Traffic information potential and necessary penetration rates

Transcription

1 390 Extended Floating Car Data Traffic information potential and necessary Susanne Breitenberger, Bernhard Grüber, Martina Neuherz and Ronald Kates, BMW Group Demands made on the quality of traffic information are increasing due to high road network utilisation. The successful operation of traffic information systems requires acquisition of up-to-date and reliable data. This prerequisite is not satisfactorily met throughout wide areas of the traffic network. Often, road traffic incidents are detected too late and are poorly localised in space and time. The need to ensure the quality of traffic telematics services has thus led BMW Research to search for new methods of improving networkwide traffic data acquisition. Efforts have focussed on the floating car approach, in which suitably equipped probe vehicles serve as mobile sensors. This article considers the potential of vehicle-generated traffic data acquisition for the generation of traffic information and local hazard warnings. A method for estimating the required floating car on the basis of traffic volume and arrival probabilities is presented. Finally, three scenarios for extending the use of vehicle-generated traffic data acquisition for the time horizon up to 2015 are developed as an illustration. Introduction Drivers are able to access a multitude of information sources in the vehicle: route guidance with current traffic congestion messages, weather information and local hazard warnings will become standard in high-end vehicle classes. The quality of these services requires up-to-date, comprehensive, and reliable traffic and weather data. At present, these requirements are not met satisfactorily throughout wide areas of the traffic network. Traffic data are obtained mainly from stationary detectors, whose coverage is inadequate and uneven. The traffic situation between these detectors remains highly uncertain. As a consequence, traffic incidents are detected too late and are poorly localized in time and space. Moreover, few outside of the motorway network are equipped with data acquisition and recording facilities. The need to ensure the quality of traffic telematics services has thus led BMW Research to search for new methods of improving network-wide traffic data acquisition to ensure the quality of telematics service products. Efforts have focussed on the floating car approach, in which suitably equipped probe vehicles serve as mobile sensors. This article addresses the potential of vehicle-based traffic data acquisition for generating traffic information and local hazard warnings. It presents a method for estimating the required floating car on the basis of traffic volume and arrival probabilities. The method will be used to derive required fleet penetration for mobile data acquisition on motorways, state and urban areas. Finally, three scenarios for extending the use of vehiclegenerated traffic data acquisition within a time horizon up to 2015 will be developed as an illustration. Whereas the business trend scenario reveals developments anticipated by trend-based extrapolation, the stable growth and optimistic growth scenarios assume a clear intention to co-operate on the part of the proponents as well as targeted implementation of the floating car approach. Vehicles anonymously collect traffic data during operation Since 1999, the BMW Group has been using probe vehicles, known as floating cars, as mobile sensors in order to collect site-independent, up-to-date traffic data from the entire roadway network 1, 2, 3. Participating vehicles not only supply up-to-date traffic data, but also benefit from the information that is acquired, combined, and processed in the system. Each floating car is equipped with a GPS receiver for tracking its position and generates section speed estimates on the basis of its measurements. Vehicle data relevant to traffic flow are aggregated and transmitted anonymously to a traffic data centre by a mobile telephone module. The data aggregation algorithm in the vehicle was defined by Mannesmann Autocom and TEGARON in the mid-90s (GATS - Global Automotive Telematics Standard). At present, GATS-FCD messages in Germany are received and processed by the DDG (Gesellschaft für Verkehrsdaten mbh, Bonn/ Germany). On the basis of this data, the DDG estimates the current traffic state on specified sections of the network. At present, approximately 40,000 active FCD vehicles (= 0.09% of total passenger cars) are underway on the roadway network in Germany. The DDG also maintains a private, stationary traffic data acquisition system (SES) for Germany s motorway network. At present, traffic situation information is primarily available from the motorway network. Demand-based section information (ie travel time on a section of road) and localised hazard warnings are not feasible due to inadequate coverage particularly in the lower-level network. An increasing number of vehicles are being equipped with telematics applications such as automatic emergency calling, ide breakdown assistance, or dynamic traffic information services. As the electronic components integrated into the vehicle for this purpose can also be used to acquire FCD, the number of potentially useable floating cars will also increase in the future. Traffic situation recording and local hazard warning with XFCD In order to improve traffic state reconstruction and measure weather conditions, BMW has refined the FCD approach to

2 391 acquire additional vehicle-generated traffic data by a system known as Extended Floating Car Data (XFCD). Current data generated in the electronic control units and sub-systems and present on the vehicle data buses are read out and processed into XFCD 4, 5. The algorithms for evaluating and further processing the relevant data can be integrated into the vehicle s existing telematics platform. Using raw data sources such as switching states of the low and main beam headlights and fog lights, ABS, ASC, the external thermometer, air conditioning system, navigation system, brakes, rain sensor, windscreen wipers and hazard warning flashers, etc. together with the current speed values, the following event and situation data can be generated by XFCD as compared with FCD: XFCD message contents Traffic state Entering traffic jam Exiting traffic jam Jam travel time / classification (congested, stop-and-go, queued) Precipitation, aquaplaning Slippery roadway, ice Impeded visibility, fog FCD message contents Travel times The following services can be supported or improved by XFCD message contents: XFCD-supported services Traffic information services (network-wide support) Dynamic route guidance Local hazard warning (queue tail/length estimation, slippery roadways, aquaplaning, fog,...) Road weather information In the primary roadway network, XFCD could contribute towards supplementing the data generated from stationary detectors to achieve more accurate, up-to-the-minute traffic state reconstruction and improve localisation (entrance and exit position) and travel time estimation for traffic jams. In the secondary network, it is anticipated that XFCD would primarily serve to detect local incidents. XFCD due to reasons of cost-effectiveness XFCD makes use of mobile communication interfaces to transmit traffic-based events (eg traffic jam approaches and exits) via SMS. XFCD additionally offers the option of feedback channel referencing, ie upon entering congestion, a vehicle that has already been informed via TMC (Traffic Message Channel), verifies that this information is correct and hence no longer needs to report the traffic jam. This process supports XFCD s cost efficiency. The potential saving rises as the penetration rate increases (Figure 1). Although this saving may be affected by use of the packetoriented GPRS a data transmission standard, XFCD will retain its advantage over FCD. The reason being that mobile data communication with GPRS will no longer be charged according to on-line time, but according to transfer volume (up to 115 kbit/s): XFCD messages are often shorter than the travel times generated by FCD, which are only transmitted as a sequence of positions with time stamps. dissipation of congestion. In urban areas, existing, stationary on-line traffic data acquisition is frequently incomplete and, to a certain extent, subject to quite high failure and error rates. XFCD could contribute towards supplementing existing recording systems. To achieve a sufficiently high traffic situation detection quality within the entire road network using XFCD, a certain percentage of the total number of vehicles on the road need to be XFCD-equipped. Procedures for determining the necessary penetration rate depending on the information quality to be achieved will be presented in the following. Determination of XFCD via arrival probabilities The quality of the information that is generated depends on the following factors: Detection rate - network-wide completeness Detection delay Reliability - functional capability of information acquisition and message transmission Different statistical approaches for estimating the optimal penetration rate specify update times of 5 to 10 minutes and calculate an arrival probability for one or more XFCD vehicles in case of an incident for this period (see results of the LISB, RHAPIT, VERDI projects). Keeping costs in mind, a detection period of ( t = 10 minutes has been selected in this study; this value yields a satisfactory precision for assessing traffic attributes. The need to be increased if higher statistical certainty is required. The method makes the simplifying assumption that the proportion of XFCD vehicles in traffic volume throughout the roadway network is equal to their percentage within the total number of vehicles. This implies that many XFCD vehicles can be found wherever there are many other vehicles. HUBER 5 determines the arrival probability P [%] of X [number of vehicles] XFCD vehicles at an incident location (arbitrary cross-section point ) depending on the traffic volume and detection time ( t (update period): P( t, x) = e µ x µ x! as µ = q XFCD t / t x = the number of XFCD vehicles anticipated within t If at least one XFCD vehicle is to arrive within a time interval ( t, the probability of the arrival of one or more vehicles can, according to HUBER, be calculated from the sum of all probabilities of arrival P(1) to P(n) minus the probability that no vehicle will arrive P(0). This results in the following: P( x 1) = 1 P(0) = 1 e µ x µ x! For the arrival of at least one XFCD vehicle within at most XFCD supports conventional traffic message management Feedback channel referencing not only enables the information calculated on-board to be compared with the incidents reported by radio via TMC, but also helps to check whether the messages being broadcasted are sufficiently current. If there is a discrepancy, information can be transmitted to the traffic data centre, leading to correction of the messages, eg Figure 1: Message quantity per vehicle on substitution of FCD by XFCD

3 392 Extended Floating Car Data Figure 2: Arrival probability for a XFCD vehicle at an incident location within 10 minutes Table 2: Average daily traffic (DTV) and passenger car mileages according to road categories; nonurban from , urban from , 8 Table 1: XFCD penetration rates depending on the traffic volume with a 10 minute detection time XFCD veh. t [min] q XFCD (P=95%) 250 t = 10 minutes, ie x = 1, the above formula results in the graph presented in Figure 2 for the arrival probability P [%] depending on the relative traffic volume q XFCD [XFCD veh/h]: Figure 2 shows the relative traffic volume q XFCD at which the arrival of a XFCD vehicle can be anticipated within 10 minutes with a 95% arrival probability: q XFCD = 18. Formulated in general terms, the relative traffic volume q XFCD can be calculated from the product of the traffic volume q and the penetration rate D XFCD [%]: q XFCD The necessary penetration rate for the 95% arrival probability within 10 minutes for a traffic volume of q = 2000 is therefore D XFCD D[%] q = [%] = = 0.9% 2000[ veh / h] The probability that several XFCD vehicles will arrive is based on a Poisson distribution. With stationary traffic flow q, the probability of occurrence for three vehicles can therefore be calculated according to the following condition: Px ( 3) = 1 P( 0) P( 1) P( 2) If a probability of 95% is required for the arrival of three vehicles within a period of 10 minutes, this results in a required, relative traffic volume of q XFCD = 38 [XFCD veh/h]. Penetration rates D XFCD [%] with a 10 minute detection time. Table Penetration rates depending on traffic volume The quality of the services supported by XFCD depends on the speed with which a defined number of vehicles in the flow of traffic reports an event. In principle, the detection time decreases as the traffic volume increases. Table 1 shows the calculation results for the XFCD penetration rate D XFCD [%] for the arrival of one and three XFCD vehicle(s) at an arbitrary cross-section within 10 minutes with a 95% arrival probability in the case of different traffic volumes q. As can be seen from Table 1, mean passenger car traffic volumes of 1000 passenger cars/h and direction require penetration rates of 3.8%, in order to enable the reliable detection of an incident by three XFCD vehicles within 10 min. The necessary are roughly halved in the case of a detection time of 20 min. At the start of implementation, however, lower penetration rates, e.g. 1% to 2%, are usual. These only lead to the desired, high-quality information in the case of very high traffic densities (see Table 1). However, such high traffic densities are not present at all times of day or everywhere throughout the road network; ie low would imply either long detection times, only certain periods of time (eg commuter traffic) or only certain network sections (eg urban areas). Traffic volume and mileages To enable statements regarding in Germany to be made, the frequency with which traffic volumes occur within the road network must first be analysed. Studies conducted by the Bundesanstalt für Straßenwesen (Federal Office of Roads), such as the 2000 road traffic survey (SVZ) 6, contain historic traffic volumes in form of DTV values (mean, daily traffic for both directions [veh/d]). In the year 2000, the mean DTV value for motorways in Germany was 47,800 veh/d. Individual motorway sections revealed extreme values with more than 120,000 veh/d. In 2000, the average daily traffic volumes on Federal was 9270 veh/d and 3920 veh/d 6 on country, whereby traffic volumes fluctuate severely depending on the day of the week and time of day (Figure 3). As a mean value over 24 hours does not enable statisfactory statements to be made regarding instantaneous traffic volume during day time, the main traffic time will be analysed in the following. This involves the peak hourly volumes (q Peak, 6-12% of DTV on average) and the mean volume over 20 hours of the day (q d20 ). In combination with the penetration percentages determined in Table 1, the necessary floating car can be derived from the peak hourly traffic volumes (q Peak ) and mean daily volumes (q d20 ) depicted in Table 2 and applied to the road network sections (Figure 4). (Conversion to hourly traffic volumes on the basis of peak hour factors and daily mean values). Table 2 Passenger cars 2000 Km covered [million veh. km] Network length [km] Federal motorways Federal Country District (old federal states only) Main urban 163,010 93,113 82,560 35, ,256 11,613 31,879 65,414 55,400 45,237 DTV [passenger cars/d] 38, Peak hourly values [% of DTV per direction] 1 Mean traffic volume q Peak [passenger cars/h and direction] Mean daily traffic volume q d20 [passenger cars/h and direction] 8.3% (determined from SVZ 2000 [6]) 9.2% (determined from SVZ 2000 [6]) 9.5% (determined from SVZ 2000 [6]) 10% (estimated) Penetration rates on Federal motorways With a detection time of 10 minutes and a peak traffic volume q Peak of 1593 passenger cars/h and direction (see Table 2), on 50% of the Federal motorway network an incident can be detected with great precision and low delay with a XFCD 10% (estimated) Determination of peak hourly percentages q as a weekly average for all survey points according to [7] based on daily frequency curves (types A G, Fig. 2) specific to the section and day of the week averaged throughout all DTV values on the various road categories

4 393 penetration rate of just 2.4% of the vehicle population (95% arrival probability ensured by the arrival of three XFCD vehicles). During off-peak hours, q d20, at a mean traffic volume of 960 passenger cars/h and direction, the same detection quality can be achieved at a penetration rate of 3.9%. Network coverage of around 80% can therefore be achieved on Federal motorways (Figure 4). Throughout wide sections of the highly-frequented Federal motorway network, particularly in metropolitan areas, information is available from stationary detectors or traffic jam callers. As a result, three vehicles are not necessarily required to detect an incident in this case. In many cases, a message from one XFCD vehicle is sufficient to verify a traffic jam. If recording by one XFCD vehicle is assumed, traffic incidents can be satisfactorily detected on 90% of the Federal motorway network at a penetration rate of 2.5%, which can be achieved within a few years (q=720, 95% arrival probability ensured by one XFCD vehicle, Figure 4). Federal In the case of non-urban, preference will be given to detailed treatment of Federal, because these supplement Federal motorways as the backbone network and are subject to more traffic jams than district and country. During peak traffic hours q Peak, mean volumes of approx. 367 veh/h and direction occur on Federal throughout Germany. This means that reliable XFCD data would be available from the Federal road network at an XFCD penetration rate exceeding about 10% (detection time of 10 min., reported by three vehicles). As can be seen from Figure 5, around 65% of the Federal road network could be reliably detected with q Peak 367 veh/h and 10% XFCD penetration. During the off-peak traffic time q d20 with a mean traffic volume of 200 veh/h and direction, the same detection quality can only be achieved at a penetration rate of some 19%, whereby 80% of the Federal road network are then detected at the specified quality (Figure 5). Urban Within urban centres, the analysis focuses on occurrences on main, which have a mean volume of 373 passenger cars/hour during the main traffic times (Table 2, 8 ). Like the process on Federal motorways and Federal, a XFCD penetration rate of 10.2% with reference to the DTV Veh value is required (detection time 10 min, reported three times) for the main traffic time q Peak. In off-peak traffic, q d20, 20.3% fleet penetration becomes necessary in urban centres. XFCD penetration using the Munich conurbation as an example With some 1.3 million inhabitants, the state capital, Munich, is a significant European traffic node point. With a traffic volume of up to 150,000 veh/24 hrs., the centre ring road plays a dominant role in the main road network. Using Munich as an example, the XFCD required to achieve satisfactory traffic situation detection in high-volume urban areas are explained in the following. Figure 6 shows the traffic volume in Munich s main road network over a daily average of 20 hours q d20 (DTV values ), differentiating between the primary and secondary network b and the centre ring road. Whereas average traffic volumes of 814 passenger cars/h and direction occur in the primary network (excluding Federal motorways) during offpeak traffic hours (q d20 ), mean traffic volumes of 425 passenger cars/h and direction occur in the secondary network. Over its length of around 19 km, the centre ring road accommodates some 2260 passenger cars/h and direction during the hours of off-peak traffic q d20, with significantly higher values during peak times. At a penetration rate of just 9%, traffic occurrences on 50% of the secondary network, which serves primarily to accommodate internal traffic and to distribute origin/destination traffic, are detected according to the required quality characteristics. If only the primary network, which is particularly susceptible to incidents, is analysed, a XFCD fleet rate of just 5% is sufficient to reliably detect an event with three XFCD Figure 3: Various daily frequency characteristics for different traffic volumes q on motorways (Tuesday to Thursday), Road types A - G in accordance with 6 Figure 4: XFCD penetration rates on Federal motorways, differentiated according to different traffic volumes (passenger cars/h at peak hours totalled) and recording qualities Figure 5: XFCD penetration rates on Federal, differentiated according to different traffic volumes (passenger cars/h at peak hours totalled) Fig. 6: XFCD on the main road network of state capital Munich (passenger cars/h at peak hours totalled)

5 394 Extended Floating Car Data Figure 7: Necessary XFCD for the north-west sector of the city of Munich 10 Figure 8: Summary of necessary XFCD for different road network categories vehicles on two-thirds of the network within 10 min. On the centre ring road, this can be achieved at a penetration rate of approx. 2% during off-peak traffic hours (q d20 ). In summary, it can be stated that a XFCD-capable vehicle fleet amounting to 7.3% of the total number of passenger cars is sufficient to detect traffic conditions on over 80% of the main road network (primary network without Federal motorways, secondary network and centre ring road) within the state capital of Munich. Effective incident detection on sections of road that are particularly susceptible to traffic jams can be achieved with far lower. Summary of necessary When analysing the traffic volumes listed in Table 2, focus must be placed primarily on the road types of greatest importance to driver assistance services, ie Federal motorways, Federal and the main urban road network, on which almost 80% of total mileage take place. The minimum penetration rates required for these network segments are depicted in summarised form in Figure 8. Problem of mean analyses In determining the above mentioned, it was not possible to take the following factors into consideration: a) The fact that not all of the 45 million passenger cars registered in Germany are on the road to the same extent. b) The additional increase in the detection rate and meaningfulness of each XFCD message due to: Centre-based fusion of the FCD / XFCD informtion with data from other recording facilities Messages from several FCD / XFCD vehicles on the same incident or hazard point Distribution of traffic volume (uniform distribution or high-mileage drivers) Detection sensitivity and reliability of XFCD, c) The phenomenon that the vehicles are not equally distributed throughout the road network, but are frequently accumulated wherever traffic jams occur. With reference to traffic jam events, uniform distribution is an unfavourable assumption, because the 95% probability of a XFCD vehicle s detecting a traffic jam within 10 minutes is frequently achieved at far lower than 2%. After all, the majority of traffic jams occur on road sections with a high traffic volume, ie where traffic volumes of 3000 vehicles per hour and over can be detected over two lanes. For q= 3000 veh/h, an event is detected by one XFCD vehicle within 10 minutes with a traffic flow penetration rate as low as 0.06% XFCD is not only a promising method for detecting traffic flow incidents; detection of local, weather-related situations can also be improved with XFCD. Particularly in the lower-level network, which is equipped with hardly any recording facilities, XFCD may also contribute towards the detection of hazards and the generation of warnings in the case of fog, slippery conditions, etc. even in the event of a penetration rate which is lower than average. How quickly can of 2%, 4% and 10% be achieved? Some 78,000 FCD-capable BMW vehicles are currently underway on Germany s. XFCD test vehicles are undergoing trials at BMW. Based on three different future scenarios, conceivable trends for XFCD fleet penetration will be revealed in the following: The business trend scenario assumes that automobile manufacturers are establishing and constantly expanding telematics-capable vehicle fleets with localisation functions. The new, significantly more high-performance mobile telephone and radio technologies (GPRS, UMTS, DAB, DVB-T) are extending the spectrum of future telematics applications. The cost of multimedia systems with a mobile telephone interface are falling, leading to the creation of a mass market. In Germany approximately five million vehicles will be equipped with a navigation system this year; according to the ADAC (German Automobile Club), growth over the next few years will lie between 15-19% / year 9. A trend towards consistent and personalised applications, in which the Internet will be networked with the navigation system and portable handsets, is emerging. The technological equipment available in the vehicle for dynamic navigation will be used to introduce XFCD. Based on the market penetration anticipated for navigation systems, it is assumed that around 15% of mid-size class, upper mid-size class and luxury class vehicles will transmit XFCD by With 2,160,000 passenger cars c, this corresponds to a penetration rate of 4.3% of total passenger cars (forecast: 49.8 million passenger cars in 2015 according to BVWP). The stable growth scenario assumes that the German automobile manufacturers (OEM d ) will agree, in the form of a memorandum of understanding, to push forward the creation of XFCD-capable vehicle fleets. The improvement of data acquisition is one of the central tasks facing market leaders in the automobile sector. Premium quality at the telematics service level, particularly with regard to dynamic route guidance, should be achieved via co-operation with service providers. The OEMs are creating a multi-disciplinary, pan- European technology platform, thereby enabling the achievement of economies of scale. It is assumed that onethird of all mid-size, upper mid-size and luxury class vehicles will have XFCD functions by With 4,975,000 passen-

6 395 ger cars, this corresponds to a penetration rate of around 10% of total passenger cars. In addition to the assumed in the stable growth scenario, the optimistic growth scenario assumes that passenger cars XFCD capability will be promoted with public support. In this scenario, the government will support XFCD capability, as XFCD actively contributes towards increasing road traffic safety (detection of traffic jam ends, hazard warning, savings over and above conventional detection). Efforts will be made to integrate data from public and private data facilities. Thanks to public support, the penetration rate can be extended to 20% of total passenger cars. At approximately ten million passenger cars, this penetration corresponds to the anticipated saturation rate. This penetration level will not only enable significantly better, extensive data acquisition in Germany, but will also offer a significant basis for improving traffic information in Europe, particularly in the light of EU expansion. The scenarios presented above are designed to illustrate conditions that have not yet been achieved and promote discussion of their expected consequences. The penetration rates and technological trends mentioned in this contribution are intended to intensify consideration of XFCD and to quantify and extend existing approaches. Conclusion Based on the current state of affairs, the desired quality for various services such as traffic information, dynamic routing, road weather and hazard warning can only be achieved using data content from XFCD. In particular, local hazard warning services can be provided satisfactorily at a penetration rate of just 1% on motorways via traffic tailback detection. Penetration rates of at least 2% are required for more extensive quality improvements. These would provide good incident detection on the federal motorway network at peak traffic times. Current traffic and road situation information can be generated on 80% of the motorway network at penetration rates of around 4%. A penetration rate of 7-10% is required throughout Germany for high information quality in urban areas and on Federal. It must be taken into consideration that the determined may achieve even greater efficiency, as vehicles with 50% higher annual mileages will initially be (X)FCD-capable. In simplified terms a 2% penetration rate (~ 1 million passenger cars) e of premium vehicles is equivalent to a 3% penetration among all vehicles. However, the related information potential is not the only aspect in favour of rapid implementation of the XFCD method, but also its cost-effectiveness, even if information cannot initially be generated with full coverage. One further advantage is revealed at European level. XFCD will help, in the short term, to expand data acquisition on major, particularly in the new EU member states, and to create a corresponding quality of service. Co-operation between the automobile manufacturers and the telecommunications/it industry is required to achieve significant XFCD fleets. High-quality traffic information can only be generated at low cost via co-operation with strong market players and joint venture partners. The following measures may additionally support the breakthrough of XFCD: Increasing perception of the potential of XFCD Decreasing price of on-board telematics hardware, particularly amongst OEMs Multi-OEM, open and mobile telephone provider-independent technology platforms Government promotion of data acquisition with XFCD However, the prerequisite of accelerated vehicle penetration rates is not only OEM commitment but also increased co-operation with customers. In addition to (X)FCD capability at the vehicle level, the customer s willingness to participate is particularly vital. More extensive marketing measures are required in this regard to raise the implementation rates. References [1] HUBER, W.; LÄDKE, M.; OGGER, R.: Extended floating car data for the acquisition of traffic information. ity/ 8_2_mobilitaet_verkehr/pdf/XFCD_englisch.pdf [2] BREITENBERGER, S. (1997): Assessment of current traffic situations from vehicle-generated data, Degree dissertation. Munich [3] BMW AG: Detection of traffic situations on Federal motorways. Patent document PA DE. [4] HAUSCHILD, M. (2003): Real time processing of vehicle bus data for traffic situation detection in the public road traffic network, Degree dissertation. Munich [5] HUBER, W. (2001): Vehicle-generated data for acquiring traffic information. Dissertation. TU Munich - Faculty of traffic technology and traffic planning, Munich [6] BUNDESANSTALT FÜR STRASSENWESEN (Federal Office of Roads): Road traffic survey (SVZ) 2000, Bergisch-Gladbach, February 2003 [7] HAUTZINGER, H., HEIDEMANN, D. KRÄMER, B.: Mileage survey Short report for research project 8902 of the Federal Office of Roads, Bergisch Gladbach, July 1992 [8] PALM, I., REGNIET, G., SCHMIDT, G. (1996): Detection of annual mileage of passenger cars and commercial vehicles 1993 on all in Germany, FE No /94, Aachen March 1996 [9] ADAC (2002): Motor vehicle navigation market growth. ADAC estimate and forecast to Data based on internal updates by ADAC. [10] State capital Munich, Planning department (2003): Average daily traffic - Values from the Munich urban area, status: Jan Munich Footnotes a GPRS (General Packet Radio Service). b Primary network: main with national and regional link function (without Federal motorways in this case, but with B304, B13-N, B11-S, B2-W, centre ring road, various radials and tangentials); Secondary network: all other main with primarily local link function according to draft traffic development plan 2001 c Extrapolation of 30% mid-size class, upper mid-size class and luxury class within total passenger cars on d OEM - Original Equipment Manufacturer e Vehicle equipment for FCD refers to high-quality vehicles, principally vehicles with cubic capacities of 2l and over. Whilst these only make up approx 20% of registrations, they have a higher mean annual mileage (in 1990, this was around 50% higher than the mean 7 ). Authors: Susanne Breitenberger Traffic Technology, BMW Group, D Munich susanne at: susanne. breitenberger@ bmw.de Tel:+49 (0) Bernhard Grüber, BMW Group Martina Neuherz, on behalf of BMW Group Ronald Kates, on behalf of BMW Group