John Hourdos

Outline for a causal model of traffic conflicts and crashes

Road crashes tend to be infrequent but with nontrivial consequences, leading to a long-running interest in identifying surrogate events, such as traffic conflicts, that can support a timely recognition and correction of safety deficiencies. Although a variety of possible surrogates have been suggested, questions remain regarding how crashes and surrogates are related. Using recent developments in causal analysis we propose a simple model which represents crashes and conflicts as resulting from interactions between initiating conditions and evasive actions, and then use this model to identify relationships between these types of events. Our first set of results expresses the probability of a crash as a mixture of probabilities over different sets of initiating conditions, where the mixing probabilities are governed by the evasive action. Our second set of results considers situations where sampling is restricted to non-crash events, and gives conditions where these truncated probabilities can serve as proxies for crash probabilities. We then illustrate how trajectory-based reconstruction can be used to classify initiating events with respect to severity, and to estimate a proxy for the crash probability from a set of incompletely observed non-crash events.

The Less-Than-Perfect Driver: A Model of Collision-Inclusive Car-Following Behavior

Unlike traditional car-following models that preclude vehicle collisions, a proposed model aims to emulate less-than-perfect everyday driving while capturing both safe and unsafe driver behavior. Most important, a realistic perception-response process is incorporated into the model on the basis of developments from visual perception studies. Driver inattention is characterized by a driver-specific variable called the scanning interval. This variable, when coupled with the drivers visual perception-response process, results in variable reaction times that are dependent not only on each drivers individual characteristics but also on instantaneous traffic conditions such as speed and density. This allows closer emulation of real-life human driving and its interactions with surrounding vehicles. Both inter- and intradriver variations in reaction time are captured in a plausible and coherent manner; in earlier studies, reaction time either was presumed fixed or was of limited variability. Furthermore, parameters of this model have a direct physical and behavioral meaning; this implies that vehicle collisions, if any, can be analyzed for behavioral patterns rather than simply being treated as numerical artifacts. In all, 54 detailed and accurate vehicle trajectories extracted from 10 real-life crashes were used to test the models capability of replicating freeway rear-end collisions. High-resolution crash-free trajectory data were used to validate the model against normal driving behavior. Test results indicate that the proposed model is able to replicate both normal and unsafe driving behavior that could lead to vehicle collisions. The feasibility of integrating the proposed model with existing microsimulators is discussed. The outcome of this work could facilitate studying crash mechanisms at a high-definition microscopic level and could enable safety-related system design improvements and evaluation through microsimulation software.

Opportunities for Preventing Rear-End Crashes: Findings from the Analysis of Actual Freeway Crash Data

To date most traffic crash analyses have been conducted using aggregated crash data. The main focus was given to determining the relationship between crashes and corresponding variables such as traffic volume, speed, speed variance, and geometry conditions. Few studies have focused on the cause of crashes at the individual vehicular level. Recently, the Minnesota Traffic Observatory at the University of Minnesota developed a set of vehicle trajectory data containing five actual rear-end crashes. This article analyzes these data and attempts to establish a trigger factor for preventing crashes. An inverse of time-to-collision value of 0.4 detected all five actual crashes before the collision, but with a large number of false alarms. An additional trigger factor, the deceleration rate difference between leading and following vehicles greater than 15 ft/sec2, completely eliminated those false alarms. In addition, it was found that an advanced warning intended to alert the driver offers little help in preventing the crashes. This is because a driver reaction time of about 0.57 sec is required before initiating deceleration. Thus, the deceleration rate required to avoid a crash became impractical, resulting in actual avoidance of only 20% crashes. This indicated that an automated braking system should be applied to prevent crashes or effectively mitigate the crash impacts.

Improving Minnesota's Stratified Ramp Control Strategy

Minnesotas most recent stratified zone ramp metering (SZM) strategy, the successor of the ZONE metering algorithm, has been deployed in the Minneapolis-St. Paul area since late 2002. The SZM strategy is intended to maximize freeway throughput while keeping ramp waiting times below a predetermined threshold. Preliminary evaluation results confirmed the overall effectiveness of the SZM strategy but also revealed that the freeway performance was compromised by the restrictive maximum ramp waiting constraint. Consequently, improvements were sought. These focus on a better determination of the minimum release rate for each ramp and its integration with the overall SZM strategy. Both the current and the enhanced SZM strategies are tested in two freeway sites under various demand scenarios through a state-of-the-art microscopic simulator. The simulation results indicate that the enhanced SZM strategy is effective for delaying and decreasing the freeway congestion and results in smoother freeway traffic flow com...

Comprehensive Evaluation of New Integrated Freeway Ramp Control Strategy

A new integrated ramp control strategy, recently deployed by the Minnesota Department of Transportation in the Twin Cities metropolitan area, is evaluated. This strategy, stratified zone metering (SZM), takes into account real-time ramp demand and queue size information and aims to strike a balance between two competing objectives: improving freeway efficiency and preventing excessive ramp delays. In this study, the SZM strategy was compared to the earlier ZONE metering strategy as well as to the no-control alternative. Comprehensive metrics were generated through rigorous microsimulation to assess critical aspects of the strategys performance. The evaluation results are consistent with qualitative field observations and confirm that SZM strategy improves freeway efficiency when compared with the no-control alternative, reduces freeway travel time and delay, improves freeway speed, smooths freeway flow, and reduces the number of stops. More important, excessive queue spillbacks and ramp delays were signi...

Preprocessing Volume Input Data for Improved Traffic Simulation

Microsimulation has become an increasingly indispensable tool in demanding intelligent transportation systems and planning applications. To build reliable and realistic simulation models, high-quality input data, including roadway geometry, vehicle and driver characteristics, traffic volumes, and composition, are required. Volumes of these data are important but hard to obtain; even when collected with advanced surveillance systems, they are susceptible to miscounting, gaps in time and space, and other inaccuracies. Data that appear to be accurate often do not balance out (i.e., they are inconsistent in terms of maintaining conservation throughout the system). These problems could lead to anomalies or errors during the simulation, seriously tainting the reliability and accuracy of the outputs and weakening the credibility of the conclusions. A comprehensive methodology is proposed for improving the quality of freeway traffic volumes for simulation purposes. Established and enhanced procedures for checking...

Directional Sound for Long-Distance Auditory Warnings From a Highway Construction Work Zone

Directional sound can be used to provide warnings to specific vehicles without disturbing other vehicles on the highway. An example of such an application is the need to alert a vehicle that is likely to intrude into a highway construction work zone. Long-distance auditory warnings potentially reduce the time for the driver to visually locate the work zone. This paper reviews the currently available technologies that can potentially be used to develop a long-distance auditory warning system for highway work-zone applications. Of these, ultrasound-based parameter arrays and time-delay-controlled arrays of compact ordinary speakers are taken up for detailed analysis and experimental evaluation. An ultrasound-based parametric array is found to be effective at generating highly directional sound. However, when issues of cost, installation, maintenance, and price are considered, the more suitable technology for work zones is found to be arrays of flat-panel loudspeakers with time delay control. Such a system is inexpensive and can be used to effectively generate directional sound for long-distance auditory warnings. This paper shows that an annular pattern of flat-panel speakers can provide directional sound along a highway lane, with no real-time control of time delay necessary. Hence, an extremely inexpensive and portable system can be obtained, with components consisting of compact flat-panel speakers, a battery, power supply, and inexpensive electronics. In terms of performance, the developed system can provide a difference in the sound level of 6 dB or higher between adjacent lanes at all frequencies in the range of 2-4 kHz at distances of up to 40 m from the location of the warning system.

Evaluation of the Roundabout Capacity Model in HCM6 Edition and HCM 2010 on a Multilane Roundabout

This study evaluated the performance of the new Highway Capacity Manual 6th Edition (HCM6) roundabout capacity model at a multilane roundabout in Richfield, Minnesota. Traffic flow rates and gap acceptance data were collected during 20u2009hours of afternoon peak period traffic over four days. The observed critical headways for the left and right lanes was equal to 4.43 and 3.99u2009s while the observed follow-up headways were 3.05 and 2.96u2009s, respectively. Roundabout capacity curves for both lanes were built through the collected flow information, as well as through the observed headways and the HCM formula. As compared with the field observed results, the default model in HCM6 overestimated the capacity of the study roundabout. The default model in HCM 2010 closely estimated the capacity for the right lane, but overestimated the capacity of the left lane when the circulating flow rate was high. The HCM6 model was calibrated with the observed critical and follow-up headways, as well as by calibrating the intercept of the model using only the follow-up headway. The fully calibrated model overestimated the capacity of the right lane by 7–10% as the circulating flow decreases and overestimated the capacity of the left lane by 7% and 31%. The partially calibrated model overestimated the capacity of the left lane by 6% and overestimated the capacity of the right lane by at most 10% under low circulating flow rate, but it underestimated the capacity by at most 21% under high circulating flow rate.

Design, specification, implementation and evaluation of a freeway queue warning system

The formation and propagation of queues on urban freeways is an unavoidable result of the ever-increasing traffic demand. This paper presents the design, specification, implementation and evaluation of an infrastructure based Queue Warning system (QWARN) that is capable of detecting dangerous traffic conditions, i.e. crash-prone conditions, on freeways and delivering warning messages to drivers, in order to increase their alertness to these traffic conditions and ultimately reduce the crash frequency on urban freeways. This study utilizes measurements of individual vehicle speeds and time headways at two fixed locations on the freeway mainline. The Queue Warning system was implemented at the right lane of a 1.7-mile-long freeway segment of Interstate 94 WB where the event frequency prior to the systems installation was 11.9 crashes per million vehicles traveled and 111.8 near crashes per million vehicles traveled. The control algorithm assesses the dangerousness of the given traffic condition and responds with a warning result based on a multi-metrics traffic evaluation model. The system translates the warning result into readable messages and delivers them to the two sets of signs located upstream of the detection zone. A three-month investigation of the operations of the QWARN system showed the event frequency reduced to 9.34 crashes per million vehicle traveled and 51.8 near crashes per million vehicle traveled. The result shows a 20% decrease of crash frequency and the feasibility of the proposed methodology.

Developing a Large-Scale Hybrid Simulation Model: Lessons Learned

Long-term, regional travel demand models are essential tools used by planning organizations for resource management, project scheduling, and impact studies. Developed primarily at the macroscopic level, these tools lack sufficient detail to capture the influence of local geometry, dynamic traffic controls, or advanced transportation demand management strategies. To bridge the gap, a hybrid mesoscopic–microscopic model was developed. The core of the model, surrounding two light rail corridors in Minneapolis–Saint Paul, Minnesota (the Twin Cities), was developed at high resolution for microscopic simulation to capture the interaction between traffic signals, transit systems, and the road network. The remainder of the greater Twin Cities area was implemented according to the regional planning model (RPM) maintained by the Metropolitan Council. Interfacing the AIMSUN-based hybrid model with the CUBE-based RPM, the Twin Cities metro hybrid simulation was used to improve mode choice and traffic assignment iteratively to achieve a dynamic user equilibrium state. Significant lessons were learned about the effort needed to develop and to maintain such a model, with implications for future large-scale regional modeling.