Current map-matching algorithms for transport applications: State-of-the art and future research directions

Map-matching algorithms integrate positioning data with spatial road network data (roadway centrelines) to identify the correct link on which a vehicle is travelling and to determine the location of a vehicle on a link. A map-matching algorithm could be used as a key component to improve the performance of systems that support the navigation function of intelligent transport systems (ITS). The required horizontal positioning accuracy of such ITS applications is in the range of 1 m to 40 m (95%) with relatively stringent requirements placed on integrity (quality), continuity and system availability. A number of map-matching algorithms have been developed by researchers around the world using different techniques such as topological analysis of spatial road network data, probabilistic theory, Kalman filter, fuzzy logic, and belief theory. The performances of these algorithms have improved over the years due to the application of advanced techniques in the map matching processes and improvements in the quality of both positioning and spatial road network data. However, these algorithms are not always capable of supporting ITS applications with high required navigation performance, especially in difficult and complex environments such as dense urban areas. This suggests that research should be directed at identifying any constraints and limitations of existing map matching algorithms as a prerequisite for the formulation of algorithm improvements. The objectives of this paper are thus to uncover the constraints and limitations by an in-depth literature review and to recommend ideas to address them. This paper also highlights the potential impacts of the forthcoming European Galileo system and the European Geostationary Overlay Service (EGNOS) on the performance of map matching algorithms. Although not addressed in detail, the paper also presents some ideas for monitoring the integrity of map-matching algorithms. The map-matching algorithms considered in this paper are generic and do not assume knowledge of ‘future’ information (i.e. based on either cost or time). Clearly, such data would result in relatively simple map-matching algorithms.     

Developing an enhanced weight-based topological map-matching algorithm for intelligent transport systems

Map-matching (MM) algorithms integrate positioning data from a Global Positioning System (or a number of other positioning sensors) with a spatial road map with the aim of identifying the road segment on which a user (or a vehicle) is travelling and the location on that segment. Amongst the family of MM algorithms consisting of geometric, topological, probabilistic and advanced, topological MM (tMM) algorithms are relatively simple, easy and quick, enabling them to be implemented in real-time. Therefore, a tMM algorithm is used in many navigation devices manufactured by industry. However, existing tMM algorithms have a number of limitations which affect their performance relative to advanced MM algorithms. This paper demonstrates that it is possible by addressing these issues to significantly improve the performance of a tMM algorithm. This paper describes the development of an enhanced weight-based tMM algorithm in which the weights are determined from real-world field data using an optimisation technique. Two new weights for turn-restriction at junctions and link connectivity are introduced to improve the performance of matching, especially at junctions. A new procedure is developed for the initial map-matching process. Two consistency checks are introduced to minimise mismatches. The enhanced map-matching algorithm was tested using field data from dense urban areas and suburban areas. The algorithm identified 96.8% and 95.93% of the links correctly for positioning data collected in urban areas of central London and Washington, DC, respectively. In case of suburban area, in the west of London, the algorithm succeeded with 96.71% correct link identification with a horizontal accuracy of 9.81 m (2σ). This is superior to most existing topological MM algorithms and has the potential to support the navigation modules of many Intelligent Transport System (ITS) services.     

Vehicle Positioning for Improving Road Safety

Abstract

Vehicle positioning is a key requirement for many safety applications. Active safety systems require precise vehicle positioning in order to assess the safety threats accurately, especially for those systems which are developed for warning/intervention in safety critical situations. When warning drivers of a local hazard (e.g. an accident site), accurate vehicle location information is important for warning the right driver groups at the right time. Global positioning system and digital maps have become major tools for vehicle positioning providing not only vehicle location information but also geometry preview of the road being used. Advances in wireless communication have made it possible for a vehicle to share its location information with other vehicles and traffic operation centres which greatly increases the opportunities to apply vehicle positioning technologies for improving road safety. This paper presents a state‐of‐the‐art review of vehicle positioning requirements for safety applications and vehicle positioning technologies. The paper also examines key issues relating to current and potential future applications of vehicle positioning technologies for improving road safety.     

Generating lane-based intersection maps from crowdsourcing big trace data

Lane-based road information plays a critical role in transportation systems, a lane-based intersection map is the most important component in a detailed road map of the transportation infrastructure. Researchers have developed various algorithms to detect the spatial layout of intersections based on sensor data such as high-definition images/videos, laser point cloud data, and GPS traces, which can recognize intersections and road segments; however, most approaches do not automatically generate Lane-based Intersection Maps (LIMs). The objective of our study is to generate LIMs automatically from crowdsourced big trace data using a multi-hierarchy feature extraction strategy. The LIM automatic generation method proposed in this paper consists of the initial recognition of road intersections, intersection layout detection, and lane-based intersection map-generation. The initial recognition process identifies intersection and non-intersection areas using spatial clustering algorithms based on the similarity of angle and distance. The intersection layout is composed of exit and entry points, obtained by combining trajectory integration algorithms and turn rules at road intersections. The LIM generation step is finally derived from the intersection layout detection results and lane-based road information, based on geometric matching algorithms. The effectiveness of our proposed LIM generation method is demonstrated using crowdsourced vehicle traces. Additional comparisons and analysis are also conducted to confirm recognition results. Experiments show that the proposed method saves time and facilitates LIM refinement from crowdsourced traces more efficiently than methods based on other types of sensor data.     

Map-matching poor-quality GPS data in urban environments: the pgMapMatch package

Global Positioning System (GPS) data have become ubiquitous in many areas of transportation planning and research. The usefulness of GPS data often depends on the points being matched to the true sequence of edges on the underlying street network – a process known as ‘map matching.’ This paper presents a new map-matching algorithm that is designed for use with poor-quality GPS traces in urban environments, where drivers may circle for parking and GPS quality may be affected by underground parking and tall buildings. The paper is accompanied by open-source Python code that is designed to work with a PostGIS spatial database. In a test dataset that includes many poor-quality traces, our new algorithm accurately matches about one-third more traces than a widely available alternative. Our algorithm also provides a ‘match score’ that evaluates the likelihood that the match for an individual trace is correct, reducing the need for manual inspection.     

A driving cycle detection approach using map service API

Following advancements in smartphone and portable global positioning system (GPS) data collection, wearable GPS data have realized extensive use in transportation surveys and studies. The task of detecting driving cycles (driving or car-mode trajectory segments) from wearable GPS data has been the subject of much research. Specifically, distinguishing driving cycles from other motorized trips (such as taking a bus) is the main research problem in this paper. Many mode detection methods only focus on raw GPS speed data while some studies apply additional information, such as geographic information system (GIS) data, to obtain better detection performance. Procuring and maintaining dedicated road GIS data are costly and not trivial, whereas the technical maturity and broad use of map service application program interface (API) queries offers opportunities for mode detection tasks. The proposed driving cycle detection method takes advantage of map service APIs to obtain high-quality car-mode API route information and uses a trajectory segmentation algorithm to find the best-matched API route. The car-mode API route data combined with the actual route information, including the actual mode information, are used to train a logistic regression machine learning model, which estimates car modes and non-car modes with probability rates. The experimental results show promise for the proposed method’s ability to detect vehicle mode accurately.     

A lane-level road network model with global continuity

An increasing number of Intelligent Transportation System (ITS) applications require high accurate vehicle positioning, e.g., positioning at the lane-level. This requirement motives the development of modeling the road network at the lane-level. In this paper we propose a novel lane-level road network model. It can be considered an improvement to existing models in its capability of representing the road and intersection details at the lane-level in a uniform and precise way. As a result, the model can guarantee the global continuity for arbitrary map route, which better approximates the real vehicle trajectory. In addition, the map construction algorithms are also developed. Following the proposed methods, the lane parameters can be extracted efficiently under flexible precision requirement, and intersections with varying appearances can be precisely modeled with limited extra data. Experiments were performed to verify the proposed model in representing the lane-level geometrical and topological details of an urban area of Milan. The results also demonstrate the effectiveness of the map construction methods.     

A Trip Reconstruction Tool for GPS-based Personal Travel Surveys

ABSTRACT

This article reports on the development of a trip reconstruction software tool for use in GPS-based personal travel surveys. Specifically, the tool enables the automatic processing of GPS traces of individual survey respondents in order to identify the road links traveled and modes used by each respondent for individual trips. Identifying the links is based on a conventional GIS-based map-matching algorithm and identifying the modes is a rule-based algorithm using attributes of four modes (walk, bicycle, bus and passenger-car). The tool was evaluated using GPS travel data collected for the study and a multi-modal transportation network model of downtown Toronto. The results show that the tool correctly detected about 79% of all links traveled and 92% of all trip modes.     

Urban road transport navigation: performance of the global positioning system after selective availability

Satellite navigation systems have a potential to support multi-modal transport navigation requirements. In road transport, the global positioning system (GPS) is currently supporting a wide variety of in-car navigation and transport telematics systems. The performance of GPS has in the past been limited by the artificial degradation of the signal through the process of selective availability (SA). With SA operational, the instantaneous horizontal positional accuracy was 100 m 95% of the time. Additional infrastructure was used with the differential concept (where range errors are determined at a known location and transmitted to users) to improve this to the level of a few metres. The US Government on 1 May 2000 removed SA. This paper presents the results of a study to assess and characterise the post-SA performance of GPS for positioning vehicles in urban areas. This is an important functionality of advanced transport telematics systems that aim to address everyday problems associated with road transport, particularly in urban areas. The performance assessment addresses, in varying levels of detail, the issues of service coverage, positioning accuracy, integrity and availability of service. Comparison is made with the results of a previous study conducted when SA was turned on. The results show an improvement in stand-alone navigation accuracy without SA compared to the period when SA was operational. Furthermore, no significant difference is seen between the level of accuracy achievable with differential positioning and post-SA stand-alone navigation. The parameters that characterise the performance of GPS determined at the analysis stage have been used to specify an architecture for a local navigation system for urban areas.     

Impact of road gradient on energy consumption of electric vehicles

We investigate the impact of road gradient on the electricity consumption of electric vehicles (EVs) by combining long-term GPS tracking data with digital elevation map (DEM) data for roads in Aichi prefecture, Japan. Eight regression models are constructed and analysed to compare the differences between linear and logarithmic forms of trip energy consumption, differences between considering the road gradient or not, and differences between considering the fixed effects of EVs or not. By categorizing gradients and assigning a percentage of the trip distance to each category, a significantly better model of electricity consumption can be achieved. The results of this study are a novel contribution toward understanding the challenges and benefits associated with downgrade braking on energy regeneration.     

Likelihood-based offline map matching of GPS recordings using global trace information

In batch map matching the objective is to derive from a time series of position data the sequence of road segments visited by the traveler for posterior analysis. Taking into account the limited accuracy of both the map and the measurement devices several different movements over network links may have generated the observed measurements. The set of candidate solutions can be reduced by adding assumptions about the traveller’s behavior (e.g. respecting speed limits, using shortest paths, etc.). The set of feasible assumptions however, is constrained by the intended posterior analysis of the link sequences produced by map matching. This paper proposes a method that only uses the spatio-temporal information contained in the input data (GPS recordings) not reduced by any additional assumption.The method partitions the trace of GPS recordings so that all recordings in a part are chronologically consecutive and match the same set of road segments. Each such trace part leads to a collection of partial routes that can be qualified by their likelihood to have generated the trace part. Since the trace parts are chronologically ordered, an acyclic directed graph can be used to find the best chain of partial routes. It is used to enumerate candidate solutions to the map matching problem.Qualification based on behavioral assumptions is added in a separate later stage. Separating the stages helps to make the underlying assumptions explicit and adaptable to the purpose of the map matched results. The proposed technique is a multi-hypothesis technique (MHT) that does not discard any hypothesized path until the second stage.A road network extracted from OpenStreetMap (OSM) is used. In order to validate the method, synthetic realistic GPS traces were generated from randomly generated routes for different combinations of device accuracy and recording period. Comparing the base truth to the map matched link sequences shows that the proposed technique achieves a state of the art accuracy level.     

Classifying the purpose of stopped truck events: An application of entropy to GPS data

This paper applies the concept of entropy to mine large volumes of global positioning system (GPS) data in order to determine the purpose of stopped truck events. Typical GPS data does not provide detailed activity information for a given stop or vehicle movement. We categorize stop events into two types: (1) primary stops where goods are transferred and (2) secondary stops where vehicle and driver needs are met, such as rest stations. The proposed entropy technique measures the diversity of truck carriers with trucks that dwell for 15 min or longer at a given location. Larger entropy arises from a greater variety of carriers and an even distribution of stop events among these carriers. An analysis confirms our initial hypothesis that the stop locations used for secondary purposes such as fuel refills and rest breaks tend to have higher entropy, reflecting the diversity of trucks and carriers that use these facilities. Conversely, primary shipping depots and other locations where goods are transferred tend to have lower entropy due to the lower variety of carriers that utilize such locations.     

Repeatability & reproducibility: Implications of using GPS data for freight activity chains

As transport modellers we are interested in capturing the behaviour of freight vehicles that includes the locations at which vehicles perform their activities, the duration of activities, how often these locations are visited, and the sequence in which they are visited. With disaggregated freight behaviour data being scarce, transport modellers have identified vehicle tracking and fleet management companies as ideal third party sources for GPS travel data. GPS data does not provide us with behavioural information, but allows us to infer and extract behavioural knowledge using a variety of processing techniques. Many researchers remain sceptical as specific human intervention, referred to as ‘expert knowledge’, is often required during the processing phase: each GPS data set has unique characteristics and requires unique processing techniques and validation to extract the necessary behavioural information. Although much of the GPS data processing is automated through algorithms, human scrutiny is required to decide what algorithmic parameters as considered ‘best’, or at least ‘good’. In this paper we investigate the repeatability and reproducibility (R&R) of a method that entails variable human intervention in processing GPS data. More specifically, the judgement made by an observer with domain expertise on what clustering parameters applied to GPS data best identify the facilities where commercial vehicles perform their activities. By studying repeatability we want to answer the question ‘if the same expert analyses the GPS data more than once, how similar are the outcomes?’, and with reproducibility we want to answer the question ‘if different experts analyse the same GPS data, how similar are their outcomes?’ We follow two approaches to quantify the R&R and conclude in both cases that the measurement system is accurate. The use of GPS data and the associated expert judgements can hence be applied with confidence in freight transport models.     

Vehicle classification from low-frequency GPS data with recurrent neural networks

The categorization of the type of vehicles on a road network is typically achieved using external sensors, like weight sensors, or from images captured by surveillance cameras. In this paper, we leverage the nowadays widespread adoption of Global Positioning System (GPS) trackers and investigate the use of sequences of GPS points to recognize the type of vehicle producing them (namely, small-duty, medium-duty and heavy-duty vehicles). The few works which already exploited GPS data for vehicle classification rely on hand-crafted features and traditional machine learning algorithms like Support Vector Machines. In this work, we study how performance can be improved by deploying deep learning methods, which are recently achieving state of the art results in the classification of signals from various domains. In particular, we propose an approach based on Long Short-Term Memory (LSTM) recurrent neural networks that are able to learn effective hierarchical and stateful representations for temporal sequences. We provide several insights on what the network learns when trained with GPS data and contextual information, and report experiments on a very large dataset of GPS tracks, where we show how the proposed model significantly improves upon state-of-the-art results.     

High accuracy crash mapping using fuzzy logic

Accurate crash location data in crash databases can be shown to be essential for crash modelling, crash mapping, hazardous road segment identification and other studies that aim to decrease the number of crashes within a network area. In this paper a generic and high-accuracy automatic crash mapping method is developed and presented. The methodology is based on a transformed map-matching method for candidate road segment identification and on a fuzzy logic inference system for the final road segment selection. The method is implemented by employing all injury and fatal crashes that occurred during 2012 in the UK Strategic Road Network but can be transferred to other network/crash data. The accuracy of the developed method is estimated to be 98.9% (±1.1%) correct matches. The results of this method are compared to other less advanced crash mapping methods.     

Proposed collision warning system for right-turning vehicles at two-way stop-controlled rural intersections

At two-way stop-controlled (TWSC) rural intersections, a right-turning driver who is departing the minor road may select an improper gap and subsequently may be involved in a rear-end collision with another vehicle approaching on the rightmost lane on the major road. This paper provides perceptual framework and algorithm design of a proposed infrastructure-based collision warning system that has the potential to aid unprotected right-turning drivers at TWSC rural intersections. The proposed system utilizes a radar sensor that measures the location, speed, and acceleration of the approaching vehicle on the major road. Based on these measurements, the system’s algorithm determines if there will be any potential conflict between the approaching and the turning vehicles and warns the driver of the latter vehicle if such a conflict is found. The algorithm is based on realistic acceleration profile of the turning vehicle to estimate its acceleration rates at different times so that the system can accurately estimate the time and distance needed for the departing vehicle to accelerate to the same speed as for the approaching vehicle. That realistic acceleration profile is established using actual experimental data collected by a Global Positioning System (GPS) data logger device that was used to record the positions and instantaneous speeds of different right-turning vehicles at 1-s intervals. The algorithm also gives consideration to the time needed by the driver of the departing vehicle to perceive the message displayed by the system and react to it (to start departure) where it was found that 95% of drivers have a perception–reaction time of 1.89 s or less. A methodology is also illustrated to select the maximum measurement errors suggested for the detectors in measuring the locations of the approaching vehicle on the major road where it was found that the accuracy of the system significantly deteriorates if the errors in measuring the distance and the azimuth angle exceed 0.1 m and 0.2°, respectively. An application example is provided to illustrate the algorithm used by the proposed system.     

Making mode detection transferable: extracting activity and travel episodes from GPS data using the multinomial logit model and Python

The increasing popularity of global positioning systems (GPSs) has prompted transportation researchers to develop methods that can automatically extract and classify episodes from GPS data. This paper presents a transferable and efficient method of extracting and classifying activity episodes from GPS data, without additional information. The proposed method, developed using Python®, introduces the use of the multinomial logit (MNL) model in classifying extracted episodes into different types: stop, car, walk, bus, and other (travel) episodes. The proposed method is demonstrated using a GPS dataset from the Space-Time Activity Research project in Halifax, Canada. The GPS data consisted of 5127 person-days (about 47 million points). With input requirements directly derived from GPS data and the efficiency provided by the MNL model, the proposed method looks promising as a transferable and efficient method of extracting activity and travel episodes from GPS data.     

Optimizing the locations of electric taxi charging stations: A spatial–temporal demand coverage approach

Vehicle electrification is a promising approach towards attaining green transportation. However, the absence of charging stations limits the penetration of electric vehicles. Current approaches for optimizing the locations of charging stations suffer from challenges associated with spatial–temporal dynamic travel demands and the lengthy period required for the charging process. The present article uses the electric taxi (ET) as an example to develop a spatial–temporal demand coverage approach for optimizing the placement of ET charging stations in the space–time context. To this end, public taxi demands with spatial and temporal attributes are extracted from massive taxi GPS data. The cyclical interactions between taxi demands, ETs, and charging stations are modeled with a spatial–temporal path tool. A location model is developed to maximize the level of ET service on the road network and the level of charging service at the stations under spatial and temporal constraints such as the ET range, the charging time, and the capacity of charging stations. The reduced carbon emission generated by used ETs with located charging stations is also evaluated. An experiment conducted in Shenzhen, China demonstrates that the proposed approach not only exhibits good performance in determining ET charging station locations by considering temporal attributes, but also achieves a high quality trade-off between the levels of ET service and charging service. The proposed approach and obtained results help the decision-making of urban ET charging station siting.     

Incident detection methods using probe vehicles with on-board GPS equipment

Mobile communication instruments have made detecting traffic incidents possible by using floating traffic data. This paper studies the properties of traffic flow dynamics during incidents and proposes incident detection methods using floating data collected by probe vehicles equipped with on-board global positioning system (GPS) equipment. The proposed algorithms predict the time and location of traffic congestion caused by an incident. The detection rate and false rate of the models are examined using a traffic flow simulator, and the performance measures of the proposed methods are compared with those of previous methods.     

Predicting travel time reliability using mobile phone GPS data

Estimates of road speeds have become commonplace and central to route planning, but few systems in production provide information about the reliability of the prediction. Probabilistic forecasts of travel time capture reliability and can be used for risk-averse routing, for reporting travel time reliability to a user, or as a component of fleet vehicle decision-support systems. Many of these uses (such as those for mapping services like Bing or Google Maps) require predictions for routes in the road network, at arbitrary times; the highest-volume source of data for this purpose is GPS data from mobile phones. We introduce a method (TRIP) to predict the probability distribution of travel time on an arbitrary route in a road network at an arbitrary time, using GPS data from mobile phones or other probe vehicles. TRIP captures weekly cycles in congestion levels, gives informed predictions for parts of the road network with little data, and is computationally efficient, even for very large road networks and datasets. We apply TRIP to predict travel time on the road network of the Seattle metropolitan region, based on large volumes of GPS data from Windows phones. TRIP provides improved interval predictions (forecast ranges for travel time) relative to Microsoft’s engine for travel time prediction as used in Bing Maps. It also provides deterministic predictions that are as accurate as Bing Maps predictions, despite using fewer explanatory variables, and differing from the observed travel times by only 10.1% on average over 35,190 test trips. To our knowledge TRIP is the first method to provide accurate predictions of travel time reliability for complete, large-scale road networks.