International Transportation (67) 1 | 2015 48 SCIENCE & RESEARCH Pedestrians Navigation Navigation of blind and visually impaired people Smartphone-assisted navigation and crossing of signalized intersections using Car2x Communication technologies Smartphone-assisted navigation, intra-urban mobility, handicapped persons The mobility of the blind and visually impaired is associated with many barriers and risks. To secure crossings, signalized intersections are partially equipped with acoustic or tactile indicators. However, environmental conditions might interfere with the acoustic identiication of the green time. Furthermore, information such as intersection topology, bicycle traic or the curb structure is not accessible to visually impaired road users. Therefore, most trips are limited to trained routes. Within the research project InMoBS (intra-urban mobility support for the blind and visually impaired) a prototype of a route planning and navigation system has been developed and evaluated in an exploratory manner. Authors: Stefen Axer, Jörg Belz, Kathrin Leske, Bernhard Friedrich, Tobias Hesse, Mark Vollrath T o identify technical and functional requirements and to meet the user needs, a requirement analysis was carried out that considered the preand on-trip functionality of the assistance system. In this process diferent techniques like in-depth interviews and video documentation of typical street crossing situations provided important implications for the engineering of a web-based route planning and mobile navigation assistance system. The collected qualitative results have been veriied as efectively as possible by an online survey of 719 visually impaired and blind users. Extracts of the most important functional and technical aspects are qualitatively summarized in the following. The functionality of a web-based pre-trip route planning tool for the given user group is basically similar to already established and well-known online routing services used by people without any visual impairment. However, to ensure content interaction that is compatible with the use of a keyboard and braille terminal, speciic accessibility guidelines need to be applied [1]. Moreover, standard routing services mainly allow a shortest-path routing [2] that minimizes the total costs (i.e. geometric route distance) between two nodes of a road network. Typical underlying road network models are primarily designed for turn-byturn vehicle navigation. In this case network topology and geometry are modelled in a very simpliied manner [3]. The existence of sidewalks needed for pedestrian navigation is usually logically generalized information linked to the modeled intraurban street. Compared to this standard approach, a routing system for blind and visually impaired users requires a much more detailed network model [4] that allows the explicit modelling and attribution of sidewalks and intersections with a high spatial resolution. As a minimum, the routing calculation needs to take parameters like accessibility and route distance into account, and the accessibility level should cover aspects like the availability of acoustic or tactile indicators at signalized intersections and the existence of pedestrian crossings. In addition, obstacles and points of interest tagged during the user’s daily mobility should be incorporated in the route planning, which will allow not only the exclusion of dangerous walkways, but also the planning of the route to cover supportive waypoints. Last but not least, users desire a storing functionality that simpliies the use of regularly used routes. When talking about the needs of on-trip assistance, blind and visually impaired test persons often reported feeling uncertain in heavily motorized, complex intra-urban intersection scenarios, especially on intersections with asymmetric geometry. Traic noise masks potential orientation signals at intersections. Missing information about curbs or mostly lowered curbs without tactile indicators makes it very diicult to securely identify the transition line between the sidewalk and the street. Therefore complementary and seamless mobile assistance is needed to compensate for these factors and provide the users with important situational information. Apart from the abovementioned aspects, the availability of a continuous and reliable direction indicator, the current position (street name and number) and nearby POIs during the whole navigation are some of the most important requirements users brought up during the surveys. This multifaceted list of information and functionality, reported by the users, clearly emphasizes the need for a sophisticated prioritization of the information provided in order to avoid a mental overload during the navigation process. Basic system design Based on these requirements a system architecture composed of smartphone navigation “InMoBS mobile”, web-based route planning “InMoBS home”, server-sided routing and Wi-Fi-equipped traic lights has been developed (figure 1). Within this architecture, a Component Server Platform (CSP) acts as a central information provider that supplies calculated routes on a highprecision digital map. The digital walkway network has been generated by processing standardized land register data from the city of Braunschweig. Three Wi-Fi-equipped traic lights, as a part of the AIM-infrastructure [5] allowed for a unidirectional communication of real traic signal states and timing, queried with a frequency of 1 Hz. For International Transportation (67) 1 | 2015 49 Pedestrians Navigation SCIENCE & RESEARCH communication purposes, a standardized protocol stack [6, 7], originally designed for 802.11 p vehicular communication, has been adapted to consumer Wi-Fi (802.11 b/ g) enabled devices. To obtain more reliable and accurate position information in urban areas, a Google Nexus 5 smartphone was connected via Bluetooth to a small external diferential GPS (Global Positioning System), which could be fastened on the upper arm of a test participant. In urban areas, the positioning system (Alberding A07) reached an accuracy of 1.3 - 2.2 m (1 σ ), depending on the surrounding environment and satellite constellation. Smartphone application “InMoBS Mobile” is the digital companion of the blind or visually impaired and guides the user safely to the chosen destination. In the scope of an iterative development process, blind and visually impaired participants were invited to test the app at various development stages. Their feedback was used to design the app’s human machine interface (HMI). Several approaches were tested and led to the inal interface, which is based on a simple layout scheme of ive tiles used throughout the application (igure 2). The app’s main menu ofers the user the option to start a navigation process based on current position and desired destination (point of interest, street and house number). Furthermore, the user can select a previously deined route, stored by using the web-based route planner “InMoBS home”. The route is calculated on the server, transmitted to the cell phone via mobile internet connection and handed over to the app’s navigation process (see igure 1). The downloaded route allows the app to execute a sophisticated navigation algorithm that keeps the users on track, provides information about the surroundings and supports them in street-crossing situations. Route guidance The computed route represents a safe path to the destination. Deviations from this path may result in dangerous situations such as leaving the sidewalk and entering the vehicle lane. Accordingly, helping the user to remain on track is important. At any time during the navigation, “InMoBS mobile” helps the user to align the direction of movement with the help of vibration impulses, emitted by the smartphone, similar to the approach suggested by Pielot et al [8]. The impulse frequency indicates the extent of the deviation from the correct heading. The higher the deviation, the higher the impulse frequency. The actual heading is determined by using the smartphone’s magnetic compass or internal GPS, whereas the desired heading can be obtained from combining the routing data with the user’s current position. In addition to this Figure 2: Test participant (left) and InMoBS Mobile user interface Component Server Platform (CSP) InMoBS home Ext. D-GPS (Bluetooth) Simpli fied route planning Naviga i on Complex route planning Traffic light (Wi-Fi) WLAN Access Point Signal phase and i ming Intersec i on Topology InMoBS mobile (Wi-Fi / GSM,UMTS / Bluetooth) Map visualiza i on User management Crossing assistance Posi i oning Enhancing Posi i oning Accuracy Routing Server Web Server User storage Server Nav-data Server Map Server Hos i ng Web Map Service R oui ng algorithms Planned Routes GNSS corr eci on informa i on POI / Obstacles Network Graph Stored POI and Obstacles 3 1 2 4 1 2 3 4 Requests from InMoBS home and InMoBS mobile, provides calculated routes (GeoJson) Provides unidir eci onal signal phase and i ming over local Wi-Fi (802.11 bg) Corr eci on informa i on, provides enhanced posi i on informa i on (NMEA) Figure 1: Basic system architecture and communication interfaces International Transportation (67) 1 | 2015 50 SCIENCE & RESEARCH Pedestrians Navigation instant feedback, the app provides turning information by automatically notifying the user of upcoming curves using speech output, hence allowing the user to prepare for a direction change in advance. Information about surroundings Navigation in unfamiliar areas is a challenging task for the blind and visually impaired. Orientation is diicult, and unknown obstacles are a dangerous threat. To support the user in such situations the app supplies prestored information about important or interesting points and obstacles. The location and other information about these socalled ‘points of interest’ (POI) are part of the digital map. Public POIs, like museums, supermarkets and bus stops, are available to all users, while private POIs are only related to a single user. The app provides functionality to add private POIs and annotate them using speech input. When the navigation is started, the smartphone receives data about public and private POIs along the route. During navigation, information about a POI is provided automatically when the user approaches the POI’s location. Furthermore, the user can request information about nearby POIs and POIs that are located along the remaining route. The speech output includes a description as well as the distance and bearing of the POI relative to the user’s position and heading. Crossing support When the user approaches a crossing, crossing properties are announced automatically, for instance the type of crossing (signalized or unsignalized), the crossing distance and the availability of an acoustic or tactile indicator. The app indicates how to reach the waiting area and warns the user of crossing bicycle lanes. Acoustic information output can also be triggered manually using the accessible app interface. While standing in front of the street crossing, the haptic impulses allow the users to align their direction of movement. The Wi-Fi-equipped signalized intersections are broadcasting SPaT and TOPO messages (see figure 1). Based on these messages and the route information, the current signal phase of the crossing can be determined by the app. When the user has safely reached the waiting area and is ready to cross the street, the app monitors the crossing’s signal state and emits an acoustic green light indicator as soon as the traic light switches to green. No signal is given in case the user enters the waiting area within a running green time interval. In such cases, the user needs to wait for the next cycle to ensure that the maximum green time interval is available. Entire system evaluation Sample The entire system was evaluated between 16- September and 23 October 2014 by a group of blind and visually impaired test persons. The realization efort for the tests and the extent of support that this speciic group of users required during the test phase were relatively high. This resulted in a comparatively small sample (N = 8) and an exploratory, qualitative system assessment. All participants (four women and four men) lived in Braunschweig. Contact was established by the Deutscher Blinden- und Sehbehindertenverband e.V., the German Federation of the Blind and Partially Sighted. All subjects had previously completed trainings in orientation and mobility. Participation was voluntary. Five of the participants were blind (63 %) and three were visually impaired (38 %). The mean age was M = 60.0 years (Min = 44, Med = 63.0, Max = 67, SD = 8.1). 50 % (n = 4) of the participants had basic knowledge of working with tactile cards and smartphones and 63 % (n = 5) had experience with navigation apps or navigation devices. All users had basic knowledge of working with computers or laptops, as well as with tools for computers or laptops for blind and visually impaired and the Internet. Procedure The blind and visually impaired participants walked a 1,900 m long section of the ring road of Braunschweig twice (figure 3). Each subject walked the route once with the support of the navigation system (test setting) and once without navigation system (comparison setting). The participants experienced a walk in the direction of traic on one date and a walk against the low of traic on the other date. The combination of the characteristics of the walking direction (with and against the low of traic) and the use of the navigation app (with and without) resulted in four study settings. The subjects were randomly assigned to these settings. The study design was fully balanced. The route had diferent characteristics, which generally restrict the mobility behavior of blind and visually impaired persons in road traic. There were some intersections with acoustic or tactile signal indicators and some intersections without these features. Loud traic noise, traic islands, bicycle lanes running in parallel to the sidewalk, different curb heights and, at one intersection, crossing tram rails further added to the dificulty of the situations. In all study settings, the test walk started with a general explanation of the study process. Additionally, on the irst date an informed consent as well as the participants’ socio-demographic data were collected. In the study settings with navigation app, the examiner explained the operation of InMoBS mobile and InMoBS home to the participants. Then, the participants were given time for to practice the operation of the navigation device. The computer for the operation of InMoBS home was equipped with tools for the blind and visually impaired (e.g. braille console). InMoBS mobile was installed on a smartphone. Through this, users were given © OpenStreetMap-contributors - Published under CC-BY-SA 2.0 Route start and target, 1 st interview point 2 nd interview point 3 rd interview point InMoBS Wi-Fi equipped intersec i on InMoBS walkway network Figure 3: Overview of the InMoBS test site International Transportation (67) 1 | 2015 51 Pedestrians Navigation Science & ReSeaRch acoustic and tactile information about the route and their own position. In addition, a Bluetooth speaker was used. In the study settings with a navigation app, the users irst enabled the route on InMoBS home. They then walked the route with the assistance of the navigation system. During the walk they added a POI to a position on the route. After completing the route, they were asked to rename this POI in InMoBS home. In the study settings without a navigation app, a tactile map or highcontrast map of the route was handed out to the participants. All participants were interviewed at three route points (see igure 3). The interviews were recorded with a voice recorder. All participants were accompanied by at least one full-sighted research associate and at least one full-sighted student of the Technische Universität Braunschweig along the entire the route. An interview guide was developed for each study setting. The interviews included questions that could be answered on a 5-point scale of response, as well as open-ended questions. The participants were asked about their subjective experience with regard to managing the task of walking the route with and without a navigation app. The participants were also interviewed regarding general aspects of mobility. For example, one question was: “How well could you orient yourself about your position in respect to the entire route? ” Participants were also asked what they thought were good and bad aspects of InMoBS mobile and InMoBS home. Results All participants positively evaluated the concept of the InMoBS system. The features of InMoBS mobile that provided the participants with information at intersections without acoustic or tactile signal indicators met with especial appreciation. The same was true for the announcement of bicycle lanes, as well as the option “Where am I? ”, which provided information about the current position on the route. The information content of the menu provided at crossings and intersections was also positively assessed. This menu ofered information about traic lights and acoustic or tactile signal indicators at the intersections, as well as the topology of the intersection and the names and number of incoming roads. Furthermore, intersection-speciic information about traic islands, tram rails, bus lanes and curbs was ofered in this menu. The study participants especially liked the relatively autonomous operability and the orientation and navigation on the web page ofered by InMoBS home. Almost all of the participants were able to handle the processing of tasks in InMoBS home. On the other hand, participants criticized the message delay caused by partially inaccurate positioning. Furthermore, sometimes the operation of both systems was diicult due to the lack of experience and training with the system. conclusion The participants repeatedly expressed that the navigation app, with reservations regarding the localization problems, would considerably improve the quality of mobility for the blind and visually impaired. Also, the system would be suitable for providing assistance to additional groups, such as fullsighted elderly pedestrians. Thus, the next steps should be to develop the prototype further, including a larger area and the integration of public transport. ■ This work was supported in part by the German Federal Ministry for Economic Afairs and Energy on the basis of a decision by the German Bundestag under the grant numbers 19 P 12001 A-F. REFERENCES [1] World Wide Web Consortium. (2008). Web content accessibility guidelines (WCAG) 2.0. [2] Dijkstra, E. W. (1959). A note on two problems in connection with graphs. Numerische Mathematik, 1(1), 269-271. [3] Lv, W., Liao, W., Wu, D., & Xie, J. (2008, March). A new road network model and its application in a traic information system. In: ICAS 2008. Fourth International Conference on Autonomic and Autonomous Systems (pp. 160-164). IEEE. [4] Wieser, M., Mayerhofer, B., Pressl, B., Hofmann-Wellenhof, B., & Legat, K. (2006). GIS-gestützte Navigation blinder und sehbehinderter Personen. in: Angewandte Geoinformatik 2006 (2006), S. 747 - 756, 18. AGIT-Symposium [5] Schnieder, L., & Lemmer, K. (2012). Anwendungsplattform Intelligente Mobilität - eine Plattform für die verkehrswissenschaftliche Forschung und die Entwicklung intelligenter Mobilitätsdienste. Internationales Verkehrswesen (64), 4, 62-63. [6] ETSI, T. (2011, February). 102 636-5-1 V1.1.1, Intelligent Transport Systems (ITS) Part 5: Transport Protocols, Sub-part 1: Basic Transport Protocol [7] ETSI, T. (2011, June). 102 636-4-1 V1.1.1, Intelligent Transport Systems (ITS) Part 4: Geographical addressing and forwarding for point-topoint and point-to-multipoint communications, Sub-part 1: Media- Independent Functionality [8] Pielot, M., Poppinga, B., Heuten, W., & Boll, S. (2011). A tactile compass for eyes-free pedestrian navigation. In Human-Computer Interaction-INTERACT 2011 (pp. 640-656). Springer Berlin Heidelberg. Tobias hesse, Dr.-Ing. Team Leader Technologies for Assistance and Automation, Institute of Transportation Systems, German Aerospace Center (DLR), Braunschweig (DE) tobias.hesse@dlr.de Jörg Belz, Dipl.-Inf. Mobile Devices for Assistance and Automation, Institute of Transportation Systems, German Aerospace Center (DLR), Braunschweig (DE) joerg.belz@dlr.de Kathrin Leske, Dr. Dipl.-Psych. Research associate, Engineering & Traic Psychology, Institute of Psychology, Technische Universität Braunschweig, Braunschweig (DE) k.leske@tu-braunschweig.de Stefen axer, M.Sc. Research associate, Institute of Transportation and Urban Engineering, Technische Universität Braunschweig, Braunschweig (DE) s.axer@tu-braunschweig.de Mark Vollrath, Univ.-Prof. Dr. Dipl.-Psych. Head of Department, Engineering & Traic Psychology, Institute of Psychology, Technische Universität Braunschweig, Braunschweig (DE) mark.vollrath@tu-braunschweig.de Bernhard Friedrich, Univ.-Prof. Dr.-Ing. Head of Institute, Institute of Transportation and Urban Engineering, Technische Universität Braunschweig, Braunschweig (DE) friedrich@tu-braunschweig.de International Transportation 2/ 2015 Mobility 4.0 - advanced transportation solutions Release Date: 30 Oct 2015 | Submission deadline for articles: 20 Aug 2015 Contact: Eberhard Buhl, Managing Editor, eberhard.buhl@dvvmedia.com