The vehicle-pedestrian encounter at midblock crosswalks in urban centers is inevitable but the challenge to urban transportation planners is in achieving a balance between traffic flow efficiency and pedestrian safety. Vehicles are expected to yield to pedestrians who have a right of way at the midblock unsignalized crosswalks but, failure to yield causes conflicts that at times are fatal. This study investigated the effect of macroscopic factors on the vehicle yielding. Six environmental factors are considered: temporal gap size, number of traffic lanes, number of waiting pedestrians, position of pedestrian (whether on street kerb or median), traffic flow direction and presence (or absence) of monitoring ePolice. Video Data on six observed variables that influenced vehicle yielding was collected from 13 uncontrolled crosswalk locations in Shanghai city in the Peoples Republic of China. A Logit model with a 95.9% accuracy was developed to describe the vehicle yielding behavior. The results showed that gap size and number of traffic lanes had the highest influence on driver yielding decision and that drivers were more likely to yield if ePolice was present. The sensitivity analysis was conducted and appropriate recommendations on improving the pedestrians crossing safety were proposed accordingly.

The loading area effectiveness is a key factor for the estimation of the bus stop capacity. The basic assumption of the existing studies is that the berths of the bus stop are divided clearly. In this paper, a new loading area effectiveness model is proposed for computing the number of effective loading areas of the off-line bus stops that have no clear-cut separation of berths. The key parameters, including the critical overtaking exiting distance gap, critical overtaking entering distance gap, and dwelling space-gap between two adjacent buses were calibrated by field data. A simulation-based approach is applied to solve the model. The proposed model was validated and the sensitivity analysis of main factors were also conducted. The results reveal that the driving behavior plays an important role. On average, 4.1% increase in the loading area effectiveness can be obtained by one meter decrease of the critical overtaking entering space-gap.

This study investigates the optimization of taxi-pooling operations using the M/M/1/m queuing model, aiming to enhance efficiency and revenue for taxi service platforms. Traditional taxi operations face challenges during peak periods, including inefficiency and increased operational costs. Taxi-pooling, by accommodating multiple passengers with similar travel demands, offers a solution that reduces travel costs, operational expenses, and urban congestion. The study develops an optimization model to balance operational costs and passenger waiting times, identifying the utilization rate of taxis as a critical factor in platform revenue. By modeling the taxi-pooling service as a queuing system, we derive mathematical expressions for passenger waiting times and platform revenue under varying conditions. Our findings highlight the importance of optimal vehicle investment strategies and pricing decisions to maximize revenue. The study provides theoretical support for improving taxi-pooling platforms’ efficiency and competitiveness, contributing to better urban transportation solutions.

The rapidly rising crime rates have necessitated advanced and automated security surveillance systems capable of robust, real-time detection, recognition, and prevention of crime. Although traditional surveillance systems have largely been deployed to enhance security and safety, they are inefficient, error-prone, and incapable of effectively processing and generating meaningful insights from the vast quantities of video data they produce. Further, they are adversely affected by extreme weather conditions and subject to human vandalism. The advent of deep learning has significantly transformed earlier automated crime detection, recognition, and prevention by enabling robust extraction and analysis of complex spatial and temporal features from surveillance videos. This research presents a comprehensive review and synthesis of various state-of-the-art deep learning architectures employed in surveillance video-based crime detection and recognition systems, including 3D Convolutional Neural Networks (3D-CNN), Residual Networks (ResNets), Recurrent Neural Networks (RNN), Bidirectional Long- and Short-Term Memory (BiLSTM), Gated Recurrent Units (GRUs), and the integration of attention mechanisms of Soft attention, hard attention, dual attention, and Multi-Head Self-Attention (MHSA). The study critically examines the architectures’ contributions to enhancing detection accuracy, recognition, and the capability to prevent crime. The review further highlights challenges associated with existing systems, including data scarcity, privacy concerns, computational complexity, data class imbalance, and limited real-world adoptability. The paper finally outlines emerging research gaps and future directions in the development of intelligent deep learning-based surveillance crime detection systems.