Delhi is highly plagued by traffic congestion and is notoriously known for its traffic jams. Thus, the question of studying the mode-choice preferences of commuters in Delhi will be integral to travel demand forecasting. The study area poses a challenge in terms of heterogeneity in different types of travel modes available as well as commuters with heterogeneous backgrounds. It offers the typical mix traffic situation prevalent in developing countries, which is cumbersome to model. Eight modes of travel have been considered in this study, which is difficult to come across in previous studies found in the literature. This study proposes to capture mode-choice preferences of commuters by using an adaptive-neuro-fuzzy classifier (ANFC) with linguistic hedges. The proposed mode-choice model will have improved ‘distinguish-ability’ in terms of less overlapping amongst classes, so that the prediction ability is highly improved. Artificial neural network, fuzzy-logic and multinomial-logit models have also been used for analysing mode-choice behaviour of commuters in Delhi. This study is based on microdata collected through household survey conducted in the study area. Results depict that mode-choice model developed by ANFC performs superior to the other models in terms of prediction accuracy.

Smoky vehicle emissions remain a significant contributor in many areas where air quality standards are under threat. The existing smoky vehicle detection methods are inefficiency and with high false alarm rate. This study presents an automatic detection method of smoky vehicles from traffic surveillance video based on vehicle rear detection and multi-feature fusion. In this method, the Vibe background subtraction algorithm is utilised to detect foreground objects, and some rules are used to remove non-vehicle objects. To obtain the key region behind the vehicle rear where the most possible has black smoke in, an improved integral projection method is proposed to detect vehicle rear. To analyse if the key region has black smoke, three groups of representative features are designed and extracted to distinguish smoky vehicles and non-smoke vehicles. More specifically, the features include the artificial features based on deep investigation of smoky vehicles, the statistical features based on grey-level co-occurrence matrix, and the frequency domain features based on discrete wavelet transform (DWT). Finally, support vector machine is used as the classifier for the extracted features. The experimental results show that the proposed method achieves lower false alarm rate than the existing smoke detection methods.

Accurate traffic flow prediction under abnormal conditions, such as accidents, adverse weather, work zones, and holidays, is significant for proactive traffic control. Here, the authors focus on a special challenge of how to develop robust responsive algorithms and prediction models for short-term traffic forecasting in different traffic conditions. To this end, this study presents an ensemble learning algorithm for the short-term traffic flow prediction via the integration of gradient boosting regression trees (GBRT) and the least absolute shrinkage and selection operator (Lasso). Four different model structures whether considering the feature selection are proposed and tested for multi-step-ahead prediction under both normal and abnormal conditions. The results indicate that the proposed multi-model ensemble models are superior to the benchmark algorithms, i.e., support vector regression, and random forests, the GBRT model outperforms the Lasso model under normal traffic conditions, and the Lasso model has a better prediction accuracy under abnormal traffic conditions. In addition, the Lasso model with the feature selection is superior to the full feature model under either normal or abnormal conditions, while the GBRT model is not always better under normal conditions. The proposed integration framework is general and flexible to assemble various traffic prediction algorithms.

Traffic sign detection assists in driving by acquiring the temporal and spatial information of the potential signs for road awareness and safety. The purpose of conducting research on this topic is introduced to a novel and less complex algorithm that works for traffic signs identification, accurately. Initially, the authors estimate the global threshold value using the correlational property of the given image. In order to get red and blue traffic signs, a segmentation algorithm is developed using estimated threshold and morphological operations followed by an enhancement procedure, the net outcome of which is provided the greater number of potential signs. Moreover, remaining regions are filtered in terms of statistical measures using the non-potential regions. Furthermore, detection is performed on the basis of histogram of oriented gradient features by employing the support vector machine (SVM)–K-nearest neighbour (KNN) classifier. The denoising approach with the weighted fusion of KNN and SVM is used in order to improve the performance of the proposed algorithm by reducing the false positive. A recognition phase is performed on the GTSRB data set in order to formulate the feature vector. The proposed method performed the significant recognition with an accuracy rate of 99.32%. It is quite comparable to the existing state-of-the-art techniques.

Vehicle-to-vehicle communication is an important part of the modern intelligent transportation system. Most of the existing vehicle networks are based on central structures, which are prone to single point of failures, and may consequently result in system paralysis. In order to increase the network fault tolerance and maintain the stability of the network system, the authors propose a vehicle network model based on the peer-to-peer (P2P) network. In this model, the base station and the roadside units will no longer be necessary facilities. The vehicles can be used as relays to directly participate in the exchange and transfer of information. The relay nodes (vehicles) are selected based on the degree distribution and the consensus algorithm. The real-time capability, efficiency and cost effectiveness of the proposed P2P model is verified through experimental results.

The real driving cycle (RDC), which aims to reflect the real driving behaviour of vehicles, plays an important role in evaluating the performance or pollution of vehicles. At present, the related researches most focus on developing RDCs of different functions or regions, while the influence factors and the rules of RDC construction are not involved. In this study, through statistical analysis and theoretical analysis of RDC construction process, the influence factors of candidate RDC are explored. These factors include statistical characteristics, common factors number, cluster number, number selection principle, and order principle. A different value of factors means a different candidate RDC and different proximity of RDCs to the real driving data. Through the dynamic time warping index, the proximity of candidate RDCs is calculated, then the factors rules and the optimum RDC are obtained. When the slope is added as one statistical characteristic, the common factors number is set as 5, the cluster number is set as 6, the number selection principle is set as ratio principle, and order principle chooses positive sequence, the candidate RDC is the optimal cycle which is closest to the real vehicles driving conditions.

Currently, payment at most car parking-lots is carried out in the following manner: a ticket machine at entrance prints a ticket with a bar-code for each car entering the parking-lot. This ticket is later scanned at a payment terminal to determine the amount to be paid. The procedure can be automated using face detection and recognition technology. This automation can help with the issue of ticket loss/car theft. This study describes an automated car parking system. The proposed system consists of a camera installed at the entrance/exit of the parking-lot. Frames are continuously acquired by the camera. If there is a detected face, it is registered in the database. When a driver is leaving, the face image is captured again at the exit of parking-lot and compared in the database to conclude the identity. The system at the parking entrance/exit is composed of the following processing modules: (i) image acquisition, (ii) vehicle and face detection, and (iii) feature extraction that also includes a feature comparison/classification module for face recognition. The authors propose suitable algorithms for each module and carry out ad-hoc experiments to check the feasibility of the proposed system.

Considering some technical and economic reasons, it is not easy to directly measure the vehicular moving parameters (such as tyre–road forces and vehicle sideslip angle) in electronic stability programme systems. This study proposes a method to estimate lateral tyre–road forces and vehicle sideslip angle by utilising real-time measurements, based on the unscented Kalman filter. Direct-yaw-moment control can effectively guarantee the stability of vehicle while steering at a high speed. This study proposed a hierarchical control strategy as the solution to the problem of the yaw-moment distribution. The overloop controller is designed to calculate the desired yaw moment based on the estimated lateral tyre–road forces and sideslip angle, using the sliding mode control. The servo-loop controller is designed to optimise the torque distribution using weighted-least-squares method based on the desired yaw moment obtained from the overloop controller. MATLAB/Simulink with Carsim is applied for the simulation experiment, the results demonstrate the effectiveness of the lateral tyre–road force and sideslip angle observer, and the optimal allocation controller could improve the handling stability and energy efficiency dramatically.

As one of the key requirements in the intelligent vehicle, accurate and precise localisation is essential to ensure swift route planning during the drive. In this study, the authors would like to reduce the longitudinal positioning error that remains as a challenge in accurate localisation. To solve this, they propose a data fusion method by integrating information from visual odometry (VO), noisy GPS, and road information obtained from the publicly available digital map with particle filter. The curve of the VO trajectory trail is compared with road segments curve to increase longitudinal accuracy. This method is validated by KITTI dataset, tested with different GPS noise conditions, and the results show improved localisation for both lateral and longitudinal positioning errors.

In this study, an eco-driving strategy is proposed to enhance the fuel efficiency of the connected autonomous vehicle (CAV) in car-following scenarios. First, the longitudinal dynamic model and fuel-consumption model of the vehicle are established. The speed trajectory of the preceding vehicles is obtained via vehicle-to-vehicle/vehicle-to-infrastructure communication function of CAVs, which is used as the reference of the following vehicles. Second, a model predictive controller is presented to optimise fuel consumption of the following vehicle. Finally, simulations in urban and highway driving conditions demonstrate that the proposed controller enables effective tracking of the preceding vehicle in an energy-efficient way. Comparisons between the second and the third following vehicles verify the fuel-saving benefits of the proposed method.

Car-following behaviour is an important problem in terms of road safety, since it represents, alone, almost 70% of road accidents caused by not maintaining a safe braking distance between the moving cars. The inappropriate anticipation of drivers to keep safety distance is the main reason for accidents. In this study, the authors present an artificial intelligence anticipation model for car-following problem based on a fuzzy logic approach. This system will estimate the velocity of the leading vehicle in the near future. Moreover, they have replaced the old methods used in the third step of fuzzy logical approach, the defuzzification, by a novel method based on a metaheuristic algorithm, i.e. Tabu search, in order to adapt effectively to the environment's instability. The results of experiments, conducted using the next generation simulation dataset to validate the proposed model, indicate that the vehicle trajectories simulated based on the new model are in compliance with the actual vehicle trajectories in terms of deviation and estimated velocities. Moreover, they show that the proposed model guarantees road safety in terms of harmonisation between the gap distance and the calculated safety distance.

Motion sickness (MS) usually occurs when travelling in a moving vehicle, and especially experienced by the passengers compared to the driver. The difference in their head movements with respect to the direction of lateral acceleration affects the MS severity level. When experiencing curvature, the passengers normally tilt their head in the same direction as the lateral acceleration, while the driver tilts his/her head against it. This study proposes a correlation model between the lateral acceleration of the vehicle and the head movements of the driver and a passenger via an artificial neural network. Experimental datasets were used in the modelling process. The influence of the number of hidden neurons with respect to the model accuracy has also been investigated. Then, the correlation from the model was expressed as a mathematical equation. This mathematical representation model can be beneficial in the design of vehicle motion control systems in order to mitigate the MS effect.

The first in-depth study on the use of electrocardiogram and electrooculogram for subject-dependent classification in driver sleepiness/fatigue under realistic driving conditions is presented in this work. Since acquisitions in simulated environments may be misleading for sleepiness assessment, performing studies on road are required. For that purpose, the authors present a database resulting from a field driving study performed in the SleepEye project. Based on previous research, supervised machine learning methods are implemented and applied to 16 heart- and 25 eye-based extracted features, mostly related to heart rate variability and blink events, respectively, in order to study the influence of subject dependency in sleepiness classification, using different classifiers and dealing with imbalanced class distributions. Results showed a significantly worse performance in subject-independent classification: a decrease of ∼40 and 20% in the detection rate of the ‘sleepy’ class for two and three classes, respectively. Since physiological signals are the ones that present the most individual characteristics, a subject-independent classification can be even harder to perform. Transfer learning techniques and methods for imbalanced distributions are promising approaches and need further investigation.

Limitation of probe sensor-based technique to collect traffic data in the temporal domain tends to combine it with the fixed sensor. This combination technique involves extra equipment installation difficulties when the road segment becomes non-homogeneous due to traffic operations or lane configurations. Particularly, in developing countries, chaotic traffic pattern and the absence of a fixed sensor in the roadway create the trajectory reconstruction even more challenging. Video-sensor (also, other fixed-sensor) based trajectory reconstruction techniques require sensors for each non-homogeneous road segment and need separate fundamental diagram (FD) to estimate the required traffic parameters. These difficulties can be overcome by using probe sensors. Thus, this research proposes an approach of estimating traffic parameters from probe data considering variable road capacity. Those parameters are used as input to reconstruct the trajectories of all vehicles in the traffic stream (i.e. also non-probe vehicles, in particular) through the application of a lopsided network. The proposed approach improves the percent root mean square error of estimated travel time by around 38% compared with that which uses traffic parameters obtained from FD with respect to ground-truth travel time. This approach is very appropriate, economical and reliable, especially where the required number of fixed sensors is unavailable.

Calibration of car-following models plays an important role not only in traffic simulation but also in the estimation of traffic-related energy consumption. However, the majority of calibration studies only focus on errors on position or speed, whereas these models are used to evaluate environmental parameters associated with road traffic (e.g. pollutant emissions, energy consumption). Then, this study focuses on the ability of Gipps’ car-following model calibrated on trajectory parameters to estimate properly the fuel consumption of a heavy vehicle. First, the shape of one of the most used Goodness-of-Fit function, Theil's inequality coefficient, is investigated. It will be demonstrated that optimal domains are flat and large, and so many combinations of parameters could accurately reproduce the vehicle trajectory. Then, the authors found that Gipps model, calibrated via a multi-objective particle swarm optimisation is relevant to simulate the trajectory of a heavy vehicle, but fuel consumption estimation resulting of these trajectories exhibits large discrepancies. To solve this issue, it is proposed to add the fuel consumption estimation directly in the calibration process as a further dimension. The results show an improvement in the value of energy consumption estimation without increasing too much the error on the trajectory.

In the current literature, little research has been done in the eco-driving of vehicle platoon. To increase the fuel efficiency of vehicle platoon and ensure vehicle safety, this study proposes a switching control strategy for vehicle platoon travelling on a road with varying slopes, where the platooning and eco-driving technologies are used. To improve the fuel efficiency of vehicle platoon, the cooperative look-ahead controller is proposed by formulating a cooperative look-ahead control problem of vehicle platoon based on distributed model predictive control (CLCbDMPC), which is a constrained nonlinear non-convex multi-objective optimisation problem. To solve the CLCbDMPC problem quickly, after the minimum fuel consumption table and its corresponding optimal reduction gear ratio table are constructed and calculated off-line, the improved particle swarm optimisation algorithm with multiple dynamic populations is presented. Furthermore, the safety controller acting as the emergency brake to guarantee vehicle safety is also designed. The switch between the look-ahead controller and the safety controller forms the switching control strategy of vehicle platoon. Simulation results demonstrate, compared with benchmarks, the proposed strategy can significantly save up to 21.88% of fuel for vehicle platoon, and vehicle safety can also be guaranteed.

In the current literature, model-predictive (MP) algorithm is widely applied in autonomous vehicle trajectory planning and control but most of the current studies only apply the linear tyre model, which cannot accurately present the tyre non-linear characteristic. Furthermore, most of these studies separately consider the trajectory planning and trajectory control of the autonomous vehicle and few of them have integrated the trajectory planning and trajectory control together. To fill in above research gaps, this study proposes the integrated trajectory planning and trajectory control method using a non-linear vehicle MP algorithm. To fully utilise the advantages of four-wheel-independent-steering and four-wheel-independent-driving vehicle, the MP algorithm is proposed based on four-wheel dynamics model and non-linear Dugoff tyre model. This study also proposes the mathematical modelling of the static obstacle and dynamic obstacle for the obstacle avoidance manoeuvre of the autonomous vehicle. Finally, simulation results have been presented to show the effectiveness of the proposed control method.

Feature maps of different scales in convolutional neural networks (CNNs) can be regarded as image pyramids. In classification tasks, only the last layer of feature maps is used for making decision. However, in tasks such as road crack detection, the target objects are so small that some original information might lose during the downsampling process in CNN. The authors propose a structure that uses the information that is contained in different layers of feature maps, so all the information could contribute to the classification. This process is managed by adding the weighted values of pixels in corresponding regions of different layers in feature maps and using the sum of these values as the output. The authors apply this structure on a residual network and use it to learn the features of road cracks. Experiments have shown that with the authors’ structure, the network performs better than others at understanding and detecting road cracks.

Autonomous car requires localisation and mapping capabilities to model the surrounding environment and to plan a safe and efficient route for the autonomous driving. Implementation of the localisation and mapping software depends on the type of sensors, map formats, computing environment, and processing algorithms. These dependencies make it difficult to extend, reuse, and maintain localisation and mapping software. The common software platform and software development guideline is a way to address the software dependency issues and to enhance the software reusability, scalability, maintainability, and transferability. Therefore, this study proposes a common software platform and implementation guidelines for localisation and mapping of autonomous cars. The common software platform for localisation and mapping is built on the philosophy of the AUTomotive Open System Architecture (AUTOSAR), which is a standard automotive software platform. The software development methodology of AUTOSAR applies to implementation guidelines of the localisation and mapping software. A graph structure of localisation and mapping is used to design the common software platform. The proposed common software platform and design methodology are evaluated using practical applications of localisation and mapping in a real autonomous car. The results show that the proposed platform and methodology are able to improve the reusability, scalability, maintainability, and transferability of the localisation and mapping software.

This study introduces the concept of space preferences to enhance the operation of recurrent-users car parking systems. On campus, at KFUPM, several parking buildings are available for students, employees, and faculty. Still, substantial amount of time is wasted just looking for a suitable parking space. The authors propose to enrol all users together with their top ranked preferred parking spots and pedestrian exit of choice. Upon presentation of the user ID card, the system retrieves their preferences, matches them to the updated database of vacant spots, then directs the user to their topmost vacant parking spot. The system directs the user to the nearest parking building if no vacancy. Two options have been evaluated to detect parking vacancies: one is based on ultrasonic sensors and the second uses a set of cameras. The sensor-based system was built around an Arduino platform paired with a wireless channel for communication. For the camera-based system, the authors introduce a new set of features mixing both edge and texture information from the parking spot images. A performance analysis of both systems was carried showing that the sensor-based implementation outperforms the camera-based one for the authors’ specific application with an accuracy of 100 and 98%, respectively.