID is a novel Human Factors (HF) method for complex sociotechnical systems. This method ensures user needs are appropriately represented in the constraints based models provided by the Cognitive Work Analysis (CWA) framework. The range of methodological approaches adopted, the advantages, disadvantages, tools and training times of UCEID are described. Examples of how to apply the method to the the domain of automated driving to produce design concepts are provided, focusing on interactions between drivers and semi autonomous vehicles for a planned BASt level 3 vehicle to driver handover. Keywords: Human Factors Methods · Human-systems Integration · Autonomous Vehicles · User Centered Design · Cognitive Work Analysis · EID · Inclusive Design

1

Introduction

Human Factor (HF) methods exist to tackle problems relating to the interaction between human and other elements of the system, that existing methods of design, evaluation and procurement have failed to address. These types of problems are often resistant to purely technical interventions resulting in less effective system performance [1]. With ever increasing rates of technological advancement, it is more difficult for companies to compete on functionality, reliability or cost [2]. Human Factors

methods offer a means to provide a competitive edge by harnessing technology to enable people to accomplish meaningful, real-world tasks. Human Factors methods fall into range of categories that are relevant for application at different parts of the design process [1]. The UCEID process includes a combination of ‘Data Collection’, ‘Task Analysis’ and ‘Cognitive Task Analysis’ techniques. Figure 1 shows how different methods are suitable at different stages of the design process. The UCEID method is positioned early in the design process to allow ‘analytical prototyping’, the means of applying HF insights to systems or designs that are yet to exist in physical form. It covers a combination of ‘Identify need’ and ‘Develop Concept’ stages of the design process taking the analyst to ‘initial design concept’ stage, not final concept. Key finding from inclusive User Centered Design advise an active process of linked iteration between technology prototypes and user trials is necessary to meet the dual needs of diverse user demographics and technology delivery requirements [3] (see figure 1).

UCEID&

Fig. 1. Diagram to show where the UCEID method fits into the design process in relation to other HF methods. Amended from Stanton et al (2013).

1.1

Why use UCEID?

UCEID is a novel HF method that integrates relationships between EID [4] and inclusive Human Centered Design by combining existing methodology from the Cognitive Work Analysis (CWA) framework [5][6][7] and Inclusive User Centered Design [8][9]. Ecological Interface Design (EID) is based on Gibsonian methodology that aims to make constraints of the system and environment explicit, so that the appropriate action is apparent to the system user [4]. Whilst both EID and UCD emphasize the user at each stage of the design process, they differ in their approach. UCD has a greater focus on end users wants, needs and limitations within single user actions, whereas EID focuses on incorporating user wants and needs within a complex system, constrained by the values and purpose of the overall system. Some of the values relate to stakeholders that may at times be in conflict with the end user wants and needs. Both methods aim to provide the user with a visual ‘mental model’ to guide possible action [10], but EID’s remit within complex systems extends to providing a mental model that enables the user to troubleshooting unanticipated events [11]. UCDs focus

on usability can ensure solutions to meet the EID remit are easy to use and learn. EIDs focus on values and constraints based on the overall purpose of the system, ensure that the design solutions proposed by UCD are relevant and systematically prioritized. The UCEID approach engages with stakeholders, Subject Matter Experts (SMEs) and users to produce outputs that generate design requirements. Initial design concepts are then produced following a design workshop and concept filtering activity. The UCEID method is best suited to complex sociotechnical systems where the user plays a critical role in the interaction. Domains that are complex exude some of all the following qualities: high risk, dynamic, uncertain, with interconnected parts. Vehicle initiated vehicle to driver (VtoD) handover in a BASt Level 3 autonomous vehicle fits the criteria of a complex sociotechnical system and will be used to illustrate application of the method.

2

The UCEID Method

The UCEID method is complied from a rich range of activities starting with defining the scenario and aims of analysis, and ending with the generation of design concepts. Following the recommended criteria for depicting a method [1], figure 2 depicts a step by step process that can be followed ‘like a recipe’. The flowchart in figure 2 describes the sequence of 16 steps (rectangles) that include: literature review; data collection; thematic analysis; cognitive task analysis; consolidation and ideas generation, and; filtering and checking. Steps within boxes can be undertaken in parallel and decision points (diamonds) and feedback loops occur at different points in the process. The type of activity is shown by the line-style (figure 2). Not all aspects of the method need to be done at once, or to the same level of detail. The method can be reapplied following testing cycles as the user needs and domain constraints become more clearly specified. This section will summaries the different types of activities in the UCEID flowchart to provide the reader with examples of the types of outputs at different stages. Before embarking on the specific activities, the aims of analysis, target scenario and boundaries of analysis need to be clearly defined. For example: analysis of ‘transfer of control in autonomous vehicles’ for the scenario ‘planned transfer of control from vehicle to driver on a highway in a BASt level 3 vehicle’.

Fig. 2. Flowchart showing the sequence of steps and range of activities for the UCEID Method

2.1

Literature Review

Literature Review occurs primarily at Step 2. Initial Research (figure 2) but can continue throughout the process as needed. For novel systems, the domain under analysis, and parallel systems that have relevant features to the scenario should be considered. This can be through document analysis (such as user guides, reports), which will depend on what is currently available for the domain. Where possible, observation of the work in context is highly recommended. However, focus purely on current systems can risk producing design proposals that are simply modifications of current technology or practice. The objective of inclusive design step is to establish the opinions and preferences of likely users towards the target technology to be designed. Different members of a research group can undertake ‘Researching domain & parallel domains’ (e.g. transfer of control in the aviation domain [12] and requirements of advanced driving [13]) and ‘Inclusive design’ (e.g. reference to literature and national surveys relating to user needs for automation [14]), in parallel to scope the research. This initial research informs all subsequent steps of UCEID method. 2.2

Data Collection

Data collection occurs at Step 3. Initial Data Collection and Step 5. Focus Groups (figure 2). In Step 3, the different types of data collection can again occur in parallel. For the chosen scenario, ‘SME semi-structured interviews’ would occur with automation experts. ‘Technology analysis and benchmarking’ would be used to understand current commercial offerings and relevant technologies relating to autonomous vehicles. ‘User interviews’ would occur with target users of autonomous vehicles following standard approaches [15] [16]. In Step 5, focus groups representing an inclusive range of the target users is recommended for generalizable outputs (e.g. driving population with balanced gender mix and range of legal driving age groups, from variety of geographical regions). Two researchers (a moderator and facilitator) and a set of 57 participants per group are suggested. 2.3

Thematic Analysis

Thematic Analysis occurs at Step 4. Thematic Analysis 1, Step 6. Thematic Analysis 2 and Step 7. User Themes and Preferences. Thematic Analysis is undertaken using a grounded theory approach [15]. The data from Step 3 is thematically analyzed with the help of qualitative software such as NVIVO. Generated themes should be defined and compared with independent analysts to create a final set. Themes relating to autonomous vehicles included ‘Trust’, ‘Control’ and ‘Safety’. In Step 6, data from previous steps are consolidated and triangulated according to the data source [17]. The most robust data is tabulated to capture requirements for step 7. This takes themes from focus groups and interviews and converts them into insights (e.g. ‘auditory alert of increasing urgency’ when handover is required, as users ‘anticipate being otherwise occupied’ during automated mode).

2.4

Cognitive Task Analysis

The CWA elements include Steps 8-12 (figure 2). Of these, steps 8 and 9 fulfill the EID requirements [4] but other CWA steps are optional depending on focus and resources. Step 8 - Work Domain Analysis (WDA) Abstraction Hierarchy (AH) marks the initial phase of the CWA framework. WDA describes the constraints that govern the purpose and function of the systems under analysis. The output of this step is the creation of an AH [5][6]. This is a diagram constructed of 5 levels of abstraction, from the most abstract level ‘purpose’ (e.g. planned transfer of control back to driver), to ‘values’ (e.g. safety), ‘purpose related functions’ (e.g. ensure driver situation awareness), ‘functional purpose’ (e.g. communicated vehicle status to driver) to the most concrete form, ‘physical object’ (e.g. head up displays). Step 9 - Control Task Analysis (ConTA) looks in depth at constraints that are imposed by tasks that need to be carried out in specific situations to reveal not only the constraints, but the level of flexibility in how activities can be achieved [18]. For autonomous vehicles, three time periods were used to constrain the scenario (e.g. Pre-handover, Handover & Posthandover). Step 10 - Strategies Analysis is useful for exploring the flexibility within a system for different types of strategies within different contexts [6][19]. The strategy adopted by an agent in a particular situation may vary based on workload, or vary between agents (e.g. different strategies for different user groups (e.g. elderly versus young drivers) or the same user undergoing different journeys (e.g. work or leisure). Step 11 - Social Organisation and Cooperation Analysis (SOCA) identifies how control tasks can be distributed between human and non-human agents within the system [19]. It determines how social (e.g. driver) and technical (e.g. automation) aspects of a system can work together to enhances performance as a whole. SOCA reveals the flexibility of system to deal with unanticipated events. Step 12 - Worker Competencies Analysis (WCA) is the final CWA element and involves identifying the cognitive skills required for control task performance in the system under analysis. These are classified using the Skills, Rules and Knowledge (SRK) framework [20]. 2.5

Consolidation and Ideas Generation

Consolidation and Ideas Generation occur in Steps 13 & 14 (figure 2). Step 13 Pre-workshop takes any new insights gained from steps 9 onwards to enable recalibration of the scenario (Step 1) user preferences (Step 7) and AH (Step 8). This allows a clear and consistent presentation of the context and design tasks to participants in the Design Workshop (Step 14). The basis for concept filtering (Step 15) is also defined here. Step 14 - Design Workshop is used to shares and review the knowledge gained from previous steps about the problem and solution spaces, then to generate concept solutions to solve that problem (e.g. handover interaction designs, alerts or alarms, multimodal variations). Inclusive design process should be adopted to explore needs and create solutions. 2.6

Filtering and Checking

A key element of UCEID occurs in its method of filtering and checking. There are three stages at which the outputs are sanity checked so that they fulfill both UCD and

EID requirements. After construction of the AH in Step 8, its elements are reviewed to determine if user preferences are represented. If not, additional elements are added to accommodate these preferences. This process requires reinterpreting and rewording user preferences to fit into the most appropriate AH layer. It should be made clear that only user preferences and themes applicable to the defined Domain Purpose are added to the AH, as this constrains the analysis. User preferences will have been generated from interviews and focus groups that fall outside the constraints of the system. The AH therefore provides a means of weighing up the relative importance of themes and preferences from user centered activities. During Step 15 - Concept Filtering Ideas and concepts from the workshop are transcribed and a design criterion matrix [21] relating to the AH in step 8 is constructed. Design features are rated according to inclusion and adherence to the criteria to produce candidate ‘output concepts’. Based on the criteria for inclusion agreed in the Pre-Workshop (Step 13), the concepts need to be filtered by the agreed level of adherence to the AH (and any other factors deemed critical). This iterative process continues until only design concepts that fulfill these criteria are progressed toward Step 16 (Concept development). Following the UCEID process, final output concepts are created through an iterative process applied to the candidate concepts. Interface Analysis Methods are applied to assess their viability, and Design Methods are used to create an initial version of the concepts [1]. The Output Concepts should be documented with sufficient detail to enable their implementation and evaluation in later stages of the design process, in order to arrive at the final design. The output concepts related to interactions between drivers and semi autonomous vehicles for a planned BASt level 3 vehicle to driver handover comprised of: 1) storyboards of interactions between driver and automation, with multimodal feedback; 2) interface designs of Head Up Displays (HUD), Head Down Displays (HDD) and ambient displays to improve trust and mode awareness, and; 3) verbal scripts for automated or intelligent guidance for enhancing driver Situation Awareness (SA) prior to taking back control of the vehicle.

3

Methodological considerations

3.1

Advantages

The UCEID method combines a systems level view of analysis with user centered design to provide design requirements that support the overall purpose of the domain, as well as the needs and expectations of users. It is a generic and highly flexible method that can be applied to a variety of different domains for different purposes and to different depths depending on requirements of the analysis. The method is based on a sound underpinning of theory by incorporating the CWA framework and inclusive design principles and is particularly suited to complex systems. The diversity of the different approaches within UCEID promotes a comprehensive analysis. The emphasis on inclusive user groups ensures that a wide spectrum of capabilities is taken into consideration in the design phase that can provide a satisfactory user experience not

only for the population with capability limitations, but for a larger proportion of the population. 3.2

Disadvantages

The steps in the UCEID method that relate to the CWA framework can be complex so practitioners may require considerable training in their application if the analyst lacks this experience. As some of the steps from the CWA framework are still in their infancy, there is limited proscriptive guidance available for the novice. A number of the of the steps can be time-consuming to apply although and some outputs (e.g. the AH) can be large, unwieldy and difficult to present for highly complex systems. It can be challenging to recruit a user group which both complies to generalizability of the entire population as well as cover the range of capability limitations. It may not be reasonable for a single research or practitioner group to be skilled in the wide range of methodological approaches within UCEID. Collaborations with other research institutions or practitioners, and the use of consultants may be required for specialist skills. The number of steps extends application time making this method best suited to longer delivery timeframes. As a new method, the reliability and validity has yet to be established. 3.3

Training and Application Time

Due to the complexity of the steps of the process relating to the CWA framework elements, training time associated with this method is high. The training requirements for a full Inclusive design approach to UCEID require practitioners that have participated in a professional development course covering inclusive or Universal design. Secondarily, many of the methods used require familiarity and practice, such as interviews, focus groups and the Design Workshop. The authors recommend six months be allocated for the full method comprising of all 16 steps. For a smaller scale project, or where the optional steps (9-12) are omitted, the UCEID method may be applied in a few weeks. 3.4

Tools

The tools required for UCEID are modest. Most stages can be undertaken with pen and paper though audio and visual recording equipment is strongly recommended for interviews, focus groups and workshops to allow content to be revisited in context. The use of software tools can facilitate the method, in particular the CWA tool by the HFI-DTC, and thematic analysis and data management tools such as Nvivo and MS Excel. For workshops and focus groups, rapid prototyping materials such as building blocks, toys, colouring materials and modeling clay are beneficial to quickly understand user wants.  

4

Summary

This paper has presented a high-level overview of UCEID, incorporating the best of EID and Inclusive UCD principles in a novel method to aid HF and Design practitioners. The range of approaches within the method has been highlighted. Generic advice followed by examples relating to VtoD handover in a BASt Level 3 autonomous vehicle is provided. Practical considerations for application of the method have been described. Acknowledgments. This work was supported by Jaguar Land Rover and the UKEPSRC grant EP/011899/1 as part of the jointly funded Towards Autonomy: Smart and Connected Control (TASCC) Programme.

References 1. Stanton, N. A., Salmon, P. M., & Rafferty, L. A. Walker, G.H., Baber, C., Jenkins, D.P., (2013). Human factors methods: a practical guide for engineering and design. Ashgate Publishing Ltd. 2. Green, W.S. and Jordan, P.W. (1999) Human Factors in Product Design: Current Practice and Future Trends. London: Taylor & Francis. 3. Langdon, P.M., Lazar, J., Heylighen, A. and Dong, H. eds., 2014. Inclusive designing: Joining usability, accessibility, and inclusion. Springer. 4. McIlroy, Rich C., and Neville A. Stanton. Ecological interface design two decades on: whatever happened to the SRK taxonomy? IEEE Transactions on Human-Machine Systems 45.2 (2015): 145-163. 5. Rasmussen, J., Pejtersen, A. and Goodstein, L.P. (1994) Cognitive Systems Engineering. New York: Wiley. 6. Vicente, K.J. (1999), Cognitive Work Analysis: Toward Safe, Productive, and Healthy Computer-based Work, Mahwah, NJ: Lawrence Erlbaum Associates. 7. Jenkins, D.P., Stanton, N.A., Salmon, P.A., Walker, G.H. (2009). Cognitive work analysis: coping with complexity. Ashgate Publishing, Ltd. 8. Czajkowski, K., Fitzgerald, S., Foster, I., Kesselman, C.: Grid Information Services for Distributed Resource Sharing. In: 10th IEEE International Symposium on High Performance Distributed Computing, pp. 181--184. IEEE Press, New York (2001) 9. Langdon, P., and Thimbleby, H., (2010), Inclusion and Interaction: Designing Interaction for Inclusive Populations, Editorial for Special Edition of Interacting with Computers, V22, pp 439-448, Elsevier. 10. Norman, D., 2013. The design of everyday things: Revised and expanded edition. Basic Books (AZ). 11. Burns, C.M. and Hajdukiewicz, J., 2004. Ecological interface design. CRC Press. 12. Handover/Takeover of Operational ATC Working Positions/Responses https://www.skybrary.aero/index.php/Handover/Takeover_of_Operational_ATC_Working _Positions/Responses 13. Roadcraft: the police driver and rider handbooks for better, safer driving http://www.roadcraft.co.uk 14. IM Travellers needs Survey, (2015), https://ts.catapult.org.uk/current-projects/travellerneeds-uk-capability-study/

15. Glaser, Barney G. and Strauss, Anselm L. (1967) The discovery of grounded theory: strategies for qualitative research. Chicago:Aldine 16. Breakwell, G.M., 2000. Introducing research methods in psychology. Research methods in psychology, Sage, London 17. Ewe, F. (2006), An introduction to qualitative research (3rd Ed). Sage, London. 18. Naikar, N., Moylan, A. and Pearce, B. (2006) Analysing activity in complex systems with cognitive work analysis: Concepts, guidelines, and case study for control task analysis, Theoretical Issues in Ergonoimics Science, 7, 4, 371-394. 19. Naikar, N. (2006), An examination of the key concepts of the five phases of cognitive work analysis with examples from a familiar system, Proceedings of the Human Factors and Ergnomics Society 50th Annual Meeting, 447 – 451. 20. Rasmussen, J. (1985), The role of hierarchical knowledge representation in decision making and system management, IEEE Transactions on Systems, Man and Cybernetics, 15, 234-243. 21. Pugh, S., 1993. Total Design: Integrated Methods for Successful Product Engineering, Addison.