A hybrid Bayesian Network approach to detect driver cognitive distraction

Driver cognitive distraction (e.g., hand-free cell phone conversation) can lead to unapparent, but detrimental, impairment to driving safety. Detecting cognitive distraction represents an important function for driver distraction mitigation systems. We developed a layered algorithm that integrated two data mining methods—Dynamic Bayesian Network (DBN) and supervised clustering—to detect cognitive distraction using eye movement and driving performance measures. In this study, the algorithm was trained and tested with the data collected in a simulator-based study, where drivers drove either with or without an auditory secondary task. We calculated 19 distraction indicators and defined cognitive distraction using the experimental condition (i.e., “distraction” as in the drives with the secondary task, and “no distraction” as in the drives without the secondary task). We compared the layered algorithm with previously developed DBN and Support Vector Machine (SVM) algorithms. The results showed that the layered algorithm achieved comparable prediction performance as the two alternatives. Nonetheless, the layered algorithm shortened training and prediction time compared to the original DBN because supervised clustering improved computational efficiency by reducing the number of inputs for DBNs. Moreover, the supervised clustering of the layered algorithm revealed rich information on the relationship between driver cognitive state and performance. This study demonstrates that the layered algorithm can capitalize on the best attributes of component data mining methods and can identify human cognitive state efficiently. The study also shows the value in considering the supervised clustering method as an approach to feature reduction in data mining applications.     

Pedestrian detection for traffic safety based on Accumulate Binary Haar features and improved deep belief network algorithm

ABSTRACT

In order to improve traffic safety and protect pedestrians, an improved and efficient pedestrian detection method for auto driver assistance systems is proposed. Firstly, an improved Accumulate Binary Haar (ABH) feature extraction algorithm is proposed. In this novel feature, Haar features keep only the ordinal relationship named by binary Haar features. Then, the feature brings in the idea of a Local Binary Pattern (LBP), assembling several neighboring binary Haar features to improve discriminating power and reduce the effect of illumination. Next, a pedestrian classification method based on an improved deep belief network (DBN) classification algorithm is proposed. An improved method of input is constructed using a Restricted Bolzmann Machine (RBM) with T distribution function visible layer nodes, which can convert information on pedestrian features to a Bernoulli distribution, and the Bernoulli distribution can then be used for recognition. In addition, a middle layer of the RBM structure is created, which achieves data transfer between the hidden layer structure and keeps the key information. Finally, the cost-sensitive Support Vector Machine (SVM) classifier is used for the output of the classifier, which could address the class-imbalance problem. Extensive experiments show that the improved DBN pedestrian detection method is better than other shallow classic algorithms, and the proposed method is effective and sufficiently feasible for pedestrian detection in complex urban environments.     

Rescuing the captive [mode] user: an alternative approach to transport market segmentation

The prevailing approach to transport market segmentation which identifies two distinct groups, “captive” and “choice” users, has widely been used by professionals and scholars despite the ambiguity associated with these terms. Furthermore, conflicting interpretations from the point of view of decision makers and individuals may result in negative policy implications where the needs of captive users are neglected in favour of attracting new users. This study attempts to address these concerns by proposing an alternative segmentation framework that could be applied to any mode of transport, in any regional context, by users and decision makers alike to better guide the development of transport policies. Using the results of a large-scale transportation survey, a series of clustering techniques are employed to derive this alternative approach for segmenting walkers, cyclists, transit and automobile users. The main factors considered in the final clustering analysis are the level of trip satisfaction and practicality. The analysis yielded four market segments: captivity, utilitarianism, dedication and convenience. Using this theoretical framework to understand the distribution of travellers among market segments is essential in identifying distinct and appropriate policy interventions to improve trip conditions. It is hoped that the segmentation approach and policy framework proposed here will encourage a better balance between pragmatic and idealistic goals in transportation policy.     

Identifying tourists and analyzing spatial patterns of their destinations from location-based social media data

Reliable travel behavior data is a prerequisite for transportation planning process. In large tourism dependent cities, tourists are the most dynamic population group whose size and travel choices remain unknown to planners. Traditional travel surveys generally observe resident travel behavior and rarely target tourists. Ubiquitous uses of social media platforms in smartphones have created a tremendous opportunity to gather digital traces of tourists at a large scale. In this paper, we present a framework on how to use location-based data from social media to gather and analyze travel behavior of tourists. We have collected data of about 67,000 users from Twitter using its search interface for Florida. We first propose several filtering steps to create a reliable sample from the collected Twitter data. An ensemble classification technique is proposed to classify tourists and residents from user coordinates. The accuracy of the proposed classifier has been compared against the state-of-the-art classification methods. Finally, different clustering methods have been used to find the spatial patterns of destination choices of tourists. Promising results have been found from the output clusters as they reveal most popular tourist spots as well as some of the emerging tourist attractions in Florida. Performance of the proposed clustering techniques has been assessed using internal clustering validation indices. We have analyzed temporal patterns of tourist and resident activities to validate the classification of the users in two separate groups of tourists and residents. Proposed filtering, identification, and clustering techniques will be significantly useful for building individual-level tourist travel demand models from social media data.     

Lane changing intention recognition based on speech recognition models

Poor driving habits such as not using turn signals when changing lanes present a major challenge to advanced driver assistance systems that rely on turn signals. To address this problem, we propose a novel algorithm combining the hidden Markov model (HMM) and Bayesian filtering (BF) techniques to recognize a driver’s lane changing intention. In the HMM component, the grammar definition is inspired by speech recognition models, and the output is a preliminary behavior classification. As for the BF component, the final behavior classification is produced based on the current and preceding outputs of the HMMs. A naturalistic data set is used to train and validate the proposed algorithm. The results reveal that the proposed HMM–BF framework can achieve a recognition accuracy of 93.5% and 90.3% for right and left lane changing, respectively, which is a significant improvement compared with the HMM-only algorithm. The recognition time results show that the proposed algorithm can recognize a behavior correctly at an early stage.     

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.     

Large-scale transit market segmentation with spatial-behavioural features

Transit market segmentation enables transit providers to comprehend the commonalities and heterogeneities among different groups of passengers, so that they can cater for individual transit riders’ mobility needs. The problem has recently been attracting a great interest with the proliferation of automated data collection systems such as Smart Card Automated Fare Collection (AFC), which allow researchers to observe individual travel behaviours over a long time period. However, there is a need for an integrated market segmentation method that incorporating both spatial and behavioural features of individual transit passengers. This algorithm also needs to be efficient for large-scale implementation. This paper proposes a new algorithm named Spatial Affinity Propagation (SAP) based on the classical Affinity Propagation algorithm (AP) to enable large-scale spatial transit market segmentation with spatial-behavioural features. SAP segments transit passengers using spatial geodetic coordinates, where passengers from the same segment are located within immediate walking distance; and using behavioural features mined from AFC data. The comparison with AP and popular algorithms in literature shows that SAP provides nearly as good clustering performance as AP while being 52% more efficient in computation time. This efficient framework would enable transit operators to leverage the availability of AFC data to understand the commonalities and heterogeneities among different groups of passengers.     

Daily activity and travel sequences of students,faculty and staff at a large Canadian university

This paper examines the activity engagement, sequencing and timing of activities for student, faculty and staff commuter groups at the largest university in the Maritime Provinces of Canada. The daily activity patterns of all university community groups are modeled using the classification and regression tree classifier algorithm. The data used for this study are derived from the Environmentally Aware Travel Diary Survey (EnACT) conducted in spring 2016 at Dalhousie University, Nova Scotia. Results show that there are significant differences in activity and travel behavior between university population segments and the general population in the region, and between campus groups. For example, students participate in more recreation activities compared to faculty and staff. They also take more trips to and from campus, and are more flexible in their scheduling of trips. The insights gained from this study will provide helpful information for promoting sustainability across university campuses, and for development of campus-based travel demand management strategies.     

Education of truck drivers using the assist driver support system ‐a pilot survey‐

We have developed a driver support system, ASSIST, to decrease automobile driving accidents. Most traffic accidents involve collisions of two objects. A collision occurs when a vehicle's headway is shorter than the stopping distance. Therefore, we plan to warn the driver when the vehicle's headway is shorter than the estimated stopping distance. This driver support system performs exactly that task. Results of experiments verify that this system increases that distance gap by warning the driver to increase the headway.     

A case study of Beijing bus crew scheduling: a variable neighborhood‐based approach

Crew scheduling for bus drivers in large bus agencies is known to be a time‐consuming and cumbersome problem in transit operations planning. This paper investigates a new meta‐heuristics approach for solving real‐world bus‐driver scheduling problems. The drivers' work is represented as a series of successive pieces of work with time windows, and a variable neighborhood search (VNS) algorithm is employed to solve the problem of driver scheduling. Examination of the modeling procedure developed is performed by a case study of two depots of the Beijing Public Transport Group, one of the largest transit companies in the world. The results show that a VNS‐based algorithm can reduce total driver costs by up to 18.1%, implying that the VNS algorithm may be regarded as a good optimization technique to solve the bus‐driver scheduling problem. Copyright © 2015 John Wiley & Sons, Ltd.     

Using automated walking gait analysis for the identification of pedestrian attributes

Collecting microscopic pedestrian behavior and characteristics data is important for optimizing the design of pedestrian facilities for safety, efficiency, and comfortability. This paper provides a framework for the automated classification of pedestrian attributes such as age and gender based on information extracted from their walking gait behavior. The framework extends earlier work on the automated analysis of gait parameters to include analysis of the gait acceleration data which can enable the quantification of the variability, rhythmic pattern and stability of pedestrian’s gait. In this framework, computer vision techniques are used for the automatic detection and tracking of pedestrians in an open environment resulting in pedestrian trajectories and the speed and acceleration dynamic profiles. A collection of gait features are then derived from those dynamic profiles and used for the classification of pedestrian attributes. The gait features include conventional gait parameters such as gait length and frequency and dynamic parameters related to gait variations and stability measures. Two different techniques are used for the classification: a supervised k-Nearest Neighbors (k-NN) algorithm and a newly developed semi-supervised spectral clustering. The classification framework is demonstrated with two case studies from Vancouver, British Columbia and Oakland, California. The results show the superiority of features sets including gait variations and stability measures over features relying only on conventional gait parameters. For gender, correct classification rates (CCR) of 80% and 94% were achieved for the Vancouver and Oakland case studies, respectively. The classification accuracy for gender was higher in the Oakland case which only considered pedestrians walking alone. Pedestrian age classification resulted in a CCR of 90% for the Oakland case study.     

A rule-based neural network approach to model driver naturalistic behavior in traffic

This paper proposes a rule-based neural network model to simulate driver behavior in terms of longitudinal and lateral actions in two driving situations, namely car-following situation and safety critical events. A fuzzy rule based neural network is constructed to obtain driver individual driving rules from their vehicle trajectory data. A machine learning method reinforcement learning is used to train the neural network such that the neural network can mimic driving behavior of individual drivers. Vehicle actions by neural network are compared to actions from naturalistic data. Furthermore, this paper applies the proposed method to analyze the heterogeneities of driving behavior from different drivers’ data.Driving data in the two driving situations are extracted from Naturalistic Truck Driving Study and Naturalistic Car Driving Study databases provided by the Virginia Tech Transportation Institute according to pre-defined criteria. Driving actions were recorded in instrumented vehicles that have been equipped with specialized sensing, processing, and recording equipment.     

The effect of neighbourhood characteristics,accessibility, home–work location,and demographics on commuting distances

The goal of this paper is to better understand home-to-work travel distances throughout the Montréal Metropolitan region. A simultaneous equation modelling analysis is carried out to jointly explain commuter trip length and home–work location as a function of neighbourhood typologies, commuter socio-demographics and measures of job and worker accessibility. First, a factor and cluster analysis of urban form is performed over the entire region on a fine-scale grid pattern. The outcome of this analysis is the classification of typologies at both home and job locations. Different measures of accessibility and commuter socio-demographics are then incorporated into the analysis. Varied data sources including a detailed Montréal Origin–Destination Survey on over 30,000 home-to-work automobile trips are analyzed. Among other results, commuters that live and work in a different sub-region almost double the average trip distance and although socio-economic factors have a statistically significant correlation with commuter distance, these factors have a marginal effect. Interestingly, our results highlight the importance of urban form and job accessibility. Deciding on whether to live and work in the same sub-region was modelled as an endogenous binary random utility model; unobserved heterogeneities seem to be simultaneously influencing both the home–work location choice and trip-to-work distances. Our results underscore the importance of home–work location with respect to urban form and job accessibility. Hence, policies that support more dense and mixed land-use in suburban areas would not be enough to reduce commuter distances. These actions should be accompanied by other policy initiatives to discourage long car trips.     

A latent segmentation based multiple discrete continuous extreme value model

We examine an alternative method to incorporate potential presence of population heterogeneity within the Multiple Discrete Continuous Extreme Value (MDCEV) model structure. Towards this end, an endogenous segmentation approach is proposed that allocates decision makers probabilistically to various segments as a function of exogenous variables. Within each endogenously determined segment, a segment specific MDCEV model is estimated. This approach provides insights on the various population segments present while evaluating distinct choice regimes for each of these segments. The segmentation approach addresses two concerns: (1) ensures that the parameters are estimated employing the full sample for each segment while using all the population records for model estimation, and (2) provides valuable insights on how the exogenous variables affect segmentation. An Expectation–Maximization algorithm is proposed to address the challenges of estimating the resulting endogenous segmentation based econometric model. A prediction procedure to employ the estimated latent MDCEV models for forecasting is also developed. The proposed model is estimated using data from 2009 National Household Travel Survey (NHTS) for the New York region. The results of the model estimates and prediction exercises illustrate the benefits of employing an endogenous segmentation based MDCEV model. The challenges associated with the estimation of latent MDCEV models are also documented.     

Specific object re-identification across non-overlapping camera views in traffic accidents

ABSTRACT

To improve the robustness of object re-identification in complex outdoor environments for traffic safety systems, a novel object re-identification algorithm based on the Individual Similarity Difference Feature (ISDF) method is proposed. This method can provide reliable support for specific object tracking during traffic accidents in video surveillance networks. First, all the images in the gallery are divided into three parts according to a segmentation ratio, and six types of feature for each part are extracted. Second, prototypes for each feature of the three parts are constructed. Third, the image sequence of the same person is grouped, and then the ISDF is extracted from each image. Finally, we use the AdaBoost classifier to judge whether the two objects are matched and then output the final results. Extensive experiments are conducted on two public data sets (Eidgenössische Technische Hochschule Zürich and multi-camera object tracking). The performance of the object re-identification method is superior to the latest methods.     

A utility-theory analysis of automobile speed under uncertainty of enforcement

Highway automobile speed and uncertain enforcement of the speed limit are introduced into a standard household utility model having time and income constraints. Due to uncertainty, expected utility is maximized to obtain the optimal speed (in excess of the speed limit). The optimal amounts of all other commodities and travel are also obtained. The key feature of the model is the risk attitude of the driver and the effect on optimal speed of such attitude. A related feature is the effect of risk attitude on the amount of speed self-insurance. An important finding is that the risk avert (seeking, neutral) driver charges himself an insurance premium that is larger than (smaller than, equal to) what is actuarially sufficient. The relationship between speed, risk attitude, and efficient cost of automobile travel is developed and implications are explored. A parametric analysis is conducted to establish the effect on optimal speed (and other variables) of changes in such policy instruments as the price of gasoline, the probability of being caught exceeding the speed limit, the unit speed fine, and the speed limit. Policy implications of the theoretical results are part of the conclusions.     

Driving situations and driver decisions at road traffic signals

Driver performance in responding to the green-amber-red signal change was studied based on a sample of 2316 last crossing and first stopping vehicles collected by unobtrusive observations at 10 junction approaches in Singapore. Two schemes, a speed-distance diagram (S-D) and an acceleration-deceleration (A-D) diagram, were used to demarcate the driving situations; the driver actions, as revealed outcomes of driver decision-making, were mapped onto these diagrams. The speed-distance diagram can give some indication on what a driver would possibly do. The more complicated acceleration-deceleration diagram is useful for diagnosing the appropriateness of the driver actions. An application of the A-D diagram was demonstrated, and several situations prone to red-running were noted.     

Improving the resolution of gridded-hourly mobile emissions: incorporating spatial variability and handling missing data

To more accurately predict hourly running stabilized link volumes for emissions modeling, a new method was recently developed that disaggregates the period-based model link volumes into hourly volumes using observed traffic count data and multivariate multiple regression (MMR). This paper extends the MMR methodology with clustering and classification analyses to account for spatial variability and to accommodate model links that do not have matching observed traffic count data. The methodology was applied to data collected in the South Air Basin. The spatial analysis resulted in identifying five clusters (or 24-h profiles) for San Diego and two clusters for Los Angeles. The MMR models were then estimated with and without clustering. For San Diego, the disaggregated model volumes with clustering were much closer to the observed volumes than those without clustering, with the exception of the a.m. period. For most hours in Los Angeles, the predicted volumes with clustering were only slightly closer to the observed volumes than those predicted without clustering, suggesting that spatial effects are minimal in Los Angeles (i.e., that 24-h volume profiles are fairly similar throughout the region) and clustering is not necessary. Finally, two classification models, one for San Diego and one for Los Angeles were developed and tested for network link data that does not have matching observed count data. The results indicate the procedure is relatively good at predicting a cluster assignment for the unmatched location for Los Angeles but less accurate for San Diego.