International Transportation (71) 1 | 2019 22 PRODUCTS & SOLUTIONS Vehicle design Secure, helpful, lovable Incorporating user needs in the design of autonomous vehicles systems for public transport User-centered design, User experience, Autonomous shuttles, Demand-responsive transport, Mobility as a service Autonomous, shared, and electric - this is the vision for future transport services that enable both efficient and climate-friendly mobility. The success of such services will crucially depend on their actual use by the population, which is in turn determined by perceptions of their usefulness, ease of use, safety, and attractiveness. The new features even entail some new challenges to users. We present methods to identify user needs and potential use barriers early in the process of designing autonomous vehicles systems for public transport, and give examples from our user-centered research. Annika Dreßler, Jan Grippenkoven, Meike Jipp, Klas Ihme, Uwe Drewitz T he public transport of the near future is envisioned to support both efficient and climatefriendly mobility: powered by sustainable drive technology, with autonomous driving allowing flexible, safe, and yet cost-efficient operation, as well as digital technology enabling an intelligent pooling of passenger (and goods) transport, and seamless connections [1]. Passengers share rides and are at minimum as fast to arrive at their destinations as they would be with their individual vehicle. Small shuttle buses are apt to serve even the capillaries of cities and pick up and drop passengers on demand almost right on their doorsteps, as door-to-door services or feeders to and from high capacity transport. A promising vision - however, the success of a transport system with the above features will crucially depend on its acceptance and actual use by the population. For this, the development of these systems must not only focus on the technical and legal aspects, but also offer viable solutions to new challenges that arise from a user perspective. Moreover, the mobility services offered must be as attractive to users as to compete with motorized individual transport. New challenges to users Certain features of flexible autonomous shuttles - e.g. the absence of a driver, the relatively small size of the vehicles, the flexible routing, timing and access points - entail new challenges from a user perspective. Imagine, for example, this rather simple situation: You want to take a bus to a destination that you do not routinely travel to, such as a station in another town or an event location in your city. Have you ever asked the driver if this bus is right for you? We guess you have. In the absence of a driver, the design of the system should still allow getting this information as fast and easy. As our research on user needs revealed, from a passenger perspective, the drivers of public transport vehicles nowadays fulfil a lot more functions than just driving [2]. Apart from the service and information function that is just shortly sketched in the above example, indeed, a lot of passengers perceive the driver as an instance of supervisory control that helps them feel safe. The question follows how to design driverless systems that support the same (or better) personal experience of safety and controllability. Motivating users to prefer the new public transport systems over the individual motor vehicle will moreover require some efforts to enhance service quality - not only with regard to the availability and time dimension, but also looking at user convenience and added value. Systems design must find answers to these and further challenges. How to incorporate user needs in systems design To design new public transport systems that will be the option of choice - not only to those who run short of alternatives, but to all - the satisfaction of user needs has to be of top priority in system development. Therefore, methods to incorporate user needs have to be applied throughout the design process, beginning from the early conceptual stages and continuing through iterations of prototyping until the final product [3, 4]. In the following, we introduce a selection of methods to achieve this Example of a driverless shuttle: Project HEAT in Hamburg, Germany © iav.de International Transportation (71) 1 | 2019 23 Vehicle design PRODUCTS & SOLUTIONS and give examples from our user-centered research. Applying models of user needs and acceptance Relevant models and empirical findings from engineering psychology are a good starting point to approach the assessment of user needs and inspire design for acceptance. There are efforts like that of Venkatesh et al. [5, 6] who integrated ideas from eight models into a Unified Theory of Acceptance and Use of Technology. The model has already been applied to autonomous vehicles and demand-responsiveness in public transport [7-9]. Figure 1 shows our adaption of the model to the context of electric autonomous shuttles in public transport. To dig deeper into the nature of human needs and explore what determines the “expected usefulness”, “expected effort”, and so on, more psychological models are available, as for example the needs pyramid [10] (see figure 2), or self-determination theory that states competence, autonomy and relatedness as basic human needs [11]. Importantly, these models can not only help in understanding how public transport can be made more attractive, but also in recognizing “hard cases” of design features serving certain needs (e.g. the fun in steering a moving vehicle by oneself ). Thus, in the case of certain features, it might not be sufficient to improve the design of public transport systems, but complementary action might be necessary to dampen the attractiveness of individual motorized transport [12]. User, context and task analysis The first thing to do in applying any of the following methods is to think about who typical users are and what they try to achieve [4]. In public transport, there is a broad spectrum of users with regard to age, gender, education, language skills, physical size, physical abilities, previous experience, and so forth. A good approach is to define multiple user groups based on the available statistics. To represent their characteristics in a concrete and tangible way, relevant features can be combined in so-called personas - a description of key prototypical users including not only a name, age, and gender, but also specifics such as goals, work, typical activities, and past experience [13]. Moreover, it is important to think about different contexts of use: For instance, an autonomous shuttle without additional safety systems might be suitable to a person in one context (e.g. by day when lots of other people are around), but not in another (e.g. at night, being alone with only one other person). Finally, the task of using the system to get from A to B needs to be analyzed in the light of the defined users and use contexts [14, 15]. Focus groups Focus group are a suitable method to obtain user requirements regarding future transport systems and let users participate in the early steps of the development. It is good to involve a broad variety of users, as this sheds light on diverse interests and requirements that might sometimes even be conflicting. A focus group typically comprises six to ten participants and is moderated by a subject matter expert or a neutral person. Participants are supposed to work together in a creative, cooperative and solution oriented way. The moderator supports the meeting with an agenda, stimulates the participants’ considerations with key questions, controls the time and ensures documentation. As an example, a focus group on perceived safety of users in flexible and driverless bus shuttles was conducted by Grippenkoven et al. [2]. After an introduction to the objectives, the participants first shared their knowledge about flexible bus shuttles and autonomous driving. The moderator specified the target system, a driverless transport service based on electrified and shared busshuttles that need to be booked through an app because they operate flexibly with regard to route and schedule. The concept was illustrated with explanatory videos. The participants’ first task was to individually elaborate on the question: What would determine my personal safety experience in these shuttles? After participants had introduced their thoughts and ideas to each other, they clustered the documented ideas and identified 17 aversive scenarios (see figure 3). In a subsequent idea phase, measures that enhance the personal experience of safety in the context of the introduced public transportation were collected. These measures serve as a valuable basis for further evaluations, for example in the context of prototypical mobility services in real world environments. Figure 1: A model of user acceptance and use of electric autonomous shuttles. Primary predictors (on the left) are expected to have a direct influence on use intention. User characteristics (bottom) can influence the relation between the predictors and use intention. Figure 2: Maslow’s pyramid of human needs, with examples of the individual levels (adapted from [10]) International Transportation (71) 1 | 2019 24 PRODUCTS & SOLUTIONS Vehicle design Surveys and interviews Users as well as current non-users can also tell us a lot about what they need and why they prefer certain mobility alternatives in surveys and interviews. In the case of advanced public transport systems, which are not yet widespread as objects of interaction experience, there is the specific challenge of enabling the respondents to make valid statements about their appraisal and requirements. Therefore, in the absence of an existing system to be evaluated, some effort should be put in an initial introduction that gives the respondent a vivid impression of the relevant system features. This can be achieved, for example, by narration, illustrated with images or videos, as shown above. In cases where a more “physical” idea of the system is needed (e.g. to assess which seat layout supports the best user experience), prototypes can be used. They can be simple, as a small room with some chairs, or more sophisticated, as an existing vehicle with (some) similar features as the target system. While interviews and focus groups mainly yield qualitative data (answers to open questions on user needs like What? How? etc.), written surveys are often used to obtain quantitative data (How much? How often? etc.), e.g. by asking users to give assessments on scales. The two methods can be usefully combined to first explore relevant aspects and design ideas and then assess the importance and generalizability of these aspects in a more representative sample of users. For instance, expanding on the results of the focus group on perceived safety, Grippenkoven et al. [2] conducted a survey in which potential users first gave ratings of how intimidating they found each of the aversive scenarios that had been identified, and afterwards assessed the effectivity of each of the proposed measures to improve experienced personal safety. Observation Subjective data as it comes from surveys, interviews, and focus groups is an invaluable source of information to explore user needs. However, some aspects of behavior and experience can hardly be to put into words, e.g. because the underlying processes are implicit and unconscious. Therefore, we can learn more about preferences of users and their difficulties in dealing with certain tasks by watching their interactions with transport systems. Examples: In what order do users choose available seats? What activities do passengers perform while traveling? How do they interact with each other in the vehicle? How do they react to disturbances (e.g. bumpy road, stop-and-go, drunk person entering)? Obviously, observation requires a more advanced stage of development of the transport system than just a concept. Thus, it can be applied to evaluate the target system when it is operating. However, depending on the design question, simulations or prototypes (see above) might also be used in earlier design phases. Scientific observation is structured and systematic by defining a priori a small range of target behaviors or events to be focused on and how to document the results (e.g. by counting how often events occur or measuring their duration [16]). Therefore, to enable Figure 3: Aversive scenarios identified in a focus group on personal safety experience in shared autonomous shuttles with flexible routing and timing [4] Figure 4. Real-time user state monitoring for assessment of user needs in autonomous vehicles for public transport. Based on different sensors (e.g. cameras or physiological sensors) the user state is identified using machine learning methods and integrated with information from the context. Based on this, the current need of the user can be determined. International Transportation (71) 1 | 2019 25 Vehicle design PRODUCTS & SOLUTIONS the scoring of the same episode for different target events, video recordings are of great help if they can be attained in accordance with data protection. Some important pillars of data protection are to work with low resolution that does not allow the identification of faces, to protect the raw data from access by third parties and to delete them right after the analysis to deduct the results. In the development of new transport systems, there might also be the possibility to do a test with dedicated users who can give informed consent. User-State monitoring In addition to the human eye, sensors can help to determine user needs during the design process by evaluating the users’ expression of inner states through behavioral channels such as posture, gestures, gaze behavior, vocal and facial expressions, or physiology [17-19]. Moreover, real-time user state monitoring can be used to develop systems that recognize user experience and can therefore react adaptively to negative episodes, such as discomfort, frustration, fear, stress or uncertainty. For instance, if a passenger of an autonomous shuttle feels uncertain about whether the automation can handle a complex situation or not, this could be sensed by the system. In case of detected uncertainty, information about the current situation representation of the automation could then be displayed on an invehicle screen or via a smartphone app to satisfy the user’s need for system transparency and safety. The sensing in such systems (see figure 4) could be implemented, for example, through camera-based methods or peripheral physiology [20-23]. It must be complemented with information about the context (traffic, environment, weather, etc.) or history of use, because similar negative experiences may have different causes - e.g. too fast driving style or rude behavior of another passenger - and thus require different measures - e.g. more defensive driving or a contact person for help. The use of multiple sensor views and methods based on wearables (e.g. fitness bands) could help to take into account the specifics of public transport to develop such user-oriented automated systems. Early user focus spares trouble These are only a few examples how usercentered methods can be used to incorporate user needs in the development of advanced systems for public transport. The choice of the best methods is made based on the specific system and research question and, not least, the available budget. Methods can be scaled to the available resources [24] by various parameters, such as low-cost prototyping, smaller sample sizes, or the application of expert evaluation (Heuristic Evaluation, Cognitive Walkthroughs) where user studies appear impracticable [3]. The early focus on user needs is most important, as it is much more practicable and much cheaper to remedy design shortcomings in the development process than after system implementation. Experience shows that it is worthwhile to invest in exploring user requirements, or, as Wickens [4] put it: “Being first, being best, and even being right do not matter; what matters is understanding what your customers want and need.” ■ REFERENCES [1] Fulton, L. M. (2018). Three Revolutions in Urban Passenger Travel. Joule, 2(4), 575-578 [2] Grippenkoven, J., Fassina, Z., König, A. & Dreßler, A. (2019). Perceived Safety: a necessary precondition for successful autonomous mobility services. In D. de Waard, K. Brookhuis, D. Coelho, S. Fairclough, D. Manzey, A. Naumann, L. Onnasch, S. Röttger, A. Toffetti, and R. Wiczorek (Ed.), Proceedings of the Human Factors and Ergonomics Society Europe Chapter 2018 Annual Conference [3] Nielsen, J. (2009). Usability engineering. Amsterdam: Morgan Kaufmann [4] Wickens, C. D., Gordon, S. E. & Liu, Y. (2004). An introduction to human factors engineering. Upper Saddle River, N.J: Prentice Hall [5] Venkatesh V., Morris, M., Davis, G. & Davis, F. (2003). User Acceptance of Information Technology: Toward a Unified View. MIS Quarterly, 27(3), 425 [6] Venkatesh, V., Thong, James Y. L. & Xu, X. (2012). Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. Management information systems : MIS Quarterly, 36(1), 157-178 [7] König, A., Grippenkoven, J., & Jipp, M. (in prep.). The Actual Demand Behind Demand-Responsive Transport: Applying the Unified Theory of Acceptance and Use of Technology to Explain Usage Intentions of Demand-Responsive Bus Services [8] Madigan, R., Louw, T., Dziennus, M., Graindorge, T., Ortega, E., Graindorge, M., & Merat, N. (2016). Acceptance of Automated Road Transport Systems (ARTS): An Adaptation of the UTAUT Model. Transport Research Arena TRA2016, 14(Supplement C), 2217-2226 [9] Nordhoff, S., van Arem, B., Merat, N., Madigan, R., Ruhrort, L., Knie, A., & Happee, R. (2017). User Acceptance of Driverless Shuttles Running in an Open and Mixed Traffic Environment. In Proceedings of the 12th ITS European Congress. Strasbourg, France, 19-22 June 2017 [10] Maslow, A. (1943). A theory of human motivation. Psychological Review, 50, 370-396 [11] Ryan, R., & Deci, E. (2017). Self-Determination Theory: Basic Psychological Needs in Motivation, Development, and Wellness. New York: Guilford [12] Graham-Rowe, E., Skippon, S., Gardner, B., & Abraham, C. (2011). Can we reduce car use and, if so, how? A review of available evidence. Transportation Research Part A: Policy and Practice, 45(5), 401-418 [13] Cooper, A. (2004). The inmates are running the asylum. Indianapolis: Sams [14] Annett, J. (2005). Hierarchical Task Analysis. In N. Stanton, A. Hedge, K. Brookhuis, E. Salas, & H. Hendrick (Eds.), The handbook of human factors and ergonomics methods . New York, NY: Taylor & Francis [15] Naikar, N., Moylan, A., & Pearce, B. (2006). Analysing activity in complex systems with cognitive work analysis: concepts, guidelines and case study for control task analysis. Theoretical Issues in Ergonomics Science, 7(4), 371-394 [16] García-López, M. D. C. (2016). Systematic Observation of Behaviors and Environmental Events Using the Lag Method. Perceptual and Motor Skills, 67(1), 255-262 [17] Crivelli, C., & Fridlund, A. J. (2018). Facial Displays Are Tools for Social Influence. Trends in Cognitive Sciences, 22(5), 388-399 [18] Erickson, K., & Schulkin, J. (2003). Facial expressions of emotion: A cognitive neuroscience perspective. Brain and Cognition, 52(1), 52-60 [19] Scherer, K. R. (2016). What are emotions? And how can they be measured? Social Science Information, 44(4), 695-729 [20] Beggiato, M., Hartwich, F., & Krems, J. (2018). Using Smartbands, Pupillometry and Body Motion to Detect Discomfort in Automated Driving. Frontiers in Human Neuroscience, 12, 3138 [21] Bořil, H., Boyraz, P., & Hansen, J. H. L. (2012). Towards Multimodal Driver’s Stress Detection. In J. H. Hansen, P. Boyraz, K. Takeda, & H. Abut (Eds.), Digital Signal Processing for In-Vehicle Systems and Safety (pp. 3-19). New York, NY: Springer New York [22] Ihme, K., Unni, A., Zhang, M., Rieger, J. W., & Jipp, M. (2018). Recognizing Frustration of Drivers From Face Video Recordings and Brain Activation Measurements With Functional Near-Infrared Spectroscopy. Frontiers in Human Neuroscience, 12, 669 [23] Zhang, M., Ihme, K., & Drewitz, U. (2019). Discriminating drivers’ emotions through the dimension of power: Evidence from facial infrared thermography and peripheral physiological measurements. Transportation Research Part F: Traffic Psychology and Behaviour, 63, 135-143 [24] Nielsen, J. (1994). Guerrilla HCI: Using Discount Usability Engineering to Penetrate the Intimidation Barrier. In R. G. Bias (Ed.), Costjustifying usability (pp. 245-272). San Diego: Academic Press Jan Grippenkoven, M.Sc. Dept. Human Factors, DLR Institute of Transportation Systems, Braunschweig (DE) jan.grippenkoven@dlr.de Meike Jipp, PD Dr. Dept. Human Factors, DLR Institute of Transportation Systems, Braunschweig (DE) meike.jipp@dlr.de Annika Dreßler, Dr. Dept. Human Factors, DLR Institute of Transportation Systems, Braunschweig (DE) annika.dressler@dlr.de Klas Ihme, Dr. Dept. Human Factors, DLR Institute of Transportation Systems, Braunschweig (DE) klas.ihme@dlr.de Uwe Drewitz Dept. Human Factors, DLR Institute of Transportation Systems, Braunschweig (DE) uwe.drewitz@dlr.de