Food

Monitoring Traffic and Emissions by Floating Car Data

Categories
Published
of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Description
WORKING PAPER ITS-WP Monitoring Traffic and Emissions by Floating Car Data By Astrid Gühnemann * Ralf-Peter Schäfer Kai-Uwe Thiessenhusen
Transcript
WORKING PAPER ITS-WP Monitoring Traffic and Emissions by Floating Car Data By Astrid Gühnemann * Ralf-Peter Schäfer Kai-Uwe Thiessenhusen Peter Wagner March 2004 DLR - German Aerospace Centre Institute of Transport Research Rutherfordstr. 2 D Berlin * Visiting Scientist at ITS Sydney Sept. Dec INSTITUTE OF TRANSPORT STUDIES The Australian Key Centre in Transport Management The University of Sydney and Monash University Established under the Australian Research Council s Key Centre Program. NUMBER: TITLE: ABSTRACT: Working Paper ITS-WP Monitoring Traffic and Emissions by Floating Car Data Intelligent traffic management is widely acknowledged as a means to optimise the utilisation of existing infrastructure capacities. A major requirement for intelligent traffic management is the collection of high quality data on traffic conditions in order to generate accurate real-time traffic information. The approach to be described here generates this information by a fleet of taxis equipped with GPS which act as Floating-Car-Data (FCD) provider for a number of metropolitan areas. The first part of this paper describes the methodology of setting up this data base. The information collected enables various applications such as real-time traffic monitoring, time-dynamic routing and fleet management. The second part of the paper proposes a framework for using these data additionally to include environmental effects into intelligent traffic management systems. To this end, a mapping between travel times and traffic flows is proposed. Some challenges related to the computation of emissions from velocity profiles are discussed. Equipped with these ingredients, an environmentally friendly intelligent traffic management might be in reach. KEYWORDS: Floating-Car-Data, Traffic Monitoring, Traffic Management, Emission Modelling AUTHORS: Astrid Gühnemann,* Ralf-Peter Schäfer, Kai-Uwe Thiessenhusen and Peter Wagner. CONTACT: Institute of Transport Studies (Sydney & Monash) The Australian Key Centre in Transport Management, C37 The University of Sydney NSW 2006, Australia Telephone: Facsimile: Internet: DATE: March 2004 *The research for this paper was done Dr Astrid Gühnemann was a Visiting Scientist at ITS, Sydney during September December 2003. 1. Introduction Conventional traffic monitoring systems that use stationary detectors such as counting loops do not produce reliable information about travel times within the network. The reason for this is that especially in urban areas their density is usually fairly low (the city of Berlin has about 200 of such devices), and the disturbances in urban areas is very high, so they cannot produce velocity profiles needed to compute travel times. Therefore, different techniques have been developed to measure travel times directly. Apart from simulation approaches that try to reconstruct the missing information from scarce data, other techniques such as license plate recognition, video imaging, or airborne traffic measurement are under development or even functional. However, in order to cover the traffic of metropolitan areas completely in time and space, these technologies are yet too costly. Therefore, the floating car data (FCD) technique has moved into the focus of current research, and first commercial applications have been developed. In the FCD-approach travel times and routes are determined directly from monitoring vehicles which float with the traffic and whose position is determined by means of navigation technologies such as GPS (Global Positioning System). The pilot operation of commercial FCD based traffic information systems has shown that a sufficiently large number of vehicles have to be observed in order to derive reliable information. BURR, SIMMONS (2002) e.g. estimate these to about 1% of the total fleet. Another obstacle to providing successful commercial services based on FCD are the high communication costs which occur at today s price levels for mobile communication when repeatedly transmitting vehicle positions and speed to a central server. Not least, generally a wider dissemination of intelligent traffic information services might have been hindered due to missing consumer s acceptance, which had not been regarded sufficiently and therefore led to inadequate business models. In order to overcome some of these problems, an alternative FCD technique is described in this paper which presents a solution to the critical mass problem of probe vehicles by using data from taxis. This approach allows for a real-time, high-quality and low cost solution to traffic monitoring especially in urban areas. Furthermore, these data are an excellent basis for data mining analysing the daily variation of travel time on urban roads. Based on this information, applications have been developed such as real-time traffic jam detection and dynamic routing and navigation tools. Under the notion of establishing intelligent traffic management for environmentally sustainable transport solutions, these data are then applied for the development of dynamic emission models as a basis for an environmental monitoring and management of road traffic. 2. Establishing the Taxi-FCD Data Base 2.1 System Architecture The FCD data base in this project is fuelled by data from car fleets where the cars are equipped which GPS (Global Positioning System). The GPS-receiver in the car determines the position, with an error of 10 to 20 m, which is then transmitted to a 1 central server via a special operator s broadcast system or via mobile communication such as GSM. Nowadays an increasing number of commercial vehicles are equipped with such GPS devices. The approach described here utilises data from the fleet management systems of taxi companies. The big advantage of this system is that no additional expenses for on-board equipment of vehicles or for software at the server site of the taxi headquarters are needed. Furthermore, the transfer of data happens by piggybacking them on the communication system needed by the taxi company anyway. Figure 1 shows the general architecture of the system. Thus, the taxis can easily be used as cheap probe vehicles for an extensive floating car data base. Using those commercial vehicles has a further advantage, which is their much higher utilization than private vehicles. In a most recent survey for Germany by IVS ET AL. (2003) the data show that vehicles which are used for commercial purposes travel about twice the distance of privately owned vehicles. A special advantage of using Taxis is their coverage of cities. In order to prove this concept a pilot system running in different German and European cities has been established. Since April 2001, more than 30 million GPS data have been recorded for Berlin already, other applications of this system are running in Vienna, Munich, Nuremberg, Stuttgart and Regensburg. This data provides the basis for further exploitation of traffic information in a high coverage and resolution of time and space as well as for the further development of service applications. Control Room Dispatch Management FCD Data Base Data Processing Service Provider Data Utilization Trunked Radio Channel, approx. 1-4 pos./ min GPRS, SMS, RDS-TMC, HTTP... Figure 1: System Architecture of the Taxi-FCD-System 2 2.2 Estimation of Travel Times From the taxi positioning data a data base on the daily variations of travel times on the street network has been generated. This data-base is subsequently used for the development of a dynamic route planning system. Furthermore, they provide the basis for an in-depth analysis of the traffic fluctuation patterns in the cities to identify the influencing factors for the variation of travel times. The positioning data is transmitted from the taxi vehicles to the dispatch system repeatedly up to four times a minute, with a default rate of one per minute due to bandwidth limitations.. The data consist of the vehicle ID, the timestamp, its geographical position (WGS-84) as well as an information code about the current status of the Taxi (occupied, waiting etc.). Therefore, after transmitting the data from the taxi head quarter to the DLR FCD data base server, these data have to be further processed in order to derive travel times on each link of the network. 1 In a first step the way points from each vehicle for a given time frame (e.g. 30, 60 ) are used to construct trajectories for each vehicle using the vehicle IDs and restricting the analysis to only those data of which the status information is relevant for representing traffic flow information. These trajectories are then matched to a digital road map. GPS positioning error is corrected according to a given confidential interval around the nearest road edge or neglected for bigger positioning errors. In a next step the most likely path between two positions is reconstructed by a routing algorithm based on Dijkstra (1959) and Zhan (1997). The mean velocity is calculated according to the path length and the time difference between two GPS positions. The velocity for each road section is reconstructed according to the maximum speed of the passed road segments which is provided in the digital road networks. The corrected speed assignment is quite important in order to avoid misinterpretations even in case of road type changes (e.g. city road - freeway) between two measured GPS positions. After processing the trajectories of all probe vehicles, a dynamic map with link-based velocities is calculated by a harmonic velocity mean for each road link where hits have been encountered. Having received the vehicle data over a time-period of more than two years, a very accurate segmentation of the daily travel time variation (e.g. for typical weekdays) for each road link can be derived. In Figures 2 recorded positions and derived velocities are depicted for a one-hour period for the Berlin and Vienna data, respectively. The velocities have been calculated based on data from 132 registered vehicles in Berlin and 212 registered vehicles in Vienna. 1 for a more detailed description of the travel time estimation approach, see LORKOWSKI ET AL. (2003) 3 Figure 2: Positions (left) and derived velocities (right) from FCD in Berlin (top) and Vienna (bottom). Data are sampled over one hour. (SCHÄFER ET AL. (2003)) Following this procedure, a time-dynamic road map of historic and real-time link-based velocities becomes available, which is the input for several service applications as well as traffic monitoring purposes. 3. Applications for Individual Travel Services and Traffic Monitoring Traffic congestion is perceived as a major problem in urban areas today. Clearly, only moderate extensions of the transport infrastructure are possible. Therefore, ITS (Intelligent Transport Systems) are regarded a potential alternative means for traffic management for decreasing congestion. Depending on the intensity with which measures can be implemented (e.g. share of vehicles equipped with dynamic routing systems or the utilization of adaptive traffic signal control systems) and the specific situation, PROGNOS AND KELLER (2001) estimate an increase in network capacities by 3% to 10% for Germany. In this view, several metropolitan areas and cities have set up traffic management centres with the aim to provide central access to improved traffic information for users (e.g. in Berlin) or even carrying out adaptive traffic control measures actively (e.g. in Sydney or Tokyo). 4 In order to design appropriate traffic management strategies, monitoring the performance of the transport network as a whole is necessary. This requires comprehensive data on the traffic states of the networks. Presently, the transport management centres rely on data that is collected from specific known hot spots by stationary measures, camera surveillance, as well as police observations and recordings. FCD technology provides an additional data base for monitoring the complete network. Furthermore, it is important to develop performance indicators that quantify congestion and are easy to understand. From the user s perspective the main impact of congestion is the loss of travel time. Therefore, as e.g. QUIROGA (2000, p. 290) points out, traveltime based measures are advantageous compared to capacity or queuing-related measures because they are easily understood by user s as well as traffic managers. Given sufficient data, they can be applied for specific locations as well as whole networks, allow for inter-temporal comparisons and can be easily translated into evaluation measures such as user costs. The travel-time based traffic maps that has been developed in this project are up to this task. These maps can then be applied for traffic monitoring purposes as well as individual travel services such as dynamic routing and navigation. In contrast to conventional router systems where static travel times are used, the routing service based on FCD use real-time information and if these are not available the experiences of the associated taxi fleets to provide users with highquality routing information. Figure 3: Traffic state in Vienna at , early in the morning (7:29) and three hours later (10:29). At this day, the employees of the public transport went on strike. 3.1 Dynamic Routing and Navigation In order to calculate the fastest route for a given journey, conventional routing and navigation systems usually apply average velocities which are derived from default and static link-based speed values. For the Navtech digital network of Berlin, on the average these values show a good representation of the average daily values which have observed using the Taxi FCD. A more detailed analysis shows, however, that for urban freeways, arterial and major routes, FCD velocities were 6.3 km/h lower than the speeds provided by the Navtech data-base. This is less than the km/h speed category 5 band width that Navtech applies. However, on smaller urban roads the difference reaches more than 20 km/h. Furthermore, the average velocities calculated from the FCD data vary considerably during the hours of the day and between links. In consequence, there is a high probability that the average speed values differ considerably from the real values for specific links and during the day and hence, the proposed route and travel time estimation deliver an unrealistic advice. Of course, this difficulty could be overcome if the digital road map attributes are changed from static velocity parameter to real-time measurement from the probe vehicles or at least to historic measurements from a similar period of the past. For demonstration of the differences between the static link-velocity estimations and the link velocities derived from FCD a plot of the Berlin city centre is visualised. Figure 4: Berlin, historical measurements by Taxi-FCD for the afternoon rush-hour (left), Navtech Speed types (right) In order to demonstrate the described benefits of a dynamic routing and navigation system a web-based portal has been developed. The routing server, using a Dijkstra optimisation algorithm, can be operated in different modes, one for real-time optimisation and the other mode for pre-trip planning of a certain journey. In the realtime mode each link of the digital road network is loaded by an average travel time of Taxi floating car data measurements on this road link during the last 30 minutes, by default. In case that no Taxi passed this road link, historical travel time measurements related to the current weekday and time of the day are used. The route optimisation can be chosen by various criteria, calculating the shortest, fastest or cost-efficient path. The routing server can be accessed by various interfaces and devices like a Web Browser, mobile phone or personal digital assistants (PDA). In order to use the routing service on-board the car, a PDA-based off-board navigation system was implemented. This solution has several advantages. The real-time use allows an on-trip optimisation during a journey. Due to the dynamic route optimisation outside the vehicle, real-time traffic data can be considered allowing to bypass a congested region of the city. Furthermore, the digital road map has only to be updated on the server site. Due to the fact that the route is calculated on the server site for each optimisation a certain amount of data have to be transferred between the server and the vehicle. For the data transfer a package-oriented wireless communication service (GPRS) is used which guarantees low expenses for the data transfer. 6 The FCD basis and the route planning system described above are accessible at On the one hand, future commercial services such as fleet management systems could also profit from this data base, on the other hand, this data can be applied to develop improved methodologies for measuring the transport networks performance in urban areas, in particular with respect to congestion and environmental impacts. 4. Application for Emission Monitoring From the public perspective, a further requirement for the development of traffic management strategies is information on environmental and other adverse effects, e.g. to prevent routing through sensitive areas. This requires additional information that has to be generated for the total quantities of traffic on the network segments and for the corresponding environmental impacts. In the following, an approach is presented for computing the emissions of traffic which takes advantage of the temporal and spatial resolution of the Taxi floating car data base. There are three building blocks to this approach: in a first step, the dependence between traffic flow and travel times has to be analyzed. This will lead to a rule-set for deriving flows from travel times. This is necessary, since the FCD provide travel times, but no traffic volumes. Second, the travel times need to be broken down into speed profiles, which of course is very difficult if the distance between subsequent positions becomes too big. Third, the emissions have to be computed from these profiles. This is made more complicated by the fact that usually emission factors are meant as kind of averages, but here truly time- and load-dependent emissions are needed. 4.1 Data Analysis of GPS Taxi-FCD As discussed, to compute emissions first of all the corresponding flows are needed. The floating car data contain just the travel times and velocities that have been observed in the network, but flows are not directly measured. The relation between these quantities is in principle provided by the fundamental diagram, which comes in many flavours for different types of road, speed limits, signalized and un-signalized arterials to name but a few as e.g. applied in the US Highway Capacity Manual. Fundamental diagrams for these conditions show that in a broad range of fairly free traffic flow, speed is nearly constant and independent of traffic density. With the increase of traffic density disturbances between vehicles grow and the speed decreases down to traffic jams and total traffic breakdown at a maximum density (and zero flows). This corresponds with the well-known basic principles of traffic flow theory (e.g. published in LEUTZBACH, 1972). There are two caveats, though. Firstly, fundamental diagrams usually apply to spot measurements, so within the context to be derived here, a kind of fundamental diagram between flow and space mean speed (proportional to inverse travel time) has to be derived. Secondly, inverting the function (usually speed depends on density) in order to estimate traffic densities or flows from observed velocities is difficult. Especially in the 7 low density region with relatively constant speeds, the slope of the function is very steep such that observed velocities fit to a wide range of densities and might indicate a much too high or too low traffic density if only slightly varying from the theoretical values. Thirdly, these functions gener
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks