HMI: An Emotional Experience

Transcription

1 Electronics Human Machine Interface HMI: An Emotional Experience Ficosa has developed a new concept of a Human Machine Interface (HMI) controlled by a multifunctional shift element, with system information displayed on a Head-Up Display (HUD). Two alternative designs of this concept have been evaluated in order to assess their usability and the emotion experienced by users.

2 1 Introduction The development of an HMI is subject to diverse, sometimes competing, demands of customers and society. The HMI is expected to improve drivers and passengers comfort, vehicle performance and active safety through the automatic configuration and adjustment of the vehicle interior, engine elements, adaptive systems and enhanced information. But the introduction of these new functions may increase the complexity of controls and information screens. This has an important effect on the driver s workload and control of the primary task of driving, and may cause dangerous situations. It has been demonstrated that risk perception affects a driver s emotions [1], and that a positive emotional experience in a user is an important key for the success of a product in the market and one that should be taken into account by the major car manufacturers [2]. 2 HMI Design Criteria In the present project, a set of new concepts for HMIs has been developed to improve user safety and encourage emotional satisfaction. These HMIs are part of an innovative concept of communication between the car and the driver, in which user actions are centred on a multifunctional control device with system information displayed on an HUD. The user has already been taken into account from the initial steps, considering three aspects of the user/product interaction: Physical Ergonomics (anthropometry, range of movements, reach), Cognitive Ergonomics (learning and ease-of-use, memory requirements, attention) and Emotion (perceived characteristics, user satisfaction). The following design principles were applied: The shape and proportions of the control device (joystick) were adapted to the hand dimensions of the objective population, considering the 5 th and 95 th percentiles in anthropometric databases, in order to allow comfortable and stable postures which permitted precise movements. The layout of the buttons on the control device was defined according to the necessary separation and orientation, to ensure that they are easily reachable in a comfortable posture and to prevent accidental activation of the wrong button. The material texture was selected to provide a comfortable and safe grip, and to help in differentiation of the different parts by tactile clues. The activation force was defined to prevent accidental actions and improve the efficiency of use. HMI menus were designed combining fixed and changing elements in order to provide reference constants and at the same time to allow users to focus their attention on the relevant options at every moment. Navigation procedures across the HMI menus were designed according to mental models of users, providing contextual information and reducing menu levels and options to a balanced minimum for moderate memory use. Two alternative graphical menus and control devices were designed for the test with these criteria, including 17 possible functions with more than 60 discrete selections and six continuous selectors. 3 Description of the Interfaces In both designs, the main function of the HMI as a gearshift is similarly controlled: the longitudinal movement of the control device (in the X direction) controls the automatic transmission mode, and a transversal movement of the control device (in the Y direction) switches between automatic and manual modes, which are then controlled by longitudinal movements. The transmission mode is displayed on the central panel. The main difference between the two HMIs is the control of other functions, which include the configuration of the suspension height and hardness, speed limit, Automatic Cruise Control (ACC), FCW and properties of the graphical interface, as well as the selection of different driving modes with certain combinations of those configurations. The user receives visual feedback of the navigation through these functions on the HUD. For safety reasons, only the configuration of the speed limit, ACC and FCW are accessible while driving. In or- The Authors Jordi Jornet Research Director, Ficosa International, SA Ferran Esquius Advanced Engineering Designer, Ficosa International, SA Manel Seren Advanced Engineering Designer, Ficosa International, SA David Tuset Embedded Software Supervisor, Ficosa International S.A. Jose S. Solaz Automotive and Mass Transport Industry Manager, RTD Area, IBV Helios de Rosario Researcher, RTD Area, IBV 3

3 Electronics Human Machine Interface Available functions are continuously shown around the sphere, and the current function is highlighted. When the user releases the button, the HMI automatically returns to the transmission function. An exception to this way of accessing functions is the ACC, which, like the manual transmission, is accessed from the automatic transmission mode of the HMI. 4 Cognitive and Experimental Evaluation The cognitive complexity of the designs has been evaluated using Markov models [4] that quantify the intrinsic difficulty of achieving a desired state of the interface from any other state. Eq. (1) and Eq. (2) represent the Minimum Path (MP) matrices for A and B designs, considering only the navigation between advanced functions (gearshift function and user selection or configuration in the case of Interface A are not represented in these matrices). ( MP(A) = Eq. (1) ) Figure 1 & 2: Intuitive interfaces providing at the same time a wide range of sensorial feedback ( MP(B) = Eq. (2) ) der to use other functions, the vehicle must be stopped. Interface A includes a larger set of functions, and allows custom settings to be defined as shortcuts for a list of users. The menu to access these functions is displayed on the HUD when a button on the joystick is pressed. This menu is designed following the hierarchical approach used in most mobile phones, and is controlled by the longitudinal and transversal movements of the stick and the side OK / Cancel buttons, which are operated with the thumb [3]. The design of Interface B provides a less conventional navigation procedure. The movement of the control device is graphically represented as the rotation of a sphere displayed on the HUD. While the menu button on the control device is held, the movements of the device make the sphere roll, thus accessing different functions and selecting options of these functions depending on the direction of the movement. If the transitions between any two states of the HMI are considered, B turns out to be intrinsically more efficient: the longest transitions in Interface B are three steps long, while Interface A needs nine steps between certain pairs of functions. In terms of the time required, assuming that the shifter strokes consume the same time as pressing the buttons (about 630 ms [5]), and adding the time needed for handling the control device and mental processing (400 ms and 1350 ms, respectively [5]), Interface A needs 5.0 sec for switching between functions, while Interface B needs only 2.9 sec. 4

4 The reason is that Interface A has many more states than Interface B. This allows B to gather all the information of its state in two levels (current function and current option), and any of these sub-states may be changed by a single movement of the control device without intermediate actions. Moreover, thanks to the circular navigation procedure, all functions are visible at one glance, and any function is a maximum of three steps from any other one. Interface A has between two and four levels of information, and going up or down these levels requires pressing the OK or Cancel button. Thus, the path from any two states of the HMI is longer. What is more, since not all sub-functions may be viewed at the same time, the memory demand is potentially greater. The advantage of the A design is that it allows the setting of customized configurations that can be quickly accessed by shortcuts. If this option is used, the need to access advanced functions is kept to a minimum, which means that it may be potentially simpler to use it once the HMI is appropriately configured. Additionally, the navigation model based on the use of a cellular phone is well known by users, and this may improve user judgements in the early learning phase. This evaluation, which is based on the cognitive theoretical model of the user and the product, has been validated by experimental testing. The developed HMI models have been tested for usability and emotion in laboratory simulation experiments. Usability has been tested through Figure 3 & 4: Flexibility and system comprehension are essential to get a good experience of use

5 Electronics Human Machine Interface Figure 5a: Sample of the variation in skin conductivity (blue) and facial EMG (red/green) during two minutes of the trial. Figure 5b: Users are requested to perform a set of operations using the HMI while driving. the performance analysis of a set of tasks to be executed using the shift. Driver s emotion has been quantified by gathering information on the user s voice through questionnaires and by psychophysiological measurements. These measurements were compared in order to assess the user s emotional experience with the different design alternatives. The experimental procedure consisted of a simulation trial for each model, in which users had to perform a set of operations with the HMI while driving, according to instructions that were given at certain moments of the course. These driver operations were: adjust the car height and suspension hardness configure the HMI graphical features switch among driving modes ( city and comfort ) activate the ACC and set a speed limit. 5 Conclusion An adaptation trial was performed before the test trial for each interface. The operation time and errors, which are often used as indicators of the workload, were measured during the tests [6, 7]. Multivariate measurements were used for the assessment of emotion, combining physiological and subjective responses, [8]. As a physiological measurement, the heart rate is one of the most frequently used variables for assessing arousal, workload and emotion [1, 9 12]. This has been combined with other measurements, such as skin conductivity [13] and facial electromyography [14], which are also related with arousal and emotional experience while using technological products. For the subjective assessment, a reduced semantic universe of twelve adjectives was evaluated: innovative, comfortable, complex, coherent, elegant, sporty, intuitive, precise, intelligent, functional, ergonomic and personal. The results of these experiments were used to confirm the validity of the hypotheses based on Cognitive Engineering theory: that both interfaces create a positive arousal in novice users thanks to the familiarity with the navigation model of Interface A and the intrinsic efficiency of Interface B. References [1] Mesken, J.; Hagenzieker, M.P.; Rothengatter, T.; de Waard, D.: Frequency, determinants, and consequences of different drivers emotions: An on-theroad study using self-reports, (observed) behaviour, and physiology. In: Transportation Research Part F: Traffic Psychology and Behaviour (2007), vol. 10, no. 6, pp [2] Calvosa, A.: Where will the user drive future technology? In: Probing Experience Symposium. Philips High-Tech Campus, June 8 th 2006 [3] Buchanan, G.; Farrant, S.; Jones, M.; Thimbleby, H.; Mardsen, G.; Pazzani, M.: Improving mobile Internet usability. Proceedings of the 10 th international conference on World Wide Web (2001), pp [4] Thimbleby, H.; Cairns, P.; Jones, M.: Usability analysis with Markov models. In: AMC Transactions on Computer-Human Interaction (2001), vol. 8, no. 2, pp [5] Card, S.; Moran, T.P.; Newell, A.: The Psychology of Human-Computer Interaction. Hillsdale: Lawrence Erlbaum Associates, 1983 [6] Blanco, M.; Biever, W.J.; Gallagher, J.P.; Dingus, T.A.: The impact of secondary task cognitive processing demand on driving performance. In: Accident Analysis and Prevention (2006), vol. 38, no. 5, pp [7] Gallagher, J.P.: An assessment of the attention demand associated with the processing of information for In-Vehicle Information Systems (IVIS). Faculty of the Virginia Polytechnic Institute and State University, Ph.D. Dissertation, 2001 [8] Miyake, S.: Multivariate workload evaluation combining physiological and subjective measures. In: International Journal of Psychophysiology (2001), vol. 40, no. 3, pp [9] Fairclough, S.H.; Venables, L.; Tattersall, A.: The influence of task demand and learning on the psycho physiological response. In: International Journal of Psychophysiology (2005), vol. 56, no. 2, pp [10] Ryu, K.; Myung, R.: Evaluation of mental workload with a combined measure based on physiological indices during a dual task of tracking and mental arithmetic. In: International Journal of Industrial Ergonomics (2005), vol. 35, no. 11, pp [11] Veltman, J.A.; Gaillard, A.W.K.: Physiological indices of workload in a simulated flight task. In: Biological Psychology (1996), vol. 42, no. 3, pp [12] Vincent, A.; Craik, F.I.M.; Furedy, J.J.: Relations among memory performance, mental workload and cardiovascular responses. In: International Journal of Psychophysiology (1996), vol. 23, no. 3, pp [13] Scheirer, J.; Fernandez, R.; Klein, J.; Picard, R.W.: Frustrating the user on purpose: a step toward building an affective computer. In: Interacting with Computers (2002), vol. 14, no. 2, pp [14] Hazlett, R.L.; Benedek, J.: Measuring emotional valence to understand the user s experience of software. In: International Journal of Human-Computer Studies (2007), vol. 65, no. 4, pp

6