Travel time is inherently uncertain in urban networks due to volatile traffic flows, signal controls, bus stops and disturbances from pedestrians. An effective way to characterize such uncertainty is by estimating Travel Time Distribution (TTD). However, conventional TTD models are lack of considering the interactions between turning movements within the link. Whereas the phenomenon of the Channelization Section Spillover (CSS) is very common, leading to strong interactions between turning movements. In this study, by incorporating the correlation of turning movements in the TTD model, that is, considering CSS, copula-based link-level and path-level TTD models are built. First, based on the empirical data, the correlations of turning movements are analyzed, and then the applicability of the various copula models are examined. The marginal distribution of each turning movement is described using parametric and nonparametric regression analysis, respectively. Then, the best-fitting copula is determined based on correlation parameters and the goodness-of-fit tests. As a case study, the chosen model is applied in estimating link-level and path-level TTD in an arterial road in Hangzhou, China, and compared with the model that did not consider the CSS before. Both results indicate that the copula-based approach can precisely capture the positively correlated relationship between turning movements during peak hours. Furthermore, higher TTD estimation accuracy demonstrates the significance to consider CSS, particularly in peak hours.

Narrow and closed spaces like high-speed train cabins are at great risk for airborne infectious disease transmission. With the threat of COVID-19 as well as other potential contagious diseases, it is necessary to protect passengers from infection. Except for the traditional preventions such as increasing ventilation or wearing masks, this paper proposes a novel measurement that optimizes passenger-to-car assignment schemes to reduce the infection risk for high-speed railway passengers. First, we estimated the probability of an infected person boarding the train at any station. Once infectors occur, the non-steady-state Wells–Riley equation is used to model the airborne transmission intercar cabin. The expected number of susceptible passengers infected on the train can be calculated, which is the so-called overall infection risk. The model to minimize overall infection risk, as a pure integer quadratic programming problem, is solved by LINGO software and tested on several scenarios compared with the classical sequential and discrete assignment strategies used in China. The results show that the proposed model can reduce 67.6% and 56.8% of the infection risk in the base case compared to the sequential and discrete assignment, respectively. In other scenarios, the reduction lies mostly between 10% and 90%. The optimized assignment scheme suggests that the cotravel itinerary among passengers from high-risk and low-risk areas should be reduced, as well as passengers with long- and short-distance trips. Sensitivity analysis shows that our model works better when the incidence is higher at downstream or low-flow stations. Increasing the number of cars and car service capacity can also improve the optimization effect. Moreover, the model is applicable to other epidemics since it is insensitive to the Wells–Riley equation parameters. The results can provide a guideline for railway operators during the post-COVID-19 and other epidemic periods.

The link dynamic vehicle count is a spatial variable that measures the traffic state of road sections, which reflects the actual traffic demand. This paper presents a hybrid deep learning method that combines the gated recurrent unit (GRU) neural network model with automatic hyperparameter tuning based on Bayesian optimization (BO) and the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) model. There are four steps in this hybrid approach. First, the ICEEMDAN is employed to decompose the link dynamic vehicle count time series data into several intrinsic components. Second, the components are predicted by the GRU model. At the same time, the Bayesian optimization method is utilized to automatically optimize the hyperparameters of the GRU model. Finally, the predicted subcomponents are reconstructed to obtain the final prediction results. The proposed hybrid deep learning method is tested on two roads of Hangzhou, China. Results show that, compared with the 12 benchmark models, the proposed hybrid deep learning model achieves the best performance in link dynamic vehicle count forecasting.

The principle of system resilience is its ability to withstand disruptions and maintain an equilibrium state. In urban network systems, adaptive traffic signal control (ATSC) has been an effective countermeasure to mitigate traffic flow disturbance and improve resilience. This research has explored the usage of a decentralised advantage actor‐critic (a2c) algorithm‐based ATSC in mitigating disruptions, particularly nonrecurring congestion caused by car accidents. A reward function has also been proposed, combining deduced resilience metric, safety indicator time to collision (TTC) and system performance. A virtual simulation environment was created using simulation of urban mobility (SUMO) to facilitate the evaluation of the proposed approach. In the grid simulation environment, an overall 5.8% improvement is achieved, exceeding benchmark algorithms in three metrics, especially performance with a margin of over 5.2%. Robustness against different levels of car accidents are proven as well. Further evaluation is also implemented based on a real‐world case study and demonstrates an improvement of 20.08%, highlighting the correlation of proposed method’s efficiency on the traffic flow rate and road structure.

With the emergence of Physical AI, artificial intelligence is extending beyond screen-based applications to embodied systems that perceive, interact with, and act in the physical world. Unlike traditional AI, Physical AI operates under real-time safety constraints, continuously interacts with dynamic environments, and coexists with humans. While current governance approaches have largely been developed for digital AI systems, they do not explicitly account for embodiment, physical safety, lifecycle adaptation, or human-machine interaction. To bridge this gap, we present a comprehensive survey of Physical AI governance that integrates both scientific and operational perspectives. We first develop a governance framework that unifies foundational governance principles. We then derive a five-stage Physical AI lifecycle and demonstrate how governance can be systematically operationalized across research, design, data, model development, and deployment. By translating governance principles into actionable implementation practices, our paper provides a systematic foundation for developing Physical AI systems that are safe, trustworthy, and socially beneficial while supporting continued scientific and technological advancement.