The in-vehicular networked control system is among the most critical embedded processes. The controller area network (CAN) has prevailed intra-vehicle communication for decades. Meanwhile, requirements of future transportation systems are expected to emphasise the in-vehicle communication complexity, which endangers the reliability/safety of the intelligent navigation. At first, this study reviews the recent solutions proposed to overcome the CAN expanding complexity. Challenges that tomorrow's intelligent vehicles may raise for CAN reliability are investigated. The comprehensive coverage of current research efforts to remove the impact of these challenges is presented. Further, the in-vehicle system reliability of future automated vehicles is also related to the fault diagnosis performances. Hence, different classes of system-level diagnosis strategies are compared relatively to the requirements of automotive embedded networks. Furthermore, to thoroughly cover CAN reliability engineering issues, focus is given to the automotive validation techniques. The hardware in the loop, real-time analysis and computer-aided-design tools intervene in various phases along the in-vehicular network life cycle. Parameters that stand behind the efficiency and accuracy of these techniques in validating the new generation of vehicles are analysed. The authors finally draw some deductive predictions about the future directions related to the reliability of the intelligent transportation system in-vehicular communication.

Over the past few decades, the increasing number of vehicles and imperfect road traffic management have been sources of congestion in cities and reasons for deteriorating health of its inhabitants. With the help of computer simulations, transport engineers optimise and improve the capacity of city streets. However, with an enormous number of possible simulation types, it is difficult to grasp valuable, innovative solutions which are of the greatest value to city citizens. In this work, the authors expose various problems within this area having reviewed and analysed over 130 papers selected out of 1200 works in the field of urban simulations. The study describes the selection process of important papers and highlights characteristics of microsimulations, macrosimulations, computation optimisations and other approaches found in the literature which are especially useful and should be further built on in the future. They present and compare results provided in reviewed works in terms of throughput improvement, queue, waiting and travel time reduction, vehicle speed increase, speed-ups as well as assumed simulation parameters. Finally, they focus on research gaps, such as a small number of works considering crisis simulations, few real-world scale simulations as well as on software architectural changes and low-level optimizations.

The market penetration of intelligent vehicles is a long-term process. In recent years, significant attention has been paid to the influence of technology market penetration rate (MPR) on efficiency at signalised intersections, but little attention has been given to such an influence on safety at non-signalised intersections. In this study, the influence of the intersection collision warning (ICW) system MPR on safety at non-signalised intersections is investigated. The authors built a Matlab-based simulation platform where an ICW algorithm was implemented. The simulation was firstly conducted to verify their models. Then the simulation results of different MPRs were obtained and compared statistically. Collision probability, conflict index, and collision rate were analysed for safety evaluation. The overall results showed that vehicle safety at non-signalised intersections improves with the increase of the ICW system MPR. Without considerations of inappropriateness and otherness of driver reaction to warnings, when the MPR is 20% and all vehicles are connected by vehicle-to-everything, the collision probability, conflict index, and collision rate can be reduced by around 20, 20, and 35%, respectively. The simulation method can establish a mapping relation between the ICW system MPRs and vehicle safety indices at non-signalised intersections.

The study was designed to assess the impacts of cooperative intelligent transport systems (C-ITS) on traffic safety. The aim was also to review assumptions made in some previous studies and verify or update the earlier safety estimates. Seven informative C-ITS services warning about various situations and hazards were selected in the study. The expert estimates provided were grounded on new evidence about driver behaviour, the crash categories created in the European Risk Calculation tool (ERiC), and literature. The C-ITS services were assessed to reduce the number of fatalities and injury crashes. The clearly biggest drop in fatalities was assessed for ‘In-vehicle signage, speed limit’, ‘Weather warning’ being the second most effective. The next effective were assessed to be ‘Warning of emergency braking ahead’ and ‘Road works warning’. The smallest but still positive impacts were assessed for ‘Traffic jam ahead warning’, ‘Car breakdown warning’ and ‘In-vehicle signage, child and pedestrian crossing ahead’. Generally, the effectiveness was assessed to be somewhat smaller for injury crashes than fatalities. The authors conclude that the results are positive both for implementing C-ITS services as such and as supporting measure for the deployment of more intervening ITS and automated driving.

A novel concept of a dual-layer-oriented control strategy for fully automated vehicles’ lane-keeping system is proposed which consists of an inner controller with a modified preview driver model and an outer cooperative copilot controller. With designed weighting function of displacement error based on a hyperbolic tangent and adjustable preview horizon according to road geometry, the inner controller is supposed to track the lane centreline precisely and efficiently through optimal control framework. The outer controller is specially designed for situations where the vehicle may run out of the lane and cause a collision. Only when the vehicle is at high risk of lane-crossing, the outer controller is activated to guide the vehicle back to lane centreline by exerting proper steering command, which is calculated by solving a model predictive control-based constrained optimisation problem with a designed quadratic cost function. Finally, simulation tests based on CarSim-Matlab joint platform are carried out to verify the proposed strategy. Results demonstrate that the modified preview driver model is able to improve path following performance and the dual-layer-oriented control strategy with less computational burden can effectively prevent the vehicle from crossing lane boundaries.

Driver's intention is a self-internal state that represents a commitment to carrying out driving action at the next moment, which could be affected by driver's emotion. Therefore, understanding driver's emotion is an important basis for developing driver intention recognition models. This study aims to gain a better insight of the characteristics of driver intention transition trigged by driver's emotion. The Hidden Markov model was used to develop a driver intention recognition model with the involvement of driver's emotions. Assorted materials including visual, auditory and olfactory stimuli were used to evoke driver's emotions before the driving experiments, as well as keep and increase the emotional level during driving. Real and virtual driving experiments were conducted to collect human-vehicle-environment dynamic data in two-lane roads. The results show that the proposed model can achieve high accuracy and reliability in estimating driver's intention transition with the evolution of driver emotion. Our findings of this study can be used to develop the personalized driving warning system and intelligent human-machine interaction in vehicles. This study would be of great theoretical significance for improving road traffic safety.

This study proposes a deep learning framework for driver identity identification by extracting information from the vehicular controller area network (CAN) bus signals. First, naturalistic driving data of 20 drivers were collected under a fixed testing route with different road types and different traffic conditions. Then, a one-dimensional convolutional neural network was constructed for driver identification, which consists of two convolutional-pooling layers, a fully connected layer, and a SoftMax layer. Model optimisation algorithms were applied to improve accuracy and speed up the training process. Also, the model parameters were optimised by evaluating their influences on the model results. Furthermore, the performance of the proposed algorithm was compared with that of the K-nearest neighbour, support vector machine, multi-layer perceptron, and long short-term memory model. The authors used the score as an evaluation criterion and the identification score of the authors' proposed model reaches 99.10% under 20 testing subjects where the data time window size is one second and the sample data overlap is 80%. The results show that the model's performance is significantly better than the other algorithms, which can effectively identify driver identities with stability and robustness.

The objective of this work is the calibration of a generalised cost function for the bicycle network links to be used in conjunction with assignment methods for uncongested networks, as cyclists are generally much less delayed by traffic congestions with respect to auto-traffic. The calibration's goal is to find a coefficient vector for a linear link-cost function that maximises the overlap between routes obtained with a minimum cost routing of cyclists' demand and the relative cyclists' chosen routes, identified by a map-matching procedure of recorded global positioning system (GPS) traces. The calibration focuses on minimising an objective function through different established evolution-based optimisation algorithms, thus avoiding the generation of route choice sets. Link cost functions are calibrated for a modified Openstreet network of Bologna, Italy, using GPS data from the European cycling challenge and Bella Mossa campaign. Results show an improvement of up to 30% of overlapping routes with respect to pure distance-based routing. It is also demonstrated that the calibrated link-costs are transferable to a different scenario.

Driven by the era of big data and the development of artificial intelligence, potential traffic patterns can be obtained by analysing the numerous data. Metro has become an essential transport infrastructure and the passenger volume provides the basic support for the optimisation of the metro system. Thus, accurate forecasting of the volume is extremely required. In this study, a model for improving the accuracy and stability of metro passenger volume prediction named VMD-TPE-LightGBM (light gradient boosting machine) is proposed. The original dataset is firstly regrouped both in the station and chronological order while the time interval is reset as 10-minute. Time features for extracting the hidden patterns are extracted by analysing the variation tendency of the passenger volume. For enhancing the precision, the variational mode decomposition algorithm is applied to decompose the original data series. Then each of the modes is regarded as the input of the LightGBM model, which are optimised by a tuning method named the tree of Parzen estimators and K-fold cross-validation. According to this process, the final forecasting results are acquired by reconstructing the predicted modes. The experimental results demonstrate that the proposed model performs superior to all the comparisons and has an impressive effect on short-term metro passenger volume forecasting.

This study presents a radar-based predictive kinetic energy management (PKEM) framework that is applicable as an add-on driver assistance module for a heavy vehicle with an internal combustion engine powertrain. The proposed framework attempts to minimise fuel consumption by estimating the motion of the leading vehicle from radar information and optimising the inputs to the ego vehicle in a predictive manner. The PKEM framework consists of a driver-pedal pre-filter, an interacting multiple model radar-based filter and predictor of traffic object states, and a non-linear model predictive controller. The framework is integrated with established human-driver car-following models representing various driving styles and evaluated over a set of standardised driving cycles. The authors found that the energy-saving benefits can be as much as 23% over the baseline driver-only case with minimal compromises on travel time in urban environments, while the benefits are nearly negligible on the highway cycle. The results included also show the potential trade-offs in accommodating driver-desired inputs.

For attenuating the shimmy phenomenon appeared in an electric vehicle (EV) with independent suspension, this study proposes a finite-time active shimmy stability control method based on an uncertainty estimation observer. Firstly, a four-degree-of-freedoms shimmy model of an EV with independent suspension is constructed. Secondly, in order to deal with the uncertainties in the shimmy model, a finite-time control method via a non-linear uncertain disturbance observer is proposed. The direct Lyapunov function method is used to analyse the global stability of the closed-loop system, and the results show that the system outputs globally converge to zero. Simulation and hardware-in-the-loop simulation test results verify the built shimmy model and show the effectiveness of the designed control method compared with the sliding mode control method.

The control of self-driving cars has received growing attention recently. Although existing research shows promising results in the vehicle control using video from a monocular dash camera, there has been very limited work on directly learning vehicle control from motion-based cues. Such cues are powerful features for visual representations, as they encode the per-pixel movement between two consecutive images, allowing a system to effectively map the features into the control signal. The authors propose a new framework that exploits the use of a motion-based feature known as optical flow extracted from the dash camera and demonstrates that such a feature is effective in significantly improving the accuracy of the control signals. The proposed framework involves two main components. The flow predictor, as a self-supervised deep network, models the underlying scene structure from consecutive frames and generates the optical flow. The controller, as a supervised multi-task deep network, predicts both steer angle and speed. The authors demonstrate that the proposed framework using the optical flow features can effectively predict control signals from a dash camera video. Using the Cityscapes data set, the authors validate that the system prediction has errors as low as 0.0130 rad/s on steer angle and 0.0615 m/s on speed, outperforming existing research.

It is of great interest for autonomous vehicles to predict the trajectory of other vehicles when planning a safe trajectory. To accurately predict the trajectory of the target vehicle, the interaction between vehicles must be considered. Interaction aware prediction methods track the previous trajectories of both the target vehicle and its surrounding vehicles. In this study, the authors consider trajectory prediction as a sequence-to-sequence prediction problem. They tackle this problem with an LSTM encoder–decoder framework. Moreover, they propose two spatial-attention mechanisms to account for the interaction between vehicles, i.e. context attention and lane attention. Spatial-attention mechanisms adopt the selective-attention mechanism of human drivers. They choose context vectors to help the model understand the surrounding environment better and thus improve its prediction accuracy. They evaluate the authors’ methods on the highD data set recorded in German highways with root mean squared error metric. Their experimental results show superior performance to other state-of-the-art methods. Code is available at https://github.com/momo1986/Spatial-attention.

Discretionary lane change is an essential part of connected and automated vehicles (CAVs) on freeway segments. Most existing studies were conducted to optimise the individual decision of discretionary lane change of CAVs. However, the effects of motion states and discretionary lane change decisions from the surrounding vehicles via vehicle to vehicle communication were ignored. To address such a problem, a game theory-based lane change strategy is proposed to collaborate and optimise decisions of discretionary lane change between the CAVs. The payoff functions are formulated for three types of decision games and the payoff of each decision is quantitatively calculated considering the state information of surrounding vehicles. The Nash equilibrium is applied to find the optimal decision set for players. A simulation platform of a CAV environment built is used to conduct the simulation experiments. Various metrics are employed to evaluate the proposed strategy, such as total travel delay, surrogate safety measurement and wave number. The results show that the proposed lane change strategy using a game-theoretical approach can effectively improve traffic operation, safety and oscillations compared to the baseline strategy. The proposed lane change strategy can further benefit the implementations of the CAVs.

The recent growth in real-time, high-capacity ride-sharing has made on-demand public transit (ODPT) a reality. ODPT systems serving passengers using a vehicle fleet that operates with flexible routes, strive to minimise fleet travel distance. Heuristic routing algorithms have been integrated in ODPT systems in order to improve responsiveness. However, route computation time in such algorithms depends on problem complexity and hence increases for large scale problems. Thus, network segmentation techniques that exploit parallel computing have been proposed in order to reduce route computation time. Even though computation time can be reduced using segmentation in existing techniques, it comes at the cost of degradation of route quality due to static demarcation of boundaries and disregarding real road network distances. Thus, this work proposes, a directionality-centric bus transit network segmentation technique that exploits parallel computation capable of computing routes in near real-time while providing high scalability. Additionally, a dynamic fleet allocation algorithm that exploits proximity and flexibility to minimise vehicle detours while maximising fleet utilisation is proposed. Experimental evaluations on a real road network confirm that the proposed method achieves notable speed-up in flexible route computation without compromising route quality compared to a widely used unsupervised learning technique.

This study proposes an effective trajectory planning algorithm based on the quartic Bézier curve and dangerous potential field for automatic vehicles. To generate collision-free trajectories, potential field functions are introduced to evaluate the collision risk of path candidates. However, many studies on artificial potential field approaches primarily focus on static and straight roads, and attach less importance to more complex driving scenarios, such as curving roads. In this study, a novel method based on the Frenet coordinate system is proposed to address such limitations. Moreover, to balance the driving comfortability and the driving safety of the path candidate, the path-planning problem is converted to an optimisation problem, and sequential quadratic programming algorithm is employed to tackle this task. Another merit of this algorithm is the curvature of the generated path is continuous even at the joints of adjacent sub-trajectories by utilising several specific properties of the Bézier curve. Furthermore, to execute the generated trajectory, a framework of velocity generation is proposed while vehicle dynamic constraints are considered. Some typical traffic scenarios, including lane-changing, lane-keeping, and collision avoidance have been designed to verify the performance of the proposed algorithm, and simulations demonstrate the validity of this method.

This study aimed to develop a coach state estimation and prediction system to enhance driving safety. Different from existing vehicle stability estimation studies, the authors propose a hybrid method to estimate and predict the state of a coach in real time. First, the vehicle sideslip angle and yaw rate are estimated by a three-degrees-of-freedom vehicle model combined with an extended Kalman filter (EKF) estimation algorithm. Then, a steering system is established that replaces the front-wheel angle with the steering wheel input torque. Next, a seven-degrees-of-freedom vehicle model analyses the effects of various driving influencing factors on the vehicle sideslip angle and the boundary of the stable region of the phase plane of the vehicle sideslip angle rate, and a boundary value parameter database is obtained. A back propagation neural network (BPNN) model is then established to obtain the boundary function parameter values under multifactor coupling conditions. Furthermore, an online prediction of the steering wheel input torque in a time series is done, and the prediction value is input to the steering system and neural network model. The effectiveness of the proposed method was evaluated via simulations based on MATLAB/Simulink and TruckSim software.

The development of connected and automated vehicle technologies allows for cooperative control of vehicles and traffic signals at intersections. This study aims at exploring the cooperation between traffic signal control and eco-driving control for a connected hybrid electric vehicle (HEV) system. A two-level cooperative control method for integrating traffic signal control, vehicle speed control, and energy management is proposed with the objective of improving both traffic and fuel efficiency for HEVs at isolated intersections. In view of the energy management and recuperation system of HEV, the vehicle energy consumption characteristic is considered in the proposed method. More specifically, at the traffic level, a traffic signal control strategy is designed to minimise the total travel time and fuel consumption of all HEVs using dynamic programming, which explicitly considers the arrival time and recuperation information of vehicles. At the vehicle level, a hierarchical control architecture is applied to optimise the speed trajectories and powertrain of each HEV using model predictive control and adaptive equivalent consumption minimisation strategy. Simulation results show that compared with the fixed-time and cycle-based signal control strategies with eco-driving, the proposed method can significantly reduce the travel time and fuel consumption by up to 27 and 24%, respectively.

Classical artificial potential approach of motion planning is extended for emulating human driving behaviour in two dimensions. Different stimulus parameters including type of ego-vehicle, type of obstacles, relative velocity, relative acceleration, and lane offset are used. All the surrounding vehicles are considered to influence drivers' decisions. No emphasis is laid on vehicle control; instead, an ego vehicle is assumed to reach the desired state. The study is on human-like driving behaviour modelling. The developed motion planning algorithm formulates repulsive and attractive potentials in a data-driven way in contrast to the classical arbitrary formulation. Interaction between the stimulus parameters is explicitly considered by using multivariate cumulative distribution functions. Comparison of two-dimensional (lateral and longitudinal) performance indicators with a baseline model and generative adversarial networks indicate the effectiveness and suitability of the developed motion planning algorithm in the mixed traffic environment.

Vehicle lamps are an important image feature of the night-time vehicle detection algorithm. This study proposes a real-time night-time vehicle detection algorithm based on light attenuation characteristic analysis, which consists of vehicle lamp detection and pairing. For the detection phase, this study proposes an automatic dual-threshold method to quickly extract the attenuation regions around bright objects. This method is highly adaptable and can accurately extract attenuation regions to identify vehicle lamps and strong reflections. For the pairing phase, this study uses the moving trajectories of vehicle lamps to complete preliminary pairing. Then, the vehicle lamp pairing is optimised using the relative vertical and horizontal lamp positions in continuous images. Pixel- and object-based methods were used to verify the experimental vehicle lamp detection results. The results of both verification methods indicate that the proposed algorithm without morphological operation is superior to other algorithms. The accuracy of the pairing phase was >94.6% in scenes with multiple lamps, high-driving speed, high traffic flow, and rain. Finally, the proposed night-time vehicle detection algorithm can achieve a real-time execution speed for images 960 × 720 pixels in size.