This study mainly studies the contributing factors on residents’ travel mode choices after the emergence of bike-sharing. In contrast to existing studies, the authors divided the travellers into commuters, students, and other travellers by travel purposes, and analysed their travel mode choice by a mix logit model, respectively. It is found that the factors on residents’ travel mode choices have many similarities and differences. Gender, private car ownership, travel cost, travel distance, and travel time are the common factors for all travellers; economy and comfort preference are the factors that affect commuters and students; commuters and other travellers are affected by age, income, and safety preference. However, occupation and an environmental preference are unique significant factors on commuters; students are affected by owning a bike; and a good understanding of bike-sharing is the only significant factor that affects other travellers. In addition, comfort preference has a significant negative influence on the choice of public transport and bike-sharing for students, while it has a significant positive impact on the choice of a private car for commuters.

Vehicular traffic in most of the developing countries is a mixed traffic flow condition with non-lane-based behaviour in urban roads. Vehicles with different static and dynamic characteristics share a common lane. Signalised intersections provide sequential movements of vehicular traffic from one leg to another leg. Saturation flow is an important parameter for the measurement of the level of service at signalised intersections. The various equations, developed for the estimation of saturation flow, for a mixed traffic flow condition do not consider the non-lane-based traffic conditions. There is a need to establish a model to estimate the saturation flow of a signalised intersection for a non-lane-based vehicular movement in mixed traffic flow conditions. This research work presents the findings of the analysis of saturation flow conducted at signalised intersections in three different cities, Warangal, Raipur, and Calicut, in India. Flow rate and traffic volume data are collected using a video graphic technique and the Transportation Research Record Laboratory direct vehicle count method is used to estimate saturation flow. The developed model shows good predictability in comparison with observed data and can be used to estimate saturation flow in non-lane-based traffic conditions.

Multi-class traffic flow modelling has various approaches several of which have focused on analytical proofs. A key limitation in this field of research is the limited field data applications. This study proposes a speed-gradient-based multi-class second-order model and shows its application to three different road sections, a mid-block section, a section with a bottleneck, and a section with a signal at the end, in Chennai, India. The model captures the congestion formation and dissipation phenomena well and could predict outflow and speed fluctuations generally observed in the field scenarios accurately. The prediction of traffic flow dynamics by the proposed model is also observed to be better when compared with two existing higher-order multi-class models.

Traffic flows of the urban transport system are randomly influenced by many internal/external factors, which bring in a huge challenge to accurately forecasting road conditions. This study combines the CANDECOMP/PARAFAC weighted optimisation and diffusion convolution gated recurrent unit (DCGRU) models to conduct the traffic condition forecasting based on the sparse ride-hailing service data. A data completion method based on the tensor decomposition is modified by adding factor tensor in the regular terms, which contains the characteristics of weekday, time period, road segment. Subsequently, the DCGRU model of multiclass predicting is adopted in the data set to predict the traffic conditions. A numerical experiment is conducted based on the one-month ride-hailing service data, collected around the Nanjing South railway station. The predicting results indicate that the method in this study outperforms other traditional models in different tested traffic conditions.

Transportation engineering, as a practical engineering discipline, relies heavily on the accurate calibration of importation parameters from field data. In the real world, most transport relations possess inherent non-linearity. Two prevailing methods for handling non-linear regression are the non-linear least-squares method (LSM) with an iterative solution, and linearisation for the non-linear regression function. The second method applies a linear regression method to solve the non-linear regression problem but requires a data transformation of the observations from variant coordinates, and the objective function is suspected to be changed accordingly. This work describes the authors’ investigation into the problem of non-linear regression through two illustrative examples, the calibration of three non-linear (either exponential or logarithmic) single-regime models for fundamental diagram and the regression of non-linear (power) bunker-consumption model, by applying the weighted LSM (WLSM) and the ordinary LSM to calibrate. It is found that linearising the regression model leads to deviations, and the data transformation can create even more concern with the WLSM because the weights can be redistributed after the data transformation. A further investigation into the linear regression and the non-linear regression gives more suggestions on the choice of regression method.

Bandwidth-based models are widely used for traffic signal coordination. Maxband-based integer linear programming is the most popular and practical method of obtaining the maximum bandwidth solution and researchers have been making continuous improvements to it for decades. However, the impacts of the parameters of the model determine (i) whether the parameter needs to be optimised and (ii) whether the optimisation results are strictly optimal, and have not yet been well explained. To analyse specifically the impacts of the relevant parameters on bandwidth, this study focuses on two adjacent intersections to establish a general mathematical description, including the parameters cycle, split, travel time, phase sequence, and offset, and further characterises the bandwidth versus relative offset relationship. The authors analyse the bandwidth solutions obtained with different relative offsets and the effects of parameters on the through-band. The results of the proposed model provide a reasonable interpretation of parameter optimisation strategies in bandwidth-based models, reveal the possibility for better offset transitioning, and promote awareness of the issues involved in applying Maxband-based models.

Cooperative adaptive cruise control (CACC) has shown great potential in improving freeway capacity. Although the benefit of CACC is obvious, its potential side effects are not yet well studied. One of the major factors that have been overlooked is merging behaviour. A driving simulator study has been recently conducted at the Federal Highway Administration of the United States and reveals that there is unique driving behaviour when joining and leaving a CACC platoon. Unlike the conventional merging model which is a passive decision action, merging into a CACC platoon is a proactive action. Without simulating this unique behaviour, any simulation evaluation on CACC is biased. To improve the validity of future CACC simulation evaluation, this research constructs a merging model. The model consists of two parts: the longitudinal trajectory model and the merging duration prediction model. The model was constructed for both human manual driver and CACC automated controller. The evaluation of the proposed model shows that the model is 96.5% accurate in terms of merging duration prediction and 95.2% accurate in terms of speed prediction.

Efficient operation technique has always been one of the common goals for researches both in automobile industrial and academic areas. With the great progress of automobile technology, hybrid electric vehicle/plug-in hybrid electric vehicle (HEV/PHEV) has already become the main achievement of transportation electrification, due to its excellent fuel-saving performance. Energy management strategy (EMS) is an important link during the HEV/PHEV design procedure, which can govern the energy flow between the fuel tank and the electric energy storage by solving the energy distribution problem. As the continuous development of intelligent connected vehicle technology, designing an efficient EMS with vehicle to infrastructure/vehicle to vehicle (V2I/V2V) information for HEV/PHEV is still a challenge and hot issue. This study presents a deep review of the various EMSs for both conventional HEV/PHEV and that using V2I/V2V information, providing a thorough survey of EMSs using different methodologies. In terms of single-vehicle and multi-vehicle scenarios, the EMSs for HEV/PHEV under intelligent transport system is in-depth reviewed. Finally, the challenges for future research are also identified. This study could provide a comprehensive reference for researchers in field of HEV/PHEV.

Automatic vehicle license plate recognition (AVLPR) aims at extracting the region that contains the information of vehicle license number out of an image data and then identifying the characters apart from the human intervention. This study proposed an effective AVLPR framework where detection, segmentation and recognition of various shaped license plates have been focused. For both proper visual perception and computational processing, a pre-processing technique including grey-scaling conversion combined with close arithmetic-based dilation has been defined. Both vertical and horizontal edge densities have been enumerated by kernel matrices which enable robustness in detecting various shaped and sized license plates. For better detection of candidate region, the vertical and horizontal energy mapping features combined with Gaussian smoothing filter have been used to enable detection of license plates from both high definition and lower resolution images under various illumination conditions and crowded background. For ensuring a better character segmentation rate which is the prerequisite for higher recognition rate, a blob assessment method has been defined integrated with connected component analysis. With 400 vehicle images having varying pixels, the proposed algorithm achieves 96.5, 95.6 and 94.4% accuracy, respectively, in identifying, segmenting and recognising the plate number.

Unmanned aerial vehicle (UAV) is at the heart of modern traffic sensing research due to its advantages of low cost, high flexibility, and wide view range over traditional traffic sensors. Recently, increasing efforts in UAV-based traffic sensing have been made, and great progress has been achieved on the estimation of aggregated macroscopic traffic parameters. Compared to aggregated macroscopic traffic data, there has been extensive attention on higher-resolution traffic data such as microscopic traffic parameters and lane-level macroscopic traffic parameters since they can help deeply understand traffic patterns and individual vehicle behaviours. However, little existing research can automatically estimate microscopic traffic parameters and lane-level macroscopic traffic parameters using UAV videos with a moving background. In this study, an advanced framework is proposed to bridge the gap. Specifically, three functional modules consisting of multiple processing streams and the interconnections among them are carefully designed with the consideration of UAV video features and traffic flow characteristics. Experimental results on real-world UAV video data demonstrate promising performances of the framework in microscopic and lane-level macroscopic traffic parameters estimation. This research pushes off the boundaries of the applicability of UAVs and has an enormous potential to support advanced traffic sensing and management.

Detection of the traffic light is a key function of the automatic driving system for urban traffic. Considering the characteristics of classical and self-learning algorithms, a fusion logic is proposed to make up the shortcoming of learning algorithms by combining the known knowledge with the learning features to detect the red and yellow–green traffic light without turn indicator. The relationship of detection performance among different detectors is established analytically. Then the improvement of detection performance by fusion is analysed theoretically and optimised numerically. According to the analysis results, the hybrid detector is designed by using the colour information in hue-saturation-intensity to extract the candidate region, the hog feature to identify the shape information of traffic light classified by a support vector machine, and a comparatively simple convolutional neural network (CNN) with the classical AlexNet structure to act as the self-learned detector. The effectiveness of the hybrid method is validated by several comparative tests with single CNN detectors and other fusion methods on the training dataset, and the extensibility to new application conditions is evaluated by vehicle tests.

Real-time, on-demand mobility systems have gradually revolutionised the transportation means. However, they continue to exhibit problems on inadequate vehicles at peak times. The popularity of ‘sharing’ may ultimately solve such problems as more passengers are served over time, particularly in high-demand (high-density) locations, thereby realising efficient, comfortable, and environmentally friendly transportation. While, existing sharing methods only arrange each order based on current information and do not apply subsequently received information to pursue more optimal route arrangements. Their research explicitly improves large-scale vehicle sharing methods using subsequent information and proposes the concept of a ‘wait time threshold’ for a vehicle, to manage the constraint contradictions in this process. Based on a representative high-demand case of serving all inbound and outbound passengers at Shenzhen Bao’ an International Airport, a system with consideration of subsequent information provides significant improvements comparing to a system without it. The improvement performance varies with dates under different demand scenarios, high demand indicating a more optimistic influence. Therefore, having such a city-scale sharing model makes it possible to provide decision support to the transportation management department, which encourages to establish a low carbon city.

Vehicle detection and distance estimation are critical components of driver assistance system and self-driving system, and considerable different frameworks have been investigated such as radar, laser and camera-based. Among them, camera-based vehicle detection and distance estimation have an obvious advantage over other systems in that it needs lower cost. However, existing camera-based methods are not robust enough under complex driving scenes. In this work, an end-to-end deep convolutional neural network framework is proposed to jointly detect vehicles and estimate vehicle distance efficiently. Specifically, a monocular depth estimation method is designed to transform the RGB appearance information into depth modality information. Then the vehicle detection module takes the RGB and depth image as inputs to improve the detection performance. Finally, the distance estimation module employs the detection results and the estimated depth information to predict the distance more precisely. The whole network can be trained in an end-to-end manner with the multi-task loss function. The proposed framework is evaluated on the public vehicle detection benchmark KITTI to show the effectiveness of the proposed framework. Moreover, the performance of three proposed sub-modules are also analysed separately to give a more comprehensive evaluation of the designed framework.

In this study, a motion control algorithm based on deep imitation reinforcement learning is proposed for the unmanned underwater vehicles (UUVs). The algorithm is called imitation learning (IL) twin delay deep deterministic policy gradient (DDPG) (TD3). It combines IL with DDPG (TD3). In order to accelerate the training process of reinforcement learning, the supervised learning method is used in IL for behaviour cloning from the closed-loop control data. The deep reinforcement learning employs actor–critic architecture. The actor part executes the control strategy and the critic part evaluates current control strategy. The training efficiency of IL-TD3 is compared with DDPG and TD3. The simulation results show that the training results of IL-TD3 converge faster and the training process is more stable than both of them, the convergence rate of IL-TD3 algorithm during training is about double that of DDPG and TD3. The control performance via IL-TD3 is superior to PID in UUVs motion control tasks. The average track error of IL-TD3 is reduced by than PID control. The average tracking error under thruster fault is almost the same as under normal condition.

Digital mirrors within vehicles may improve aerodynamics and the field of view. Nevertheless, digital technology may fail. This study investigated the influences of different failures on distraction and behavioural adaptation, measured using glance and driving behaviour, as well as a workload questionnaire. Three failure conditions included a blank (no information), a degraded (hard to extract information), and a frozen (misleading information) display. In a high-fidelity simulator, 30 participants undertook three drives in a UK motorway scenario. During the second drive, the right (offside) digital mirror failed during the instruction to conduct a left–right lane-change manoeuvre, and remained until the end of that drive. Results show that failures led to longer and more glances towards the driver-side mirror, increased variability of speed and lateral position, and heightened workload; however, these distracting and behavioural effects lessened in future drives. Behavioural adaptations in the form of increased rear-view mirror or blind-spot checks could not be established. There were indications that the blanked mirror affected workload less than the other failures.

In recent years, advanced driver assistance systems (ADASs) have been used to improve the safety of vehicles by either providing additional information to the driver or by taking over complete control. The majority of ADASs currently being utilised run entirely on the vehicle, only having access to information provided by the sensors that are onboard the vehicle itself. Part of the next step in the evolution of ADAS is to incorporate information from other offsite sensors or obtain control inputs from infrastructure which can coordinate multiple vehicles simultaneously via a wireless interface. Wireless communication is inherently delayed and prone to dropped packets. This study looks at the effect of transport latencies and dropped packets on an off-site autoregressive steering controller supplying direct steering inputs to a vehicle. A fully non-linear vehicle simulation model is used to test the effect of delaying steering inputs and dropped packets to test the stability of the controller. The study shows that at dropped packet percentages of up to 40% adequate vehicle control is maintained, while transport latencies of up to 100 ms allow for moderately accurate vehicle control.

With the rapid development of artificial intelligence, the study of intelligent transportation is getting more and more attention and vision-based vehicle behaviour analysis has become an active research field. Most existing methods label vehicle behaviours with discrete labels and then use the vehicle trajectories or motion characteristics to train classifiers which identify vehicle behaviours. However, a simple discrete label cannot contain detailed information about the vehicle behaviour. So, inspired by structured learning, the authors design a structured label which is used to characterise the instantaneous behavioural state based on the vehicle image, including behaviour trend and degree simultaneously. A structured convolutional neural networks model is constructed to learn and predict structured representation of transient vehicle behaviour and preliminary experimental results justify the feasibility of vehicle behaviour structural analysis model, but it achieves only 53.3% prediction accuracy. To reduce the risk of overfitting to small-scale training data, the authors further propose an overfitting-preventing deep neural network, which exploits transfer learning and multi-task learning to achieve a much higher prediction accuracy of 91.1%.

This study investigates the robust control problem of a heterogeneous vehicular platoon subject to non-linear and (possibly fast) time-varying uncertainties. The uncertainties are induced by parameter variations and external disturbances. The bound of the uncertainty is described via a continuous function. Firstly, the platoon is modelled as a coupled uncertain dynamic system. To guarantee collision avoidance and compact formation performance, the bidirectional inequality constraints are established for the spacing error between adjacent vehicles. A mathematical transformation scheme is proposed to convert the bounded state into an unbounded one. Then, based on the Udwdia–Kalaba approach and Lyapunov stability theory, a constraint-following robust controller is designed. The controller renders the uniform boundedness and uniform ultimate boundedness performance of the unbounded state, which in turn guarantees the bidirectional restrictions for the spacing error. Moreover, an optimal design scheme for the tunable parameter of this controller is proposed to minimise a comprehensive index involving the system performance and control cost. Finally, numerical simulations are conducted to validate the efficiency of the proposed algorithm.