No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Comparing Speed Data from Stationary Detectors Against Floating-Car Data
Abstract
This paper compares speed data measured by induction loops of stationary detectors with reported speeds from floating-car data, which are based on most recent GPS observations of probe vehicles. Detector data are aggregated over a one minute time interval and that means a 30 s delay occurs on average. The time delay issues with respect to floating-car data are quite convoluted with many influences: (i) the update frequencies from vehicles to the backend server, (ii) the fleet size of floating cars, (iii) the current traffic flow, and (iv) the provider treatment. The floating-car dataset has a high spatial resolution with an average segment length of 100 m suited for large-scale traffic observation and management. The spatial dimension of detector data can only be reconstructed ex-post from spotty positions (mean detector positions distance approx. 1.3 km). The paper analyzes which source is more advantageous in terms of detecting traffic jams, high temporal availability of detector data or detailed spatial resolution of floating-car data. An algorithm is presented to compute the jam detection duration, which means we are able to recognize which data source detects the jam earliest. The results demonstrate that there exist regions along certain road stretches where floating-car data clearly outperform stationary data. However, in regions where detectors are densely placed, stationary sensor data recognize a jam situation approx. 2 min earlier than floating-car based speed data. The datasets cover a period of 80 days in 2015 for both driving directions on the German autobahn A9 in the north of Munich.
... The congestion classification introduced by Kessler et al. (2020) is then applied to the data set. The algorithm detects individual congestion elements based on the algorithm (Kessler et al., 2018) and assigns them to one of the congestion patterns defined by Karl et al. (2019). The schematic workflow of the methodology is sketched in Fig. 3. ...
... A possible approach for the fusion would be Kessler et al. (2021). We refer EURO Journal on Transportation and Logistics 13 (2024) 100144 (Kessler et al., 2020(Kessler et al., , 2018Karl et al., 2019). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) ...
... Parameter values to identify and classify congestion clusters according to Karl et al. (2019). to the original papers (Kessler et al., 2018;Karl et al., 2019;Kessler et al., 2020;Kessler, 2021) for further details. Applying this congestion classification to the speed distribution in Fig. 4, the result is depicted in Fig. 5. ...
... The methodology presented was applied to data derived from autobahn A9 near Munich, Germany. The stretch cov- [19]. The data have been interpolated applying the Adaptive Smoothing Method (ASM) [15], [16] to derive smoothed speed values for the space between the detector positions (33 detectors in NB, 27 detectors in SB) [19]. ...
... The stretch cov- [19]. The data have been interpolated applying the Adaptive Smoothing Method (ASM) [15], [16] to derive smoothed speed values for the space between the detector positions (33 detectors in NB, 27 detectors in SB) [19]. The ASM parameters used are as shown in Tab. ...
... PARAMETERS USED FOR SMOOTHING THE DETECTOR DATA[19] congestion is necessary, the parameter is called n StopandGo . Furthermore, it is parametrized how many trajectories are created. ...
... Wide Jam Mega Jam This section describes a methodology to automatically detect congestion and to assign a unique congestion type to each congestion. The algorithm works in two steps: First, coherent, isolated congestion clusters are identified based on the methods described in [15] and [16]. Second, the methodology from [1], which assigns an appropriate congestion type, is applied to each of the found clusters. ...
... Second, the methodology from [1], which assigns an appropriate congestion type, is applied to each of the found clusters. The authors of [15] and [16] propose an algorithm to identify congestion clusters. Starting from a spatio-temporally discretized speed distribution consisting of so-called speed cells, all cells containing a velocity below a congestion threshold !!"#$ are considered. ...
This paper presents an approach that increases the resilience of a freeway network while
differentiating patterns of freeway congestion events and investigating hot spots of each
pattern both spatially and temporally. Based on an automated pattern recognition, an
emerging congestion event can be identified and classified into one of four predefined
congestion patterns. Determining the spatial and temporal extensions of several congestion events, hot spots of each pattern can be localized. Additionally, possible traffic management and control measures are compiled and evaluated by expert statements to mitigate and dissolve the found congestion hot spots. This approach provides a helpful toolbox for freeway operators to classify occurring congestion into predefined categories and to select appropriate countermeasures based on the hot spot analysis to increase the resilience of the overall system. By applying the presented methodology, optimized traffic information is provided to the operator in time-critical situations, which enables an improved decisionmaking process in traffic management. The data base is three large-scale data sets from stationary detectors, vehicle re-identification sensors, and floating car data collected on a German freeway in 2019.
... To foresee motorway congestion locations, [17] built a model that combined FCD and loop detector data. [18] offered a comparison between loop detectors and FCD to determine which is able to identify a traffic event first. [19] proposed combining FCD measurements taken at crossings with loop detector data to estimate queue times and traffic flows. ...
Access to reliable data is crucial, particularly in emerging economies, where data may be scarce. Various collection methods, such as commercial Floating Car Data (FCD) and Bluetooth (BT) traffic data, which are low-cost and widely used methods making them ideal for monitoring and analyzing networks, provide valuable insights. This study presents "Link-Based Generalized Data Fusion Method (LB-GDFM)", which integrates and fuses commercial FCD with BT data to assess and manage traffic congestion on urban arterial links. By leveraging the strengths of both data sources, this fusion method provides a more accurate speed estimation and traffic state assessment of urban links and offers a robust comparative method for determining average link speed using fused FCD and BT data, along with performance results. The method, which involves determining a BT-based fused FCD speed, was applied to selected links of an arterial road in Mersin, Turkiye. Two data collection methods were compared to assess the effectiveness of measuring speed reliability across the links over a corridor. Performance metrics such as MAE FC D, BT , M APE FC D, BT , and RMSE FC D, BT were used to evaluate the accuracy of both data sources, demonstrating that BT-based fused FCD and BT data provide reliable speed estimates in both heavy and normal traffic scenarios. Results indicated that even with limited data sources, effective traffic management strategies can be developed in emerging economies. This study, therefore, provides a valuable data fusion model for traffic monitoring and emphasizes the potential to enhance the used approach by incorporating additional data sources and broader traffic patterns in developing countries.
... The authors (Kessler et al., 2018) compare data from induction loops and floating vehicles obtained from TomTom company from the German A9 motorway. Both datasets have data aggregation in 1 minute. ...
Data from floating vehicles is a modern technology and can be another source of data. There is a free data source available in the Czech Republic, which is relatively new. The addressed source of data from floating vehicles covers the whole Czech Republic, which is a promising source for future use e.g. in transport planning in logistics, estimation of travel times and other related issues. For this reason, it is appropriate to examine the qualitative parameters of the data to see if they characterize the traffic stream. The present paper deals with the size of the processed data. Furthermore, the paper compares the data quality and coverage. January data for four subsequent years was used. The period of the COVID19 pandemic, when traffic declined, was included. Finally, data from selected highways are compared and the period covered is evaluated.
... At the same time, the M. Shaygan et al. fixed-position detector data must be interpolated to reconstruct the spatio-temporal network dynamics during post-processing. Thus, FCD data is generally more helpful in elucidating spatial correlations within the data because it encompasses information about the entire stretch of a roadway link in contrast to isolated and fixed-position detectors (Kessler et al., 2018). Moreover, FCD can provide traffic information in areas of the network where no sensors are deployed. ...
Traffic prediction plays a crucial role in alleviating traffic congestion which represents a critical problem globally, resulting in negative consequences such as lost hours of additional travel time and increased fuel consumption. Integrating emerging technologies into transportation systems provides opportunities for improving traffic prediction significantly and brings about new research problems. In order to lay the foundation for understanding the open research challenges in traffic prediction, this survey aims to provide a comprehensive overview of traffic prediction methodologies. Specifically, we focus on the recent advances and emerging research opportunities in Artificial Intelligence (AI)-based traffic prediction methods, due to their recent success and potential in traffic prediction, with an emphasis on multivariate traffic time series modeling. We first provide a list and explanation of the various data types and resources used in the literature. Next, the essential data preprocessing methods within the traffic prediction context are categorized, and the prediction methods and applications are subsequently summarized. Lastly, we present primary research challenges in traffic prediction and discuss some directions for future research.
... Additionally, improved FCD segments with shorter lengths (up to 220 m) providing more consistency with the road network topology were introduced by INRIX (2018). Even a finer segmentation with lengths of up to 100 m for approaches at signalized urban arterials was created by the TomTom company for urban roads in Munich (Germany) (Kessler et al. 2018). ...
Floating Car Data (FCD) are being increasingly used as an alternative traffic data source due to its lower cost and high coverage area. FCD can be obtained by tracking vehicle trajectories individually or by processing multiple tracks anonymously to produce average speed information commercially. For commercial FCD, the spatio-temporal distribution of these vehicles in actual traffic, traffic Penetration Rate (PR) is the most important factor affecting the accuracy of speed estimations, despite the high number of registered vehicles feeding to an FCD provider, denoting the market PR. This study proposes a method for assessing the traffic PR of commercial FCD by evaluating its speed estimation quality compared to Ground Truth (GT) data. GT speed data were employed to generate different levels of traffic PR using Monte Carlo (MC) simulations, which resulted in the development of Quality-PR (Q-PR) relations for Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) as selected Measures of Effectiveness (MoE). Simulation-based FCD results at an urban road segment in Ankara (Turkey) showed that a quality of FCD with traffic PR of 15% or more would improve significantly. Use of the developed Q-PR relations suggested an approximately 5% traffic PR for the commercial FCD speeds at the location.
... With a k-means algorithm clustering and classification is applied on fundamental diagrams for dynamic traffic assignment. Kessler et al. [66] have compared the measurement results of continuous counting points in the area of congestion zones with floating car data by cluster analysis. The results show that FCD and loop detection speed data are comparable. ...
... The authors of Bachmann et al. (2013b) compared Bluetooth measurements and loop detector data in the Greater Toronto Area on a stretch of several kilometers. In Kessler et al. (2018a), the authors describe an offline comparison between loop detectors and floating cars, determining which is able to detect a traffic incident earlier. In Cohen and Christoforou (2015), the authors statistically analyze the differences between loop detectors and floating car data in the area of Lille, France. ...
This paper studies the joint reconstruction of traffic speeds and travel times by fusing sparse sensor data. Raw speed data from inductive loop detectors and floating cars as well as travel time measurements are combined using different fusion techniques. A novel fusion approach is developed, which extends existing speed reconstruction methods to integrate low-resolution travel time data. Several state-of-the-art methods and the novel approach are evaluated on their performance in reconstructing traffic speeds and travel times using various combinations of sensor data. Algorithms and sensor setups are evaluated with real loop detector, floating car and Bluetooth data collected during severe congestion on German freeway A9. Two main aspects are examined: 1) which algorithm provides the most accurate result depending on the used data and 2) which type of sensor and which combination of sensors yields highest estimation accuracy. Results show that, overall, the novel approach applied to a combination of floating-car data and loop data provides the best speed and travel time accuracy. Furthermore, a fusion of sources improves the reconstruction quality in many, but not all cases. In particular, Bluetooth data only provide a benefit for reconstruction purposes if integrated subtly.
... Shafei et al. [4] verwenden eine Clusteranalyse um Dauerzählstellen mit Haushaltsanalysedaten zu vergleichen und um Kalibrierungs-und Validierungsprozesse zu verbessern. Kessler et al. [5] vergleichen die Messergebnisse von Dauerzählstellen im Bereich von Stauzonen mit Floating Car Data unter der Verwendung der Clusteranalyse. Die Ähnlichkeiten bei Verkehrsunfallsdaten wurden von Yuan et al. [6] analysiert unter der Anwendung einer Clusteranalyse und Faktorenanalyse. ...
... An overview of datasets reported in scientific studies shows the diversity of data sources and providers used nowadays. Kessler et al. (2018) use pre-processed data provided by the Dutch company TomTom to compare congestion analysis results between FCD and SDD. Ebendt et al. (2010) use FCD from a taxi fleet in Berlin in combination with a truck fleet to optimize urban routing. ...
This paper explores the usability of floating car data (FCD) of mixed quality in congestion analysis on motorways. The specific data quality aspects that we are investigating are the number and density of trajectories, the GPS interval, and the fleet representativeness. We use a dataset provided by the German Automobile Club ADAC covering the Tyrolean road network in 2016. From this dataset, trajectories along the A12 motorway were extracted for congestion analysis. These data are characterized by high GPS time interval, low number of trajectories, and are not representative for total traffic due to overrepresentation of trucks. The influence of these quality parameters on congestion identification is explored by analyzing the parameter distribution among different congestion types. In addition, we validate the results by comparing them with congestion incidents obtained from the stationary detector data (SDD) and examining the impact of quality parameters on the validation results. We find that the given data set does not allow short-term congestion patterns to be identified due to quality flaws. Especially the low number of trajectories proved problematic, whereas the influence of other parameters was less distinct. Despite these flaws, for large-scale congestion incidents, floating car data provide outcomes similar to those derived from stationary detectors.
... On the other hand, traffic speed can be detected by high-resolution location and time measurements of floating vehicles, so-called floating-car data (FCD). This technique performs well compared to stationary sensors [1], [2] or in addition to them [3]. ...
... Most of these studies compare the level of similarity between FCD and a ground truth data source, typically stationary detector data, in terms of the relevant traffic variables, e.g., speed and travel time (Jurewicz et al. 2018;de Boer and Krootjes 2012;Clergue and Buttignol 2014;Clergue and Buttignol 2015;Hrubes and Blümelová 2015;Diependaele et al. 2015;Ambros et al. 2017). Some also look at other aspects such as the coverage of the road network (Jurewicz et al. 2018;Aarts et al. 2015) or timeliness to recognize jams (Hu et al. 2016;Kessler et al. 2018;Wang et al. 2014). It has been suggested that theoretically mean point speed from sensors would often be greater than mean link speeds from FCD (Jurewicz et al. 2018) and this has turned out to be the case in some empirical observations (Jurewicz et al. 2018;Clergue and Buttignol 2015;Hrubes and Blümelová 2015). ...
Online real-time traffic data services could effectively deliver traffic information to people all over the world and provide large benefits to the society and research about cities. Yet, city-wide road network traffic data are often hard to come by on a large scale over a longer period of time. We collect, describe, and analyze traffic data for 45 cities from HERE, a major online real-time traffic information provider. We sampled the online platform for city traffic data every 5 min during 1 year, in total more than 5 million samples covering more than 300 thousand road segments. Our aim is to describe some of the practical issues surrounding the data that we experienced in working with this type of data source, as well as to explore the data patterns and see how this data source provides information to study traffic in cities. We focus on data availability to characterize how traffic information is available for different cities; it measures the share of road segments with real-time traffic information at a given time for a given city. We describe the patterns of real-time data availability, and evaluate methods to handle filling in missing speed data for road segments when real-time information was not available. We conduct a validation case study based on Swedish traffic sensor data and point out challenges for future validation. Our findings include (i) a case study of validating the HERE data against ground truth available for roads and lanes in a Swedish city, showing that real-time traffic data tends to follow dips in travel speed but miss instantaneous higher speed measured in some sensors, typically at times when there are fewer vehicles on the road; (ii) using time series clustering, we identify four clusters of cities with different types of measurement patterns; and (iii) a k-nearest neighbor-based method consistently outperforms other methods to fill in missing real-time traffic speeds. We illustrate how to work with this kind of traffic data source that is increasingly available to researchers, travellers, and city planners. Future work is needed to broaden the scope of validation, and to apply these methods to use online data for improving our knowledge of traffic in cities.
This paper investigates the detection rate of various freeway congestion patterns and compares them across different traffic sensor technologies. Congestion events can be categorized into multiple types, ranging from short traffic disruptions (referred to as Jam Wave) to Stop and Go patterns and severe congestion scenarios like Wide Jam. We analyze multiple traffic data sets, including speed data from loop detectors, travel time measurements from Bluetooth sensors, and floating car data (FCD) collected from probe vehicles. Each combination of congestion pattern and detection technology is thoroughly examined and evaluated in terms of its capability and suitability for identifying specific traffic congestion patterns. For our experimental site, we selected the freeway A9 in Germany, which spans a length of 157 km . Our findings reveal that Bluetooth sensors, which record travel times between two locations, are barely suited for detecting short traffic incidents such as Jam Waves due to their downstream detection direction, contrasting with the upstream congestion propagation. Segment-based speed calculations prove more effective in identifying significant congestion events. FCD tend to recognize Stop and Go patterns more frequently than loop detectors but often underestimate severe congestion due to their sensitivity to penetration rates and data availability.
This paper compares speed data measured by induction loops of stationary detectors with reported speeds from floating-car data which are based on most recent GPS observations of probe vehicles. Detector data are aggregated over one minute so they are 30 s old on average. The time delay of floating-car data is more complex. Significant influences are (i) the update frequencies from vehicles to the backend server, (ii) the
fleet size of floating cars, (iii) the current traffic flow, and (iv) the provider treatment. The floating-car dataset has a high spatial resolution with an average segment length of 100m suited for large-scale traffic observation and management. The spatial dimension of detector data can only be reconstructed ex-post from spotty positions (mean detector positions distance approx. 1.3 km).
The paper analyzes which source is more advantageous in terms of detecting traffic jams, high temporal availability of detector data or detailed spatial resolution of floating-car data. The analysis includes spatiotemporal dynamics with traffic jam patterns. Furthermore, an algorithm is presented to compute the jam detection duration meaning which data source recognizes a jam earlier. The results show that regions exist along the considered road stretch where floating-car data clearly outperform stationary data because of their disadvantageous positions but in regions where detectors are placed densely, stationary sensor data recognize a jam situation approx. 2 min earlier than
floating-car based speed data.
The datasets cover a period of 80 days in 2015 for both driving directions on German autobahn A9 in the north of Munich.
We present a new method to obtain spatio-temporal information from aggregated data of stationary traffic detectors, the "adaptive smoothing method". In essential, a nonlin-ear spatio-temporal lowpass filter is applied to the input detector data. This filter exploits the fact that, in congested traffic, perturbations travel upstream at a near-constant speed, while in free traffic, information propagates downstream. As a result, one obtains velocity, flow, or other traffic variables as smooth functions of space and time. Applications include traffic-state visualization, reconstruction of traffic situations from incomplete information, fast identifica-tion of traffic breakdowns (e.g., in incident detection), and experimental verification of traffic models, and even a short-term traffic forecast. We apply the adaptive smoothing method to observed congestion patterns on several German freeways. It manages to make sense out of data where conventional visualization techniques fail. By ignoring up to 65 % of the detectors and applying the method to the reduced data set, we show that the results are robust. The method works well if the distances between neighbouring detector cross sections do not exceed 3 km.
Many real-time traffic-monitoring applications only require speed or travel time. In recent years INRIX Traffic has started collecting and selling real-time speed data collected from "a variety of sources." The clients include direct to consumer and operating agencies alike. So far the INRIX speed data have received little independent evaluation in the literature, with only a few published studies. The current study exploits a unique juncture as the Ohio Department of Transportation transitioned from loop detectors to third party traffic data for real time management. The two traffic surveillance systems operated concurrently for about half a year in Columbus, Ohio, USA. This paper uses two months of the concurrent data to evaluate INRIX performance on 14 mi of 1-71, including both recurrent and non-recurrent events. The work compared reported speeds from INRIX against the concurrent loop detector data, as detailed herein. Three issues became apparent: First, the reported INRIX speeds tend to lag the loop detector measurements by almost 6 min. This latency appears to be within INRIX specifications, but from an operational standpoint it is important that time sensitive applications account for it, e.g., traffic responsive ramp metering. Second, although INRIX reports speed every minute, most of the time the reported speed is identical to the previous sample, suggesting that INRIX is effectively calculating the speeds over a longer period than it uses to report the speeds. This work observed an effective average sampling period of 3-5 min, with many periods of repeated reported speed lasting in excess of 10 min. Third, although INRIX reports two measures of confidence, these confidence measures do not appear to reflect the latency or the occurrence of repeated INRIX reported speeds.
We present an advanced interpolation method for estimating smooth
spatiotemporal profiles for local highway traffic variables such as flow, speed
and density. The method is based on stationary detector data as typically
collected by traffic control centres, and may be augmented by floating car data
or other traffic information. The resulting profiles display transitions
between free and congested traffic in great detail, as well as fine structures
such as stop-and-go waves. We establish the accuracy and robustness of the
method and demonstrate three potential applications: 1. compensation for gaps
in data caused by detector failure; 2. separation of noise from dynamic traffic
information; and 3. the fusion of floating car data with stationary detector
data.
Qualität von Verkehrsinformationen
ABDSB (2017). http://www.abdsb.bayern.de/zahlen.
Bogenberger, K. (2003). Qualität von Verkehrsinformationen. Straßenverkehrstechnik, 47, 518-526.
(a) v crit = 40 km/h, P min = 0.1: matchable jams: 152, deviations: mean = 0.2, median = 2, quartiles at −3 and 5, max = 21, min = −31 (b) v crit = 60 km/h
- Ifac Cts
IFAC CTS
June 6-8, 2018. Savona, Italy
Lisa Kessler et al. / IFAC PapersOnLine 51-9 (2018) 299-304
(a) v crit = 40 km/h, P min = 0.1: matchable jams: 152,
deviations: mean = 0.2, median = 2, quartiles at −3 and
5, max = 21, min = −31
(b) v crit = 60 km/h, P min = 0.1: matchable jams: 244,
deviations: mean = −0.2, median = 1, quartiles at −3
and 4, max = 13, min = −31
QBench-Evaluation of Traffic Flow Quality
- C Lux
Lux, C. (2011). QBench-Evaluation of Traffic Flow
Quality. In C. Lotz and M. Luks (eds.), Qualität von
on-trip Verkehrsinformationen im Straßenverkehr, volume 82 of Berichte der Bundesanstalt für Straßenwesen
: F, Fahrzeugtechnik, 56-63. Wirtschaftsverl. NW, Verl.
für Neue Wiss.
TMCplus-Improving the TMC information chain
- B Rainer
Rainer, B. (2011). TMCplus-Improving the TMC information chain. In C. Lotz and M. Luks (eds.), Qualität
von on-trip Verkehrsinformationen im Straßenverkehr,
volume 82 of Berichte der Bundesanstalt für Straßenwesen: F, Fahrzeugtechnik, 25-31. Wirtschaftsverl. NW,
Verl. für Neue Wiss.
Assessment of data quality needs for use in transportation applications
- H Rakha
- H Chen
- A Haghani
- K F Sadabadi
Rakha, H., Chen, H., Haghani, A., and Sadabadi, K.F.
(2013). Assessment of data quality needs for use in
transportation applications. Report MAUTC-2011-01.
How can we determine the quality of traffic information
- H Rehborn
- B S Kerner
Rehborn, H., Kerner, B.S., and Palmer, J. (2011). How
can we determine the quality of traffic information?
In C. Lotz and M. Luks (eds.), Qualität von on-trip
Verkehrsinformationen im Straßenverkehr, volume 82
of Berichte der Bundesanstalt für Straßenwesen : F,
Fahrzeugtechnik, 46-55. Wirtschaftsverl. NW, Verl. für
Neue Wiss.