Automated classification based on video data at intersections with heavy pedestrian and bicycle traffic: Methodology and application

Pedestrians and cyclists are amongst the most vulnerable road users. Pedestrian and cyclist collisions involving motor-vehicles result in high injury and fatality rates for these two modes. Data for pedestrian and cyclist activity at intersections such as volumes, speeds, and space–time trajectories are essential in the field of transportation in general, and road safety in particular. However, automated data collection for these two road user types remains a challenge. Due to the constant change of orientation and appearance of pedestrians and cyclists, detecting and tracking them using video sensors is a difficult task. This is perhaps one of the main reasons why automated data collection methods are more advanced for motorized traffic. This paper presents a method based on Histogram of Oriented Gradients to extract features of an image box containing the tracked object and Support Vector Machine to classify moving objects in crowded traffic scenes. Moving objects are classified into three categories: pedestrians, cyclists, and motor vehicles. The proposed methodology is composed of three steps: (i) detecting and tracking each moving object in video data, (ii) classifying each object according to its appearance in each frame, and (iii) computing the probability of belonging to each class based on both object appearance and speed. For the last step, Bayes’ rule is used to fuse appearance and speed in order to predict the object class. Using video datasets collected in different intersections, the methodology was built and tested. The developed methodology achieved an overall classification accuracy of greater than 88%. However, the classification accuracy varies across modes and is highest for vehicles and lower for pedestrians and cyclists. The applicability of the proposed methodology is illustrated using a simple case study to analyze cyclist–vehicle conflicts at intersections with and without bicycle facilities.     

Roadway and traffic characteristics for bicycling

The promotion of bicycle transportation includes the provision of suitable infrastructure for cyclists. In order to determine if a road is suitable for bicycling or not, and what improvements need to be made to increase the level of service for bicycles on specific situations, it is important to know how cyclists perceive the characteristics that define the roadway environment. The present paper describes research developed to define which roadway and traffic characteristics are prioritized by users and potential users in the evaluation of quality of roads for bicycling in urban areas of Brazilian medium-sized cities. A focus group discussion identified 14 attributes representing characteristics that describe the quality of roads for bicycling in Brazilian cities. In addition, an attitude survey was applied with individuals to assess their perception on the attributes, along with the importance given to each one of them. The results were analyzed through the Method of Successive Intervals Analysis, which allows the transformation of categorical data into an interval scale. The analysis suggests that both the roadway and traffic characteristics related to segments and those related to intersections are important to the survey respondents. The five most important attributes, in their opinion, are: (1) lane width; (2) motor vehicle speed; (3) visibility at intersections; (4) presence of intersections; and (5) street trees (shading). Therefore, the research suggests that to promote bicycle use in Brazilian medium-sized cities, these attributes must be prioritized.     

Impact of horizontal geometric design of two-lane rural roads on vehicle co2 emissions

In 2014, highway vehicles accounted for 72.8% of all Greenhouse Gases emissions from transportation in Europe. In the United States (US), emissions follow a similar trend. Although many initiatives try to mitigate emissions by focusing on traffic operations, little is known about the relationship between emissions and road design. It is feasible that some designs may increase average flow speed and reduce accelerations, consequently minimizing emissions.This study aims to evaluate the impact of road horizontal alignment on CO₂ emissions produced by passenger cars using a new methodology based on naturalistic data collection. Individual continuous speed profiles were collected from actual drivers along eleven two-lane rural road sections that were divided into 29 homogeneous road segments. The CO₂ emission rate for each homogeneous road segment was estimated as the average of CO₂ emission rates of all vehicles driving, estimated by applying the VT-Micro model.The analysis concluded that CO₂ emission rates increase with the Curvature Change Rate. Smooth road segments normally allowed drivers to reach higher speeds and maintain them with fewer accelerations. Additionally, smother segments required less time to cover the same distance, so emissions per length were lower. It was also observed that low mean speeds produce high CO₂ emission rates and they increase even more on roads with high speed dispersions.Based on this data, several regression models were calibrated for different vehicle types to estimate CO₂ emissions on a specific road segment. These results could be used to incorporate sustainability principles to highway geometric design.     

Speed variation during peak and off-peak hours on urban arterials in Shanghai

Increased speed variation on urban arterials is associated with reductions in both operational performance and safety. Traffic flow, mean speed, traffic control parameters and geometric design features are known to affect speed variation. An exploratory study of the relationships among these variables could provide a foundation for improving the operational and safety performance of urban arterials, however, such a study has been hampered by problems in measuring speeds. The measurement of speed has traditionally been accomplished using spot speed collection methods such as radar, laser and loop detectors. These methods can cover only limited locations, and consequently are not able to capture speed distributions along an entire network, or even throughout any single road segment. In Shanghai, it is possible to acquire the speed distribution of any roadway segment, over any period of interest, by capturing data from Shanghai’s 50,000+ taxis equipped with Global Positional Systems (GPS). These data, hereafter called Floating Car Data, were used to calculate mean speed and speed variation on 234 road segments from eight urban arterials in downtown Shanghai. Hierarchical models with random variables were developed to account for spatial correlations among segments within each arterial and heterogeneities among arterials. Considering that traffic demand changes throughout the day, AM peak, Noon off-peak, and PM peak hours were studied separately. Results showed that increases in number of lanes and number of access points, the presence of bus stops and increases in mean speed were all associated with increased speed variation, and that increases in traffic volume and traffic signal green times were associated with reduced speed variation. These findings can be used by engineers to minimize speed differences during the road network planning stage and continuing through the traffic management phase.     

The relationship between segment-level built environment attributes and pedestrian activity around Bogota’s BRT stations

Few studies have examined the relationship between micro-scale features of the built environment and street segment usage. Micro-scale features of the built environment include the width of the sidewalk, the presence of amenities such as benches and trash bins, and the presence of crossing aids such as stoplights and crosswalks. This study employs segment-level primary data collected for 338 street segments in close proximity to one of 71 bus rapid transit stations in Bogotá, Colombia. We also use secondary data to control for area-level characteristics such as density, socio-economic stratum, unemployment, and crime. Factor and regression analyses are to use identify two dimensions of the built environment that are associated with higher levels of pedestrian activity: pedestrian-friendly amenities, comprised of wider and higher quality sidewalks and the presence of amenities such as benches, garbage cans, and bike paths; and connectivity, comprised of higher levels of road density, three- and four-way intersections, and density. In addition, we find greater pedestrian activity on segments with higher development intensity, with more mix of land uses, and with more crossing aids. Although the relationships identified are not causal, they are suggestive in terms of planning successful built environment interventions.     

Evaluating effects of traffic and vehicle characteristics on vehicular emissions near traffic intersections

Urban air quality is generally poor at traffic intersections due to variations in vehicles’ speeds as they approach and leave. This paper examines the effect of traffic, vehicle and road characteristics on vehicular emissions with a view to understand a link between emissions and the most likely influencing and measurable characteristics. It demonstrates the relationships of traffic, vehicle and intersection characteristics with vehicular exhaust emissions and reviews the traffic flow and emission models. Most studies have found that vehicular exhaust emissions near traffic intersections are largely dependent on fleet speed, deceleration speed, queuing time in idle mode with a red signal time, acceleration speed, queue length, traffic-flow rate and ambient conditions. The vehicular composition also affects emissions. These parameters can be quantified and incorporated into the emission models. There is no validated methodology to quantify some non-measurable parameters such as driving behaviour, pedestrian activity, and road conditions     

Where are the dangerous intersections for pedestrians and cyclists: A colocation-based approach

Pedestrians and cyclists are vulnerable road users. They are at greater risk for being killed in a crash than other road users. The percentage of fatal crashes that involve a pedestrian or cyclist is higher than the overall percentage of total trips taken by both modes. Because of this risk, finding ways to minimize problematic street environments is critical. Understanding traffic safety spatial patterns and identifying dangerous locations with significantly high crash risks for pedestrians and cyclists is essential in order to design possible countermeasures to improve road safety. This research develops two indicators for examining spatial correlation patterns between elements of the built environment (intersections) and crashes (pedestrian- or cyclist-involved). The global colocation quotient detects the overall connection in an area while the local colocation quotient identifies the locations of high-risk intersections. To illustrate our approach, we applied the methods to inspect the colocation patterns between pedestrian- or cyclist-vehicle crashes and intersections in Houston, Texas and we identified among many intersections the ones that significantly attract crashes. We also scrutinized those intersections, discussed possible attributes leading to high colocation of crashes, and proposed corresponding countermeasures.     

Bicycle commuting market analysis using attitudinal market segmentation approach

The market segmentation analysis for bicycle commuting can help identify distinct bicycle market segments and develop specific policies or strategies for increasing the bicycle usage in each segment. This study aims to use the approach of attitudinal market segmentation for identifying the potential markets of bicycle commuting. To achieve the research objective, the household survey is conducted to obtain the travelers’ attitudes towards their commuting travels. The factor analysis is used to explore the latent attitudes. The structural equation modeling (SEM) simultaneously estimates the correlations between the attitudinal factors. The K-means clustering is conducted to segment the bicycle commuting market into several submarkets. Finally, six segments of bicycle commuting market with distinct attitudes are identified by four dividing factors including the willingness to use bicycle, need for fixed schedule, desire for comfort, and environmental awareness. The attitudinal characteristics, socioeconomic features, and actual bicycle choices in each market segment are analyzed and compared. The policy implications that best serve the needs of each submarket are discussed to promote the bicycle commuting.     

Investigating the link between cyclist volumes and air pollution along bicycle facilities in a dense urban core

In this paper we explore the air pollution levels along types of bicycle facilities using a NO₂ land use model previously developed for Montreal. We explore potential associations between bicycle volumes through signalized intersections and pollution levels at those intersections. We further investigate this relationship through the comparison of over thirty cycling corridors as well as an evaluation of the potential exposure of cyclists to air pollution along five routes. We observe NO₂ concentrations to be positively correlated with bicycle flows at the intersection level. We also observe that corridors with either a bicycle path or cycle track generally rank higher in terms of bicycle volume and also have higher NO₂ concentrations than corridors without bicycle facilities. This indicates that intersections and bicycle facilities with a large number of cyclists are also those characterized with the highest air pollution levels.     

Characterizing on-road variables that affect passenger vehicle modal operation

The current research direction in transportation-related air-quality modeling is towards development and implementation of modal emissions models that correlate emission rates to specific ranges of activity. This paper describes a methodology to identify roadway characteristics at signalized intersections which affect the fraction of vehicle activity spend in specific operating modes where modal emission rate models indicate elevated emissions occur to improve vehicle activity inputs to modal emissions models. Field studies using laser guns were conducted on-road collecting second-by-second activity for individual vehicles at signal-controlled intersections and roadway segments. Hierarchical tree-based regression analysis was used to identify on-road geometric and operational characteristics that influenced the fractions of vehicle activity spent in specific modes. Results indicated that queue position, grade, downstream and upstream per-lane hourly volume, distance to the nearest downstream signalized intersection, percent heavy vehicles, and posted link speed limit were the most statistically significant variables.     

Estimating arterial link speed using conventional road detectors

Abstract

In response to an initiative to develop an advanced traffic information system in Bangkok, this paper explores practical guidelines for the optimal location of road sensors, such that the data collected on spot speeds reflect an entire link's average speed. In particular, the authors use microsimulation software to investigate optimal detector locations, using the sum of squared errors and root mean squared errors. The analysis hypothesizes that road segments are 0.4, 0.6, 0.8, 1.0, 2.0 and 3.0 km in length and are specially designed to replicate typical arterial streets in Bangkok. The results show that a single detector location can produce good estimates of link speed only for segments that are shorter than 1.0 km. For distances of 1.0 km or more, the results suggest that two detectors be used for good link speed estimates under all traffic conditions.     

Predicting speeds for rural 2-lane highways

A procedure is developed for estimating highway speeds as a combined function of both traffic volumes and the geometric and environmental conditions surrounding the highway. The proposed method applies a general set of passenger car equivalent units (pcu's) for trucks, recreational vehicles, and opposing flows, in order to reduce traffic volumes to a single variable, which can then be introduced at either an average level, or at contrasting high and low levels, along with geometric/environmental conditions into a multiple linear regression equation which treats all of the factors simultaneously. The procedure is applied to estimate various speeds om rural 2-lane highways inOntario, using a data bank compiled in 1980. Equations are developed for the 10th, 50th and 90th percentile operating speeds. Speeds in Ontario are found to be affected by: traffic volumes, by direction and type of vehicle; access to adjacent land use; access from other highways; speed limit; existence of an extra lane; and grade These Ontario findings are tempered by a lack of severe grades on high-volume roads, and by standard road widths. Thus is recommended that the described procedure be re-applied in another jurisdiction using data which incorporate a wider range of grades, and road widths.     

Reconstructing maximum likelihood trajectory of probe vehicles between sparse updates

Data from connected probe vehicles can be critical in estimating road traffic conditions. Unfortunately, current available data is usually sparse due to the low reporting frequency and the low penetration rate of probe vehicles. To help fill the gaps in data, this paper presents an approach for estimating the maximum likelihood trajectory (MLT) of a probe vehicle in between two data updates on arterial roads. A public data feed from transit buses in the city of San Francisco is used as an example data source. Low frequency updates (at every 200 m or 90 s) leaves much to be inferred. We first estimate travel time statistics along the road and queue patterns at intersections from historical probe data. The path is divided into short segments, and an Expectation Maximization (EM) algorithm is proposed for allocating travel time statistics to each segment. Then the trajectory with the maximum likelihood is generated based on segment travel time statistics. The results are compared with high frequency ground truth data in multiple scenarios, which demonstrate the effectiveness of the proposed approach, in estimating both the trajectory while moving and the stop positions and durations at intersections.     

The impact of speed post-processing methods on regional mobile emissions estimation

Average roadway segment travel speeds play an important role in estimating stabilized running vehicle emissions. Currently stabilized, or hot, running emissions are computed based on speeds produced during the travel demand modeling process. Speed data from the travel forecasting models are widely recognized as being insufficiently accurate for air quality purposes. Frequently post-processing techniques are seen as the most cost-effective means of improving the accuracy of the speed estimates. Using the Sacramento Metropolitan area, this paper focuses on the impacts of different speed post-processors on regional peak period emissions inventories. The results indicated that most post-processed speeds produce consistently and significantly higher running emissions, particularly in locations with heavy traffic. The observed differences in emissions between different types of post-processed speeds vary with congestion level, pollutant type and the underlying approach encapsulated in the speed post-processor calculations. The Sacramento case study suggests that the post-processor used to develop speeds for the purposes of calculating on-road emissions inventories can significantly influence the emissions inventories.     

Time-series model for vehicle speeds

This paper presents a time-series model for the spot speeds of vehicles on a road section. Based on time-series analysis techniques, the model incorporates information on the extent of existing dependency between the speeds of successive vehicles. The model for the data is chosen while relying heavily on the data, and thus emphasis is given to their special characteristics. The advantages of using the model are examined with regard to the relative speed of two successive vehicles along a road section. The results are compared with those obtained by using a model of independent observations; fewer errors are obtained with the time-series model. Therefore, it is concluded that the sequence of speed observations contains valuable information which should be incorporated into speed models.     

A simulation of transit bus emissions along an urban corridor: Evaluating changes under various service improvement strategies

This study investigates the impacts of transit improvement strategies on bus emissions along a busy corridor in Montreal, Canada. The local transit provider, Société de Transport de Montréal, has implemented a number of strategies which include the use of smart cards, limited-stop (express bus) service, and reserved bus lanes along this corridor. Using data collected on-board for instantaneous speeds and stop-level ridership, we estimated bus emissions of greenhouse gases and other pollutants at three levels: road segment, bus-stop, and per passenger. A regression of segment-level emissions against a number of explanatory variables reveals that reserved bus lanes and express bus service reduce emissions significantly. On the other hand, smart card use reduces idling emissions compared to other fare payment methods. Our findings are of most relevance for transit planners who are seeking to implement different strategies to reduce emissions and improve transit performance.     

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.     

Bicyclist commuters’ choice of on-street versus off-street route segments

When using limited funds on bicycle facilities, it would be helpful to know the extent to which a new facility will be used. If a bicycle lane is added to a street, how many bicyclists will no longer use the adjacent sidewalk? If a separate bicycle path is constructed, how many bicyclists will move from the street or sidewalk? This study seeks to identify factors that explain a bicyclist’s choice between available facility choices—off-street (sidewalk and bicycle path) or on-street (bicycle lane and roadway). This paper investigates these issues through a survey of bicyclists headed to Purdue University in West Lafayette, IN, USA. The first data collected to address these questions were “site-based”. Bicyclists were interviewed on campus at the end of their trips and asked which part of the cross-sections along their routes they had used—on-street or off-street. The characteristics of a particular cross-section of street right-of-way were then compared against the characteristics of each bicyclist and his/her observed choice of street, sidewalk, lane, or path. Later, “route-based” serial data were also added. The study developed a mixed logit model to analyze the bicyclists’ facility preferences and capture the unobserved heterogeneity across the population. Effective sidewalk width, traffic signals, segment length, road functional class, street pavement condition, and one-way street configuration were found to be statistically significant. A bicycle path is found to be more attractive than a bicycle lane. Predictions from the model can indicate where investments in particular bicycle facilities would have the most desirable response from bicyclists.     

Mobility and environment improvement of signalized networks through Vehicle-to-Infrastructure (V2I) communications

Traffic signals, even though crucial for safe operations of busy intersections, are one of the leading causes of travel delays in urban settings, as well as the reason why billions of gallons of fuel are burned, and tons of toxic pollutants released to the atmosphere each year by idling engines. Recent advances in cellular networks and dedicated short-range communications make Vehicle-to-Infrastructure (V2I) communications a reality, as individual cars and traffic signals can now be equipped with communication and computing devices. In this paper, we first presented an integrated simulator with V2I, a car-following model and an emission model to simulate the behavior of vehicles at signalized intersections and calculate travel delays in queues, vehicle emissions, and fuel consumption. We then present a hierarchical green driving strategy based on feedback control to smooth stop-and-go traffic in signalized networks, where signals can disseminate traffic signal information and loop detector data to connected vehicles through V2I communications. In this strategy, the control variable is an individual advisory speed limit for each equipped vehicle, which is calculated from its location, signal settings, and traffic conditions. Finally, we quantify the mobility and environment improvements of the green driving strategy with respect to market penetration rates of equipped vehicles, traffic conditions, communication characteristics, location accuracy, and the car-following model itself, both in isolated and non-isolated intersections. In particular, we demonstrate savings of around 15% in travel delays and around 8% in fuel consumption and greenhouse gas emissions. Different from many existing ecodriving strategies in signalized road networks, where vehicles’ speed profiles are totally controlled, our strategy is hierarchical, since only the speed limit is provided, and vehicles still have to follow their leaders. Such a strategy is crucial for maintaining safety with mixed vehicles.     

Updating free speed models through mechanistic principles for Indian conditions

The prediction of free speeds of vehicles is an integral part of the economic appraisal of highways. It is to be noted that speeds not only govern the travel time costs, but also have major impacts on Vehicle Operating Costs (VOC). The World Bank has proposed a mechanistic free speed model based on the limiting speed concept for Highway Design and Maintenance (HDM)‐III. This model along with some refinements has been included in HDM‐4. The underlying assumption in the HDM free speed prediction model is that the free speed at any given point of time is the minimum of possible constraining speeds. This paper mainly addresses the methodology considered to update the free speed models through mechanistic principles (based on HDM‐4). This is accomplished by calibration of the model using the current data on free speeds, road and vehicle characteristics. Subsequently, the validation of the developed models has been carried out.