Karim Ismail

Automated Analysis of Pedestrian-Vehicle Conflicts Using Video Data

Pedestrians are vulnerable road users, and despite their limited representation in traffic events, pedestrian-involved injuries and fatalities are overrepresented in traffic collisions. However, little is known about pedestrian exposure to the risk of collision, especially when compared with the amount of knowledge available for motorized traffic. More data and analysis are therefore required to understand the processes that involve pedestrians in collisions. Collision statistics alone are inadequate for the study of pedestrian–vehicle collisions because of data quantity and quality issues. Surrogate safety measures, as provided by the collection and study of traffic conflicts, were developed as a proactive complementary approach to offer more in-depth safety analysis. However, high costs and reliability issues have inhibited the extensive application of traffic conflict analysis. An automated video analysis system is presented that can (a) detect and track road users in a traffic scene and classify them as pedestrians or motorized road users, (b) identify important events that may lead to collisions, and (c) calculate several severity conflict indicators. The system seeks to classify important events and conflicts automatically but can also be used to summarize large amounts of data that can be further reviewed by safety experts. The functionality of the system is demonstrated on a video data set collected over 2 days at an intersection in downtown Vancouver, British Columbia, Canada. Four conflict indicators are automatically computed for all pedestrian–vehicle events and provide detailed insight into the conflict process. Simple detection rules on the indicators are tested to classify traffic events. This study is unique in its attempt to extract conflict indicators from video sequences in a fully automated way.

Large-Scale Automated Analysis of Vehicle Interactions and Collisions

Road collisions are a worldwide pandemic that can be addressed through the improvement of existing tools for safety analysis. A refined probabilistic framework is presented for the analysis of road-user interactions. In particular, the identification of potential collision points is used to estimate collision probabilities, and their spatial distribution can be visualized. A probabilistic time to collision is introduced, and interactions are grouped into four categories: head-on, rear-end, side, and parallel. The framework is applied to a large data set of video recordings collected in Kentucky that contains more than 300 severe interactions and collisions. The results demonstrate the usefulness of the approach for studying road-user behavior and mechanisms that may lead to collisions.

Automated Analysis of Pedestrian-Vehicle Conflicts Context for Before-and-After Studies

This paper presents a novel application of automated video analysis for a before-and-after (BA) safety evaluation of a scramble phase treatment. Data availability has been a common challenge to pedestrian studies, especially for proactive safety analysis. The traditional reliance on collision data has many shortcomings because of the quality and quantity of collision records. Qualitative and quantitative issues with road collision data are more pronounced in pedestrian safety studies. In addition, little information on the mechanism of action implicated can be drawn from collision reports. Traffic conflict techniques have been advocated as supplements or alternatives to collision-based safety analysis. Automated conflict analysis has been advocated as a new safety analysis paradigm that empowers the drawbacks of survey-based and observer-based traffic conflict analysis. One of the areas of focus of pedestrian safety that could greatly benefit from vision-based road user tracking is BA evaluation of safety treatments. This paper demonstrates the feasibility of conducting a BA analysis with video data collected from a commercial-grade camera in Chinatown, Oakland, California. Video sequences for a period of 2 h before and 2 h after scramble were automatically analyzed. The BA results of the automated analysis exhibit a declining pattern of conflict frequency, a reduction in the spatial density of conflicts, and a shift in the spatial distribution of conflicts farther from crosswalks.

Methodologies for Aggregating Indicators of Traffic Conflict

Various indicators of objective conflict have been proposed in the literature to measure the severity of traffic events. Objective conflict indicators measure various spatial and temporal aspects of proximity on the premise that proximity is a surrogate for severity. These aspects of severity may be partially overlapping and in some cases independent. Two sets of conflict indicators were used in a study conducted to demonstrate that integration of the severity cues provided by each conflict indicator could be performed to reflect better the true, yet unobservable, severity of traffic events. The first set of conflict indicators required the presence of a collision course common to the interacting road users. The second set measured severity in mere temporal proximity between road users. The study proposes a methodology with which to aggregate the event-level measurements of conflict indicators into a safety index. First, individual conflict indicator measurements are mapped into severity intervals [0, 1]. Second, these severity indices are aggregated to a safety index that includes both individual severities and exposure. The methodology is applied on individual measurements of pedestrian–vehicle conflicts.

Freeway safety estimation using extreme value theory approaches: A comparative study

The validity of traffic conflicts and other surrogate events has been a great concern in the development and application of surrogate safety measures. Extreme value theory (EVT) offers a strong modeling framework for linking surrogate measures of safety to crash frequency. This study aims at developing, validating, and comparing two EVT modeling approaches for characterizing extreme events. The two alternative EVT approaches, block maxima (BM) and peak over threshold (POT), are used to relate surrogates and lane change maneuver-related crashes on freeways. The surrogate measure is post encroachment times measured from 4189 lane change maneuvers recorded at 29 directional freeway segments with approximately 3-h observation for each segment. The sample size, serial dependency, and non-stationarity issues for both approaches are examined. The comparison of results from the two modeling approaches indicates that the POT approach performs better than BM approach from the aspects of data utilization, estimate accuracy and estimate reliability. This conclusion is drawn on condition of relatively short time observations. An additional comparison is conducted between the estimated crashes and estimated return levels from two approaches. Due to large variances in the estimated crashes, much more robust estimated return levels are recommended for freeway safety evaluation.

Risk-Based Highway Design: Case Studies from British Columbia, Canada

Existing geometric design guides provide deterministic standards for design requirements primarily based on near worst-case scenarios and conservative percentile selection of design parameters. Probabilistic geometric design analysis has been advocated to represent realistically the randomness in design parameters and variables. Probabilistic techniques can provide a measure of the degree of deviation from design standards. Collision modification factors have been advocated as quantitative measures of the impact on safety associated with changes in road features or traffic control. Often, however, no collision modification factors exist in the literature to predict the safety impact of changing particular road features. An important example is sight distance restriction on horizontal curves. Many highways in British Columbia, Canada, are located in mountainous terrain where the additional cost of earthwork or land acquisition to accommodate lateral road expansion can be prohibitive. In this constrained environment, a typical trade-off arises between design requirements (e.g., adequate sight distance on a horizontal curve) and budgetary constraint. The resolution requires comparing the consequences of every alternative. In these cases, reliability analysis can be used to evaluate the risk of deviating from the design requirements. A decision-support tool was developed to compare the risk of different deviations from sight distance requirements. Two case studies were used to investigate the safety implications of sight distance limitation on road segments, the risk associated with deviation from standards, and risk variations among the road segments. The proposed road design is associated with relatively high risk of limited sight distance, and the risk levels associated with standard design requirements vary significantly.

Feasibility of Computer Vision-Based Safety Evaluations: Case Study of a Signalized Right-Turn Safety Treatment

Traditional road safety analysis has often been undertaken with historical collision records. However, limitations on the quality and completeness of collision data gave rise to surrogate ways of measuring safety, especially the traffic conflict technique. Traditionally, traffic conflict techniques have relied on field observations, which have some reliability and repeatability problems. Therefore, successfully automating conflict detection with data extracted from video sensors could have considerable benefits for traffic safety studies. Before-and-after safety evaluations could greatly benefit from automated analysis of traffic conflicts, and the main objective of this paper is to demonstrate the use of this analysis technique for such evaluations. A right-turn safety improvement was implemented at an intersection in Edmonton, Alberta, Canada, in 2009 to mitigate the high rate of rear-end and merging collisions. The right-turn ramp was closed, and all right-turning vehicles were brought to the right-turn lane at the intersection, where a “No-Right-Turn-on-Red” sign was installed. In this study, video sensors were the primary source of conflict data. The video data were analyzed and traffic conflicts were measured with an automated traffic safety tool. The distributions of the calculated conflict indicators before and after the treatment showed a considerable reduction in the frequency and severity of traffic conflicts. This result suggests significant positive changes in rear-end, merging, and total conflicts. The results of this study show the potential benefit of adopting automated conflict analysis for before-and-after safety studies.

Estimation of Frequency and Length of Pedestrian Stride in Urban Environments with Video Sensors

An emphasis on active modes of transportation, that is, walking and cycling, has recently been renewed amid concerns for the environment and public health. However, the focus of research and practice that these modes have traditionally received is secondary to that received by motorized modes. As a consequence, the data on pedestrians (in particular, microscopic data) required for analysis and modeling are lacking. For instance, accurate data on the length of individual stride are not available in the transportation literature. This paper proposes a simple method to extract frequency and length of pedestrian stride automatically from video data collected nonintrusively in outdoor urban environments. The walking speed of a pedestrian oscillates during each stride; the oscillation can be identified through the frequency analysis of the speed signal. The method was validated with real-world data collected in Rouen, France, and Vancouver, Canada, where the root mean square errors for stride length were 6.1 and 5.7 cm, respectively. A method to distinguish pedestrians from motorized vehicles is proposed and used to analyze the 50 min of the Rouen data set to provide the distributions of stride frequency and length.

Automated Analysis of Pedestrian Crossing Speed Behavior at Scramble-phase Signalized Intersections Using Computer Vision Techniques

ABSTRACT This study investigates changes in pedestrian speed behavior following the implementation of a scramble phase. The aim is to demonstrate the feasibility of automatic collection of pedestrian data; and to study pedestrian speed variations with respect to design changes to intersection crossings. The results show that the average crossing speed is higher after the implementation of the scramble phase. Within the scramble, the average crossing speed is higher for diagonal crossing than side crosswalk crossing. The average crossing speed is lower for pedestrians crossing during the Walk interval. Pedestrians have higher speed through the first half of the crosswalk.

Methodology for safety optimization of highway cross-sections for horizontal curves with restricted sight distance

Several earlier studies have noted the shortcomings with existing geometric design guides which provide deterministic standards. In these standards the safety margin of the design output is generally unknown and there is little knowledge of the safety implications of deviating from the standards. To mitigate these shortcomings, probabilistic geometric design has been advocated where reliability analysis can be used to account for the uncertainty in the design parameters and to provide a mechanism for risk measurement to evaluate the safety impact of deviations from design standards. This paper applies reliability analysis for optimizing the safety of highway cross-sections. The paper presents an original methodology to select a suitable combination of cross-section elements with restricted sight distance to result in reduced collisions and consistent risk levels. The purpose of this optimization method is to provide designers with a proactive approach to the design of cross-section elements in order to (i) minimize the risk associated with restricted sight distance, (ii) balance the risk across the two carriageways of the highway, and (iii) reduce the expected collision frequency. A case study involving nine cross-sections that are parts of two major highway developments in British Columbia, Canada, was presented. The results showed that an additional reduction in collisions can be realized by incorporating the reliability component, P(nc) (denoting the probability of non-compliance), in the optimization process. The proposed approach results in reduced and consistent risk levels for both travel directions in addition to further collision reductions.