Automatic vehicle classification system with range sensors

Traffic management systems use inductive loop detectors and more recently video cameras to detect vehicles. Loop detectors are expensive to maintain and video-based systems are sensitive to environmental conditions and do not perform well in vehicle classification. Cameras based upon range sensors are not sensitive to lighting and may be less sensitive to other environmental conditions. In addition, range imagery should provide data to form a good basis for vehicle classification applications. In this paper, we describe methods for processing range imagery and performing vehicle detection and classification. A vehicle classification rate of over 92% accuracy was obtained in classifying vehicles into different vehicle classes.     

Length-based vehicle classification using event-based loop detector data

Length-based vehicle classification is an important topic in traffic engineering, because estimation of traffic speed from single loop detectors usually requires the knowledge of vehicle length. In this paper, we present an algorithm that can classify vehicles passing by a loop detector into two categories: long vehicles and regular cars. The proposed algorithm takes advantage of event-based loop detector data that contains every vehicle detector actuation and de-actuation “event”, therefore time gaps between consecutive vehicles and detector occupation time for each vehicle can be easily derived. The proposed algorithm is based on an intuitive observation that, for a vehicle platoon, longer vehicles in the platoon will have relatively longer detector occupation time. Therefore, we can identify longer vehicles by examining the changes of occupation time in a vehicle platoon. The method was tested using the event-based data collected from Trunk Highway 55 in Minnesota, which is a high speed arterial corridor controlled by semi-actuated coordinated traffic signals. The result shows that the proposed method can correctly classify most of the vehicles passing by a single loop detector.     

Improved waveform‐feature‐based vehicle classification using a single‐point magnetic sensor

Vehicle classification systems have important roles in applications related to real‐time traffic management. They also provide essential data and necessary information for traffic planning, pavement design, and maintenance. Among various classification techniques, the length‐based classification technique is widely used at present. However, the undesirable speed estimates provided by conventional data aggregation make it impossible to collect reliable length data from a single‐point sensor during real‐time operations. In this paper, an innovative approach of vehicle classification will be proposed, which achieved very satisfactory results on a single‐point sensor. This method has two essential parts. The first concerns with the procedure of smart feature extraction and selection according to the proposed filter–filter–wrapper model. The model of filter–filter–wrapper is adopted to make an evaluation on the extracted feature subsets. Meanwhile, the model will determine a nonredundant feature subset, which can make a complete reflection on the differences of various types of vehicles. In the second part, an algorithm for vehicle classification according to the theoretical basis of clustering support vector machines (C‐SVMs) was established with the selected optimal feature subset. The paper also uses particle swarm optimization (PSO), with the purpose of searching for an optimal kernel parameter and the slack penalty parameter in C‐SVMs. A total of 460 samples were tested through cross validation, and the result turned out that the classification accuracy was over 99%. In summary, the test results demonstrated that our vehicle classification method could enhance the efficiency of machine‐learning‐based data mining and the accuracy of vehicle classification. Copyright © 2014 John Wiley & Sons, Ltd.     

Low cost two dimension navigation using an augmented Kalman filter/Fast Orthogonal Search module for the integration of reduced inertial sensor system and Global Positioning System

Due to their complementary characteristics, Global Positioning System (GPS) is integrated with standalone navigation devices like odometers and inertial measurement units (IMU). Recently, intensive research has focused on utilizing Micro-Electro-Mechanical-System (MEMS) grade inertial sensors in the integration because of their low-cost. In this study, a low cost reduced inertial sensor system (RISS) is considered. It consists of a MEMS-grade gyroscope and the vehicle built-in odometer. The system works together with GPS to provide 2D navigation for land vehicles. With adequate accuracy, Kalman filter (KF) is the commonly used estimation technique to achieve the data fusion of GPS and inertial sensors in case of high-end IMUs. However, due to the inherent error characteristics of MEMS grade devices, MEMS-based RISS suffers from the non-stationary stochastic sensor errors and nonlinear inertial errors, which cannot be handled by KF and its linear error models. To overcome the problem, Fast Orthogonal Search (FOS), a nonlinear system identification technique, is suggested for modeling the higher order RISS errors. As a general-purpose numerical method, FOS algorithm has the ability to figure out the system nonlinearity efficiently with a tolerance of arbitrary stochastic system noise. Even using online short-term training data, this method is still able to build an accurate nonlinear model that predicts the system dynamics. Motivated by the above merits, an augmented KF/FOS module is proposed by cascading FOS algorithm to a traditional KF structure. By estimating and reducing both linear and nonlinear RISS errors, the proposed method is supposed to offer substantial enhancement on the positioning accuracy of MEMS-based RISS during GPS outages. In order to examine the effectiveness of the proposed technique, the KF/FOS module is applied on the low cost RISS together with GPS in a land vehicle for several road test trajectories. The performance of the proposed method is compared to KF-only solution, both assessed with respect to a reference offered by a high-end solution. The experimental results confirm that KF/FOS module outperforms KF-only method. The results also show the applicability of the proposed method for real-time vehicle applications.     

Tracking the position of neighboring vehicles using wireless communications

We investigate four communication schemes for Cooperative Active Safety System (CASS) and compare their performance with application level reliability metrics. The four schemes are periodic communication, periodic communication with model, variable communication, and variable communication with repetition. CASS uses information communicated from neighboring vehicles via wireless network in order to actively evaluate driving situations and provide warnings or other forms of assistance to drivers. In CASS, we assume that vehicles are equipped with a GPS receiver, a Dedicated Short Range Communications (DSRC) transceiver, and in-vehicle sensors. The messages exchanged between vehicles convey position, speed, heading, and other vehicle kinematics. This information is broadcast to all neighbors within a specified communication range. Existing literature surmises that in order for CASS to be effective, it may need a vehicle to broadcast messages periodically as often as every 100 ms. In this paper, we introduce the concept of running a kinematic model in-between message transmissions as a means of reducing the communication rate. We use traffic and network simulators to compare the performance of the four schemes. Our performance measure metrics include communication losses as well as average position errors.     

Origin-destination pattern estimation based on trajectory reconstruction using automatic license plate recognition data

Origin-destination (OD) pattern estimation is a vital step for traffic simulation applications and active urban traffic management. Many methods have been proposed to estimate OD patterns based on different data sources, such as GPS data and automatic license plate recognition (ALPR) data. These data can be used to identify vehicle IDs and estimate their trajectories by matching vehicles identified by different sensors across the network. OD pattern estimation using ALPR data remains a challenge in real-life applications due to the difficulty in reconstructing vehicle trajectories. This paper proposes an offline method for historical OD pattern estimation based on ALPR data. A particle filter is used to estimate the probability of a vehicle’s trajectory from all possible candidate trajectories. The initial particles are generated by searching potential paths in a pre-determined area based on the time geography theory. Then, the path flow estimation process is conducted through dividing the reconstructed complete trajectories of all detected vehicles into multiple trips. Finally, the OD patterns are estimated by adding up the path flows with the same ODs. The proposed method was implemented on a real-world traffic network in Kunshan, China and verified through a calibrated microscopic traffic simulation model. The results show that the MAPEs of the OD estimation are lower than 19%. Further investigation shows that there exists a minimum required ALPR sampling rate (60% in the test network) for accurately estimating the OD patterns. The findings of this study demonstrate the effectiveness of the proposed method in OD pattern estimation.     

Improved vehicle classification from dual-loop detectors in congested traffic

Vehicle classification is an important traffic parameter for transportation planning and infrastructure management. Length-based vehicle classification from dual loop detectors is among the lowest cost technologies commonly used for collecting these data. Like many vehicle classification technologies, the dual loop approach works well in free flow traffic. Effective vehicle lengths are measured from the quotient of the detector dwell time and vehicle traversal time between the paired loops. This approach implicitly assumes that vehicle acceleration is negligible, but unfortunately at low speeds this assumption is invalid and length-based classification performance degrades in congestion.To addresses this problem, we seek a solution that relies strictly on the measured effective vehicle length and measured speed. We analytically evaluate the feasible range of true effective vehicle lengths that could underlie a given combination of measured effective vehicle length, measured speed, and unobserved acceleration at a dual loop detector. From this analysis we find that there are small uncertainty zones where the measured length class can differ from the true length class, depending on the unobserved acceleration. In other words, a given combination of measured speed and measured effective vehicle length falling in the uncertainty zones could arise from vehicles with different true length classes. Outside of the uncertainty zones, any error in the measured effective vehicle length due to acceleration will not lead to an error in the measured length class. Thus, by mapping these uncertainty zones, most vehicles can be accurately sorted to a single length class, while the few vehicles that fall within the uncertainty zones are assigned to two or more classes. We find that these uncertainty zones remain small down to about 10 mph and then grow exponentially as speeds drop further.Using empirical data from stop-and-go traffic at a well-tuned loop detector station the best conventional approach does surprisingly well; however, our new approach does even better, reducing the classification error rate due to acceleration by at least a factor of four relative to the best conventional method. Meanwhile, our approach still assigns over 98% of the vehicles to a single class.     

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.     

Estimating historical hourly traffic volumes via machine learning and vehicle probe data: A Maryland case study

This paper focuses on the problem of estimating historical traffic volumes between sparsely-located traffic sensors, which transportation agencies need to accurately compute statewide performance measures. To this end, the paper examines applications of vehicle probe data, automatic traffic recorder counts, and neural network models to estimate hourly volumes in the Maryland highway network, and proposes a novel approach that combines neural networks with an existing profiling method. On average, the proposed approach yields 24% more accurate estimates than volume profiles, which are currently used by transportation agencies across the US to compute statewide performance measures. The paper also quantifies the value of using vehicle probe data in estimating hourly traffic volumes, which provides important managerial insights to transportation agencies interested in acquiring this type of data. For example, results show that volumes can be estimated with a mean absolute percent error of about 21% at locations where average number of observed probes is between 30 and 47 vehicles/h, which provides a useful guideline for assessing the value of probe vehicle data from different vendors.     

A probabilistic map matching method for smartphone GPS data

Smartphones have the capability of recording various kinds of data from built-in sensors such as GPS in a non-intrusive, systematic way. In transportation studies, such as route choice modeling, the discrete sequences of GPS data need to be associated with the transportation network to generate meaningful paths. The poor quality of GPS data collected from smartphones precludes the use of state of the art map matching methods. In this paper, we propose a probabilistic map matching approach. It generates a set of potential true paths, and associates a likelihood with each of them. Both spatial (GPS coordinates) and temporal information (speed and time) is used to calculate the likelihood of the data for a specific path. Applications and analyses on real trips illustrate the robustness and effectiveness of the proposed approach. Also, as an application example, a Path-Size Logit model is estimated based on a sample of real observations. The estimation results show the viability of applying the proposed method in a real route choice modeling context.     

Sensor location problems in path-differentiated congestion pricing

Path-differentiated congestion pricing is a tolling scheme that imposes tolls on paths instead of individual links. One way to implement this scheme is to deploy automated vehicle identification sensors, such as toll tag readers or license plate scanners, on roads in a network. These sensors collect vehicles’ location information to identify their paths and charge them accordingly. In this paper, we investigate how to optimally locate these sensors for the purpose of implementing path-differentiated pricing. We consider three relevant problems. The first is to locate a minimum number of sensors to implement a given path-differentiated scheme. The second is to design an optimal path-differentiated pricing scheme for a given set of sensors. The last problem is to find a path differentiated scheme to induce a given target link-flow distribution while requiring a minimum number of sensors.     

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.     

Noise prediction for highways in Thailand

A model of highway traffic noise is formulated based on vehicle types. The data were collected from local highways in Thailand with free-flow traffic conditions. First, data on vehicle noise was collected from individual vehicles using sound level meters placed at a reference distance. Simultaneously, measurements were made of vehicles’ spot speeds. Secondly, are data for building the highway traffic noise model. This consists of traffic noise levels, traffic volumes by vehicle classification, average spot speeds by vehicle type, and the geometric dimension of highway sections. The free-flow traffic noise model is generated from this database. A reference energy mean emission level (the basic noise) level for each type of vehicles is developed based on direct measurement of Leq (10 s) from the real running condition of each type of vehicles. Modification of terms and parameters are used to make the model fit highway traffic characteristics and different types of vehicle.     

Spatial and temporal characterization of travel patterns in a traffic network using vehicle trajectories

This paper presents a trajectory clustering method to discover spatial and temporal travel patterns in a traffic network. The study focuses on identifying spatially distinct traffic flow groups using trajectory clustering and investigating temporal traffic patterns of each spatial group. The main contribution of this paper is the development of a systematic framework for clustering and classifying vehicle trajectory data, which does not require a pre-processing step known as map-matching and directly applies to trajectory data without requiring the information on the underlying road network. The framework consists of four steps: similarity measurement, trajectory clustering, generation of cluster representative subsequences, and trajectory classification. First, we propose the use of the Longest Common Subsequence (LCS) between two vehicle trajectories as their similarity measure, assuming that the extent to which vehicles’ routes overlap indicates the level of closeness and relatedness as well as potential interactions between these vehicles. We then extend a density-based clustering algorithm, DBSCAN, to incorporate the LCS-based distance in our trajectory clustering problem. The output of the proposed clustering approach is a few spatially distinct traffic stream clusters, which together provide an informative and succinct representation of major network traffic streams. Next, we introduce the notion of Cluster Representative Subsequence (CRS), which reflects dense road segments shared by trajectories belonging to a given traffic stream cluster, and present the procedure of generating a set of CRSs by merging the pairwise LCSs via hierarchical agglomerative clustering. The CRSs are then used in the trajectory classification step to measure the similarity between a new trajectory and a cluster. The proposed framework is demonstrated using actual vehicle trajectory data collected from New York City, USA. A simple experiment was performed to illustrate the use of the proposed spatial traffic stream clustering in application areas such as network-level traffic flow pattern analysis and travel time reliability analysis.     

Empirics of multi-modal traffic networks – Using the 3D macroscopic fundamental diagram

Traffic is multi-modal in most cities. However, the impacts of different transport modes on traffic performance and on each other are unclear – especially at the network level. The recent extension of the macroscopic fundamental diagram (MFD) into the 3D-MFD offers a novel framework to address this gap at the urban scale. The 3D-MFD relates the network accumulation of cars and public transport vehicles to the network travel production, for either vehicles or passengers. No empirical 3D-MFD has been reported so far.In this paper, we present the first empirical estimate of a 3D-MFD at the urban scale. To this end, we use data from loop detectors and automatic vehicle location devices (AVL) of the public transport vehicles in the city of Zurich, Switzerland. We compare two different areas within the city, that differ in their topology and share of dedicated lanes for public transport. We propose a statistical model of the 3D-MFD, which estimates the effects of the vehicle accumulation on car and public transport speeds under multi-modal traffic conditions. The results quantify the effects of both, vehicles and passengers, and confirm that a greater share of dedicated lanes reduces the marginal effects of public transport vehicles on car speeds. Lastly, we derive a new application of the 3D-MFD by identifying the share of public transport users that maximizes the journey speeds in an urban network accounting for all motorized transport modes.     

Modelling dual carriageway lane changing using neural networks

Neural networks offer a potential alternative method of modelling driver behaviour within road traffic systems. This paper explores the application of neural networks to modelling the lane-changing decisions of drivers on dual carriageways. Two approaches are considered. The first, preliminary approach uses a prediction type of neural network with a single hidden layer and the back propagation learning algorithm to model the behaviour of an individual driver. A series of consecutive time-scan traffic patterns, which describe the driver's environment and changes over time as the selected vehicle travels along a link, are input to the neural network, which then predicts the new lane and position of the vehicle. Training data are collected from a human subject using an interactive driving simulation. The trained neural network successfully exhibited the rudiments of driving behaviour in terms of lane and speed changes. A major disadvantage of this approach was the difficulty in recording real-life data, which are required to train the neural network, for individual drivers. The second approach concentrates specifically on lane changing and makes use of a learning vector quantization classification type of neural network. Input to the neural network still consists primarily of time-scan traffic patterns, but the format is changed to facilitate the possibility of data acquisition using image processing. The neural network output classifies the input data by determining the new lane for the vehicle concerned. Performance in both testing and training was very good for data generated by the rule-based driver-decision model of a microscopic simulation. Performance in testing was less satisfactory for data taken directly from a road and highlighted the need for extensive data sets for successful training.     

Vehicle identification sensor models for origin–destination estimation

The traditional approach to origin–destination (OD) estimation based on data surveys is highly expensive. Therefore, researchers have attempted to develop reasonable low-cost approaches to estimating the OD vector, such as OD estimation based on traffic sensor data. In this estimation approach, the location problem for the sensors is critical. One type of sensor that can be used for this purpose, on which this paper focuses, is vehicle identification sensors. The information collected by these sensors that can be employed for OD estimation is discussed in this paper. We use data gathered by vehicle identification sensors that include an ID for each vehicle and the time at which the sensor detected it. Based on these data, the subset of sensors that detected a given vehicle and the order in which they detected it are available. In this paper, four location models are proposed, all of which consider the order of the sensors. The first model always yields the minimum number of sensors to ensure the uniqueness of path flows. The second model yields the maximum number of uniquely observed paths given a budget constraint on the sensors. The third model always yields the minimum number of sensors to ensure the uniqueness of OD flows. Finally, the fourth model yields the maximum number of uniquely observed OD flows given a budget constraint on the sensors. For several numerical examples, these four models were solved using the GAMS software. These numerical examples include several medium-sized examples, including an example of a real-world large-scale transportation network in Mashhad.     

Green vehicle technology to enhance the performance of a European port: A simulation model with a cost-benefit approach

In this paper, we study the impact of using a new intelligent vehicle technology on the performance and total cost of a European port, in comparison with existing vehicle systems like trucks. Intelligent autonomous vehicles (IAVs) are a new type of automated guided vehicles (AGVs) with better maneuverability and a special ability to pick up/drop off containers by themselves. To identify the most economical fleet size for each type of vehicle to satisfy the port’s performance target, and also to compare their impact on the performance/cost of container terminals, we developed a discrete-event simulation model to simulate all port activities in micro-level (low-level) details. We also developed a cost model to investigate the present values of using two types of vehicle, given the identified fleet size. Results of using the different types of vehicles are then compared based on the given performance measures such as the quay crane net moves per hour and average total discharging/loading time at berth. Besides successfully identifying the optimal fleet size for each type of vehicle, simulation results reveal two findings: first, even when not utilising their ability to pick up/drop off containers, the IAVs still have similar efficacy to regular trucks thanks to their better maneuverability. Second, enabling IAVs’ ability to pick up/drop off containers significantly improves the port performance. Given the best configuration and fleet size as identified by the simulation, we use the developed cost model to estimate the total cost needed for each type of vehicle to meet the performance target. Finally, we study the performance of the case study port with advanced real-time vehicle dispatching/scheduling and container placement strategies. This study reveals that the case study port can greatly benefit from upgrading its current vehicle dispatching/scheduling strategy to a more advanced one.     

An empirical assessment of the feasibility of battery electric vehicles for day-to-day driving

Driven by sustainability objectives, Australia like many nations in the developed world, is considering the option of battery electric vehicles (BEVs) as an alternative to conventional internal combustion engine vehicles (ICEVs). In addition to issues of capital and running costs, crucial questions remain over the specifications of such vehicles, particularly the required driving range, recharge time, re-charging infrastructure, performance, and other attributes that will be of importance to consumers. With this in mind, this paper assesses (hypothetically) the extent to which current car travel needs could be met by BEVs for a sample of motorists in Sydney assuming a home-based charging set-up, which is likely to be the primary option for early adopters of the technology. The approach uses five weeks of driving data recorded by GPS technology and builds up home-home tours to assess the distances between (in effect) charging possibilities. An energy consumption model based on characteristics of the vehicle, and the speeds recorded by the GPS is adapted to determine the charge used, while a battery recharge function is used to determine charging times based on the current battery level. Among the most pertinent findings are that over the five weeks, (i) BEVs with a range as low as 60 km and a simple home-charge set-up would be able to accommodate well over 90% of day-to-day driving, (ii) however the incidence of tours requiring out-of-home charging increases markedly for vehicles below 24 kWh (170 km range), (iii) recharge time in itself has little impact on the feasibility of BEVs because vehicles spend the majority of their time parked and (iv) effective range can be dramatically impacted by both how a vehicle is driven and use of electrical auxiliaries, and (v) while unsuitable for long, high-speed journeys without some external re-charging options, BEVs appear particularly suited for the majority of day-to-day city driving in big cities where average journey speeds of 34 km/h are close to optimal in terms of maximising vehicle range. The paper has implications for both policy-makers and auto manufacturers in breaking down some of the (perceived) barriers to greater uptake of BEVs in the future.