Journal of Transportation Engineering, Part A: Systems

The comfort threshold is one of the basic controls in designing horizontal alignment. Some studies have implied that the recommended comfort thresholds are conservative because they were determined using the traditional approach for vehicle design and corresponding driver behavior during the 1930s and 1940s. Drivers exceed the recommended comfort thresholds (because they are conservative) when traversing horizontal curves, which results in overturning and lateral skidding of the vehicle, increasing the chances of crashes on the horizontal curves. Therefore, the design guidelines need to consider comfort thresholds determined using recent approaches in data collection for modern vehicle design and corresponding driver behavior. Limited studies have determined the comfort thresholds for horizontal curves. Specifically, no studies determined the comfort thresholds for the horizontal alignment design in India. Thus, this study attempted to estimate the comfort thresholds for horizontal curves using an advanced Global Positioning System (GPS) device instrumented in passenger cars for data collection on two-lane rural highways in India. An all-subset regression approach was used to model comfort thresholds, and statistics such as Akaike information criterion (AIC), coefficient of determination, and cross-validation were used in the model selection. The study results showed no significant difference in the comfort thresholds on the right-turn and left-turn curves. Among the various geometric design features of curves, the curve radius significantly influenced the variation in the comfort threshold. The estimated comfort thresholds were higher than the side-friction demand recommended in an Indian design guideline. This study highlights the importance of determining the comfort thresholds using an advanced data collection tool for modern vehicle design and corresponding driver behavior to avoid crashes on the horizontal curves.

The horizontal alignment of existing railway must be re-created and calibrated periodically to meet the requirements of safety, stability and comfort. Most existing re-creation methods perform local section optimization, in which the whole alignment is divided into several sections, depending on the number and positions of intersection points, and then each section is re-created successively. The final whole re-created horizontal alignment is generated by combining optimized re-created sections. In this process, the re-created result of one section must be adjusted in re-creating the next section. Thus, the optimized re-created result of the previous section cannot be preserved. To overcome this problem, some methods use the result of local section optimization as the initial value, and then further optimize the whole alignment. However, the final re-created alignment is affected easily by the local section optimization, so that the re-created alignment is not optimized. This paper proposes an overall swing iteration based on points-alignment consistency integrated with mesh adaptive direct search (MADS). At each iteration, the adverse effect of re-creating successive sections is avoided by preferentially fitting all shared tangents which link nearby sections and then fitting curved segments simultaneously. To optimize curved segments further, MADS is introduced to optimize the parameters and handle constraints. After each iteration, the attributes of measured points for the whole alignment are adjusted based on points-alignment consistency until they are consistent with the attributes of re-created geometric elements. Applications demonstrated that the proposed approach can re-create an optimized horizontal alignment while satisfying multiple constraints.

Roadway improvements to reduce the frequency of crashes are of the utmost priority to transportation agencies. To a great extent, implementation of improvement programs depends on the reliable identification of roadway segments with high crash risk. Among all crash types, wrong-way driving (WWD) crashes are considered random in nature and are a major safety concern. The Federal Highway Administration defines WWD specifically for high-speed divided highways and access ramps. This definition excludes all other roadway classifications when a crash occurs in the opposing direction to the legal flow of traffic. Screening 5 years of crash data in Minnesota revealed that WWD resulted in crashes on other types of roadway functional classes. This work aimed to (1) introduce a new term/acronym to the literature for driving in the wrong direction (DWD) on all roadway functional classes, (2) apply a set of count data models to estimate the occurrence of DWD crashes, (3) identify roadway geometric features of high-risk segments for DWD crashes, (4) investigate random effects of covariates due to unobserved factors, and (5) calculate elasticity effects of variables. Final models’ specifications indicate that the negative binomial (NB) mixed effect model was found to be the best-fit model. Focusing on DWD crashes, we uncovered the factors contributing to higher DWD crash-risk segments: log of average annual daily traffic (AADT), number of lanes, sidewalk, and shoulder type. The change in frequency of crashes is also investigated using marginal effects, and safety interventions for preventing DWD crashes are also discussed. Transportation agencies can use the findings of this research, in terms of contributing factors and their relative effects on DWD crashes, to deploy appropriate countermeasures at high-risk locations.

Railway curve realignment is critical for rectifying railway alignment deviations caused by excessive train load and repeated repairs. The existing realignment methods have limitations, such as low efficiency and precision, when considering realigning curves without key parameter information (CWI). To address the CWI issues, this study proposes a range identification and adaptive simplified particle swarm optimization (RI-ASPSO) algorithm combined with the existing principle of realigning railway curves. In this algorithm, the RI is designed to identify the range of curve parameters and is the premise of the ASPSO. Moreover, an automatic update strategy of the velocity threshold and an adaptive local random search strategy are developed in the ASPSO to efficiently and stably search the final near-optimal solution. The method is applied in real-world case studies, and the results show that the RI-ASPSO outperforms the particle swarm optimization (PSO) algorithm and coordinate method with higher accuracy, higher efficiency, and less deviation.

Abstract

The origin and destination (O-D) of public transit passengers are important for the planning and operation of the transit system. However, only 46% of public transit agencies have a smart card system in the US, and most of them require an entry-only tap, which prohibits identifying passenger destinations unless utilizing an estimation model. Therefore, there is a need for a cost-effective and automated solution to facilitate the majority of the US transit agencies in recognizing the origin and destination of passengers as well as capturing passenger transfers. This paper created a novel algorithm for transit agencies to count passengers with the consideration of transfers using a cost-effective Wi-Fi sensing–based approach. Two pilot studies were conducted in the city of Louisville, Kentucky on three different bus routes to explore the feasibility of the method. A Wi-Fi detector was installed in the bus to detect passengers, and a manual counting was performed to be used as ground truth data. After data collection, the proposed algorithms were applied to optimize the detection radius and to eventually find origins, destinations, and transfers. Analysis revealed that the proposed Wi-Fi–based approach is capable of recognizing 78.7% of the total passengers as well as detecting their boarding and alighting activities. The paper demonstrates the ability of the proposed method to detect passengers with a reasonable detection rate by using Wi-Fi technology on bus routes, which makes it feasible for transit agencies to conduct frequent and low-cost network-level passenger O-D studies.

Abstract

This work utilized Wi-Fi technology to count the passengers of public transit and to identify transfers. Current practice in data collection is transitioning from traditional manual methods toward more-automated approaches. The current practice relies on a manual method such as surveys or newer technologies such as gaining information from a smart card that passengers swipe when they get on a bus. Surveys are a manual method that requires intensive effort to gather the data and process it. On the other hand, the smart card system can be expensive to install and operate, and most public transit providers in the US have limited budgets and large coverage areas. The algorithm in this work is a practical solution for public transit agencies to gain information about the boarding and alighting activities along their routes. A Wi-Fi sensor installed inside the bus can be used to estimate the number of passengers who board and alight the bus. The results of the novel method in terms of detection rate demonstrated higher accuracy than previously proposed algorithms. This work also revealed the ability of the algorithm to capture transfer activities in terms of the number of passengers and other transfer characteristics.

This paper focuses on the bus bridging service design in response to an unplanned metro line segment disruption. An integrated optimization model is constructed to address both bus bridging route layouts and bus resource allocation. Diversified routes, e.g., stop-by-stop route, express route, skip-stop route, parallel to the disrupted line segment are applied in the optimized bridging scheme to best respond to the passenger demands. An event-driven simulation-based genetic algorithm is designed to solve the optimization model. The simulation method is introduced to handle the uncertainty of passenger waiting time and the randomness of bus dwelling times at bridging stations and control the bus station capacities and guarantee the feasibility of the bridging schemes. Finally, the effectiveness of our proposed approaches is verified in a case study. Sensitivity analyses explore the impacts of fleet size and route diversity on the bridging performance. The results are instructive for transit agencies to operate bus bridging services.

Reaction delay is an indispensable factor in the operation and control process of drivers in a car-following scenario. Utilizing trajectory data obtained from an instrumented vehicle, this paper proposes a data-driven neural network car-following model incorporating instantaneous reaction delay of the drivers in nonlane-based (or disorderly) traffic systems. Considering instantaneous reaction delay as the time interval between relative speed and acceleration, a model is developed where the lateral descriptor of vehicle interaction (or centerline separation, CS) is found as a significant factor in modeling reaction delays in such disordered systems. Interestingly, reaction delays were found to increase with lateral separation between the interacting vehicles. Modeling results further indicate that the car-following model with instantaneous reaction delay outperformed the model with fixed reaction delay. In addition, the proposed model also showed better performance in terms of trajectory reproducing accuracy when compared with the classic car-following models. The results of this work justify the importance of considering CS in the development of algorithms for future autonomous traffic of disorderly traffic environments.