Currently, majorities of the robust H _∞ control methods are designed for active suspensions, and seldom take the active control of the in-wheel-motor (IWM) into consideration for IWM driven electric vehicles (EVs). In this study, a robust fault-tolerant H _∞ output feedback control strategy with finite-frequency constraint is proposed to synchronously control the active suspension and dynamic vibration absorber (DVA) for IWM driven EVs. Firstly, a DVA-based electric wheel model is developed, in which the IWM is designed as DVA. Furthermore, the spring-damper parameters of the DVA are matched by using particle swarm optimisation (PSO). Then, the robust fault-tolerant H _∞ output feedback control strategy is developed based on linear matrix inequality, in which the finite-frequency constraint is designed in the resonance frequency range of sprung mass. Finally, simulation results validate that the PSO can effectively optimise the spring-damper parameters of the DVA. The robust fault-tolerant H _∞ output feedback control with finite-frequency constraint can effectively improve the ride comfort and suppress the vertical vibration caused by IWM compared with entire frequency constraint. Meanwhile, the fault-tolerant effectiveness of the proposed method is demonstrated under the actuator faults concerning the actuator force noises and losses.

For a sustainable public transport system, it is important to unveil the spatiotemporal characteristics of ridership and identify the influence mechanisms. Some studies analysed the effects of weather and built environment separately, however, their effects when incorporated remains to be determined. Using smart card data, weather information, and point of interest data from Beijing, the Light Gradient Boosted Machine was employed to investigate the relative importance of weather and built environment variables contributing to daily ridership at the traffic analysis zone level, and investigate the non-linear relationship and interaction effects between them. Weather conditions and built environment contribute 30.22 and 55.83% to ridership fluctuations, respectively. Most variables show complex non-linear and threshold effects on ridership. The interaction effects of weather and weekend/public holiday have a more substantial influence on ridership than weekdays, indicating weather conditions have less impact on regular commuting trips than discretionary trips. The ridership fluctuations in response to changing weather conditions vary with spatial locations. Adverse weather, such as strong wind, high humidity, or heavy rainfall, has a more disruptive impact on leisure-related areas than on residence and office areas. This study can benefit stakeholders in making decisions about optimising public transport networks and scheduling service frequency.

This paper takes the vehicles scheduling of hazardous materials as the research object. First, considered the four objectives of risk minimization, cost minimization, risk equilibrium value minimization, and duration minimization, the vehicles robust scheduling of hazardous materials in multiple distribution centers with risk balance is constructed. Then, uncertain models are transformed into peer-to-peer models through the idea of robust discrete optimization. By introducing the crossover and mutation operators of genetic algorithm into particle swarm algorithm, a hybrid particle swarm algorithm is constructed, and three-stage coding rules are adopted. Finally, case study is exemplified to prove the feasibility of the model and algorithm. The results show that the obtained scheduling scheme for the vehicles of hazardous materials minimizes the risk value and makes the risk distribution more balanced for the government regulatory department, and the vehicle can avoid certain road sections with relatively large risk values. For transportation enterprises, the vehicle scheduling scheme for transporting hazardous materials minimizes the cost and shortens the task duration at the same time. In the case of risk uncertainty, different robust control parameters are introduced to make the vehicles robust scheduling of hazardous materials have different degrees of stability.

Rising energy cost and environment-friendly awareness make energy conservation a key issue for metro operation. Reducing traction energy consumption and improving the utilisation of regenerative braking energy are two efficient solutions to conserve energy. Two optimisation models are proposed in this study for metro systems to minimise the energy consumption by exploring energy-efficient speed profile and optimum timetable. An analytical formulation through finding suitable force coefficients and driving regime-switching points is developed to investigate the optimal speed profile of a single train during its movement along complex rail tracks considering gradients, curves, tunnels and speed limits. Moreover, the synchronisation of accelerating and braking trains is much necessarily implemented when a timetable optimisation model is established for the energy-efficient operation by adjusting the dwell times, running times and turnaround times of multiple trains. A combined particle swarm optimisation and genetic algorithm are addressed to solve the model. The case studies from the actual data of a Beijing Metro line are carried out to verify the feasibility and availability of the proposed approach. Their results show that the net energy consumption of trains along the metro line is reduced by 10.99% by combining the optimal speed profiles and timetable.

For many intelligent transportation applications, traffic congestion prediction is quite essential. If traffic congestion on the road ahead can be accurately and promptly predicted, and routes can be planned reasonably based on the prediction results, traffic congestion can be effectively alleviated. Aiming at the spatio-temporal correlation and evolution characteristics of traffic flow data, the Conv–BiLSTM module comprising a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) is proposed, considering the spatio-temporal features. Firstly, the obtained traffic speed data is folded according to spatio-temporal features, and a three-dimensional matrix is constructed as the input of the prediction network module. After the spatial features are extracted by the CNN, the temporal features and alignment features are extracted by the BiLSTM, followed by which the prediction results are obtained as an output. Prediction and evaluation experiments on the traffic data of the highway in Shanghai prove that the traffic congestion state predicted by this method is largely consistent with the actual state. The results demonstrate that the proposed method has a higher prediction accuracy compared with the conventional and state-of-the-art methods and is an efficient method of traffic congestion prediction.

Machine learning models have been widely adopted for passenger flow prediction in urban metros; however, the authors find machine learning models may underperform under anomalous large passenger flow conditions. In this study, they develop a prediction framework that combines the advantage of complex network models in capturing the collective behaviour of passengers and the advantage of online learning algorithms in characterising rapid changes in real-time data. The proposed method considerably improves the accuracy of passenger flow prediction under anomalous conditions. This study can also serve as an exploration of interdisciplinary methods for transportation research.

Multi-step prediction of long-term traffic speed is an important part of the intelligent transportation system. Traffic speed is affected by temporal features, spatial features, and various environmental features. The prediction of traffic speed considering the above features is a big challenge. This study proposed a multi-step prediction model named embedding graph convolutional long short-term memory network (EGC-LSTM) for urban road network traffic speed prediction which can deal with spatial–temporal correlation and auxiliary features at the same time. Firstly, a graph convolutional network (GCN) for capturing directed graph properties is proposed. Based on the GCN, the LSTM and sequence to sequence model are further applied to realise multi-step prediction considering the spatial–temporal correlation of the traffic network. To improve the performance of the model and obtain the importance of each step in the historical data, the attention mechanism is introduced. Then, one-hot encoding is applied to the category-type auxiliary features. Considering that the dimension becomes larger after the features are one-hot encoded, the dimensions are reduced using embedding. The experiment results prove that the proposed model's performance is better than other models, and the model is interpreted in detail.

The study aims at determining geo-reference locations for dynamic cooperative communications to be established, allowing the generation of dynamic vehicular gaps between the Dangerous Goods Vehicles and their surrounding vehicles, such that they could travel via a tunnel as a platoon and in isolation. This will ensure the safety of other road users in line with check-and-allow procedures at the road tunnel on Trans-European Transport Network, as per ADR regulations. The model is verified for different road layouts approaching a road tunnel in the UK, using varying traffic scenarios to determine if the identified geo-referenced location by the model is at suitable distances for gap generation. The results are compared against the simulated real-world tunnel traffic flow with conventional vehicles involving escorting of Dangerous Goods Vehicles via human operators. The mixed traffic and connected vehicles traffic flow scenarios are simulated for dynamic gap generation and compared against real-world tunnel scenario to analyse the improvements in travel time, queues and congestion.

At present, a number of smart urban mobility solutions exist and furthermore are emerging towards the design of smart cities offering equitable, green and efficient transport. As this concept addresses the needs of all travellers, it is essential to explore the relevance of existing and proposed smart solutions for distinct user categories. Powered-two-wheelers (PTWs) comprise a distinct vehicle category exhibiting specific movement dynamics and characteristics while exhibiting considerable presence in urban areas and increasing ownership trends during the recent years. Thus, PTW needs should be identified and considered in the design of smart cities. This study explores PTW diversion behaviour considering variable message signs (VMS) operation. Relevant data was collected through a stated preference questionnaire survey and diversion propensity was explored through the design of probit models. Survey results indicated that although PTWs believe that VMS fail to address their needs, their attitudes towards them are still positive. Model results exhibited that most of the information elements transmitted via VMS affect PTW diversion behaviour. Other contributory factors, considering PTW diversion included traffic code violation behaviour, rider ‘flexibility’, age, gender etc. Rider sub-populations, considering riding on the pavement or pedestrianised areas and internet use for traffic information, were also explored.

The increasing traffic congestion problem can be solved by an adaptive traffic signal control (ATSC) system as it utilises real-time traffic information to control traffic signals. Recently, deep reinforcement learning (DRL) has shown its potential in solving the traffic signal timing. However, one of the main challenges of DRL is to design a proper reward function and special attention needs for a multi-objective reward design. Since the feedback to the agent depends on the reward function, a proper design of reward function is needed for fast and stable learning. In this study, the authors proposed a new reward architecture called composite reward architecture (CRA) for multi-objective ATSC to optimise multiple objectives. It calculates multiple rewards in parallel for each action and applies the majority voting method to choose the desired action. Since the traffic signal of one intersection affects the adjacent intersections, a new coordination approach is proposed to get the overall smooth traffic flow. The proposed reward architecture CRA is compared with several existing reward functions used in the literature for different traffic scenarios. The new coordinated approach is compared with the non-coordinated approach. The authors demonstrated that the proposed approaches outperform the others concerning waiting time, halting the number of vehicles, and so on.

Performing centrality analysis on nodes from transportation networks are critical to identify important hubs, understand travel decisions, and assess system performances. Current centrality measures are based on topological characteristics of nodes and edges. When applying those measures to large-scale transportation networks, two problems remain unsolved. First, measures are computed based on simplified travel paths, which only include origins and destinations. Due to the lack of information about waypoints of routes, such network representation may not preserve fine level information about waypoints, routes, and traffic flow patterns, resulting in an inaccurate view of centrality. Second, most centrality measures are global measures that rank all nodes in a network, thus failing to detect nodes of regional importance. Therefore, this paper describes an approach that leverages the concept of sequences to identify key waypoints from frequent travel paths and detect community structures of transportation networks. This approach extends two complementary centrality measures to define the role of nodes within communities. The approach has been tested using tracking data of ships in a regional maritime transportation network. Compared to traditional measurement approaches, the proposed approach can construct compact communities, discover prominent waypoints, and add new insight with local centrality measure.

Global positioning system (GPS) trajectory map matching projects GPS coordinates to the road network. Most existing algorithms focus on the geometric and topological relationships of the road network, while did not make full use of the historical road network information and floating car data. In this study, the authors proposed a deep learning enabled vehicle trajectory map-matching method with advanced spatial–temporal analysis (DST-MM). The algorithm mainly focused on the following three aspects: (i) analyse the spatial relevancy from the prospective of geometric analysis, topology analysis and intersection analysis; (ii) to make full use of the historical and real-time data, a deep learning model was conducted to extract the road network and vehicle trajectory features and (iii) establish a speed prediction model and nest it in the temporal analysis structure. It narrows down the path search range through establishing the dynamic candidate graph. Experimental results show that the proposed DST-MM algorithm outperforms the existing algorithms in terms of matching accuracy for low-sampling frequencies GPS data, especially in the central urban area.

In order to ensure a safer and more reliable trajectory during the lane change process, the motion decision algorithm needs to predict the possibility of different interaction behaviours with surrounding vehicles and then makes an advantageous decision. For this purpose, a motion decision method of considering the interaction of surrounding vehicles is proposed. Firstly, this study builds the payoff functions to determine the driving revenue of autonomous driving vehicles. Then, an interactive motion prediction method based on game theory is established to predict the interaction behaviours possibility and future local trajectories of surrounding vehicles. Based on this, a motion decision algorithm based on Nash Q-learning for an autonomous driving vehicle is established. With externalising the main behaviours predicted by the interactive game and the greedy optimisation method, the autonomous vehicle can determine the optimal sequence of actions and take into account the interaction of the surrounding vehicles. Finally, the motion decision in this study is validated by MATLAB in the merging lane scene, and compared with the existing rule-based lane change decision algorithm. The results show that the decision method in this study not only has superiority in safety and efficiency but also can effectively predict the interaction of surrounding vehicles.

Real-time and accurate multi-lane traffic condition forecasting is of great importance to the connected and automated vehicle highway system. However, the majority of existing deep learning based traffic prediction methods focus on pursuing the precision of the methods while neglect to improve the efficiency of the methods. To achieve the high accuracy and high efficiency of multi-lane traffic flow prediction simultaneously, this study proposes a novel combination method via the integration of the clockwork recurrent neural network (CWRNN) and random forest (RF) method, which is RF-CWRNN. To the best of the authors’ knowledge, this is the first time that the CWRNN is introduced to capture the temporal feature of lane-level traffic flow and make traffic speed prediction. Meanwhile, the RF method is employed to measure the temporal relevance of the traffic flow and determine the optimal input time window. To verify the performance of the RF-CWRNN method, the ground-truth data of the expressways in Beijing were utilised to carry out experiments. The results indicate that the RF-CWRNN method is superior to the baseline models in terms of accuracy and robustness. Besides, the proposed method can save plenty of training time compared with the classical long short-term memory neural network.

This work introduces and evaluates a model for predicting driver behaviour, namely turns or proceeding straight, at traffic light intersections from driver three-dimensional gaze data and traffic light recognition. Based on vehicular data, this work relates the traffic light position, the driver's gaze, head movement, and distance from the centre of the traffic light to build a model of driver behaviour. The model can be used to predict the expected driver manoeuvre 3 to 4 s prior to arrival at the intersection. As part of this study, a framework for driving scene understanding based on driver gaze is presented. The outcomes of this study indicate that this deep learning framework for measuring, accumulating and validating different driving actions may be useful in developing models for predicting driver intent before intersections and perhaps in other key-driving situations. Such models are an essential part of advanced driving assistance systems that help drivers in the execution of manoeuvres.

The uncertainties in tire cornering stiffness can degrade the path following the performance of autonomous vehicles, especially in low adhesive conditions, to deal with this problem, a novel multi-model adaptive predictive control is proposed in this study. Firstly, a model predictive path following controller is designed based on a combined model of vehicle dynamics and road-related kinematics relationship. Then, to deal with the model uncertainties, the multiple model adaptive theory is introduced, and the recursive least adaptive law is proposed with its convergence proved by Lyapunov theory. Finally, the multiple-model adaptive law is combined with the proposed model predictive control by a convex polytope of tire cornering stiffness. In this way, the proposed algorithm can be adaptive to the uncertainties of tire cornering stiffness. Simulation results show the effectiveness and robustness of the proposed method to the uncertainties of the tire cornering stiffness resulting in an excellent performance in any road condition without introducing conservativeness.

This study proposes a multi-mode switching longitudinal autonomous driving system based on model predictive control (MPC) with acceleration estimation of proceeding vehicle. A hierarchical control framework composed of three layers is utilised. In the first layer, five longitudinal driving scenarios are defined based on emergency degree. In the second layer, the MPC for longitudinal autonomous driving is designed and serving as the upper controller. Among which a non-linear tracking differentiator is used for acceleration estimation of preceding vehicle. In the third layer, the inverse longitudinal vehicle system dynamic model with strong non-linearity is considered in the lower controller. Proportional–integral–derivative feedback and feedforward control are combined to track the desired acceleration. Simulation and hardware-in-loop test results show that the multi-mode switching longitudinal autonomous driving system is feasible and effective, and has important value for engineering application.

Parking problems caused by a lack of parking spaces have exacerbated traffic congestion and worsened environmental pollution. An analysis of the cruising process for parking can provide new perspectives to reduce cruising. Based on a parking survey conducted in Beijing, the authors collected a large amount of trajectory data of cruising vehicles. Then, fluctuation indexes of trajectories were proposed to analyse travellers’ cruising processes for parking. The spectral clustering method based on a hidden Markov model (HMM) was used to recognise the cruising trajectories. The recognition performance for three-dimensional trajectory data is better. Cruising trajectories for Clusters 1, 2, 3, 4, and 6 have large fluctuations and a weightier effect on road traffic. These groups can be taken as target groups for intelligent parking guidance and recommendations. The recognition accuracies for parking location and parking status increase with increasing intercepted trajectory lengths. 150 m from far to near the desired destination can be used as a threshold of the cruising trajectory length to accurately predict travellers’ parking location and status. These research results can be applied in intelligent parking systems to dynamically predict parking situations, formulate parking guidance schemes and information release strategies, and improve parking efficiency.

Autonomous emergency braking (AEB) is an active safety technology which aims to prevent collisions by operating harsh braking. When AEB is activated on split-μ roads, the yaw moment generated by the asymmetric braking forces will lead to losing control of the vehicle. In this study, a coordinated control scheme of steer-by-wire (SBW) and brake-by-wire (BBW) is developed to solve this problem. An anti-lock braking system is achieved by a sliding mode controller. Besides, two model predictive controllers based on a 7-degree-of-freedom vehicle dynamics model are proposed for different road adhesion conditions. According to the simulation results, the proposed scheme can maintain the stability of the vehicle and achieve a satisfying braking efficiency under various friction differences between the two-side wheels. At last, the experiments are also carried out to verify the effectiveness and the real-time performance of the proposed scheme on a hardware-in-loop bench of BBW and SBW.

This study proposes a dynamic motion planner with trajectory optimisation for automated highway lane-changing driving. Owing to the connected and automated vehicles (CAVs) technology that the real-time traffic information can be obtained, alternative trajectories can be generated to satisfy the vehicle kinematic constraints and avoid many types of potential collisions. An optimal control theory is adopted to select an optimal lane-changing path from the finite path set, and the appropriate acceleration and speed for the execution path are also determined. In order to avoid unnecessary motion re-planning process, this study puts forward a collision-avoidance monitoring algorithm to reduce the time consumption costs of the motion planner. Moreover, an online planning framework based on ‘decision-execution’ is explored. Applying this timeline framework can not only help to evaluate the dynamic planner's online performance, but also reduce the deviation between the online calculation and the actual execution caused by the time consumption. The simulations are performed in PreScan-Simulink platform and the experimental results show that the presented dynamic planner can complete the lane-changing manoeuvre safely and effectively in a high-speed environment.

Vehicular cloud computing (VCC) is a promising technology for the intelligent transport system. It provides a better quality of transport and vehicular services that will increase the safety and comfort of drivers and passengers. A group of vehicles creates VCC and offers vehicular services for its users in the absence of infrastructure to support the system. The vehicles having computing, storage, communication, and sensing devices can share these resources with other vehicles. VCC creates a resource pool by aggregating all the shared resources of nearby vehicles. The major challenge for VCC is to manage the resources from different vehicles in a resource pool and to allocate necessary resources to its user on-demand. The authors propose a semi-Markov decision process based resource allocation method for the VCC system called flexible resource allocation for vehicular cloud system to manage and allocate the resources. The proposed method finds optimal resource allocation strategies for different states of the VCC system and maximises the long-term expected reward under different parameter settings.

With the dense deployment and wide applications of the Internet of Things in railway systems, the location-based security access control scheme is becoming increasingly important. In this study, the received signal strength (RSS) and channel state information (CSI) in railway communications are measured by AR9344 network interface card. Then, based on the measurement data, the authors propose trajectory-based, neural network-based (NN-based) and ray tracing-based (RT-based) localisation algorithms, serving for location-based security access control. Specifically, the trajectory-based algorithm combined with trajectory simulation, movement detection and dynamic time warping algorithms, realises passengers enter/exit pattern detection. The NN-based algorithm leverages back-propagation network (BPN) and constructs training sets with measurement-based RSS and CSI, finishing accurate localisation. Besides, they evaluate the algorithm performance under different layers of BPN. RT-based localisation algorithm combines measurement data and simulation analysis, leveraging simulated-based multiple-input multiple-output received power and delay spread to realise lightweight localisation. After evaluation, the RT-based algorithm can achieve the highest accuracy of localisation, up to 99.9% and is designed to be straightforward for integration with commercial access points and deployment to railway communications.