Human-like decision-making and planning are crucial for advancing the decision-making level of autonomous driving and increasing acceptance in the autonomous vehicle market, as well as for achieving data closed loop for autonomous driving. However, human-like decision-making and planning methods still face challenges in safety and interpretability, particularly in multi-vehicle interaction scenarios. In light of this, this paper proposes an interpretable human-like decision-making and planning method with Transformer-based deep inverse reinforcement learning. The proposed method employs a Transformer encoder to extract features from the scenario and determine the attention assigned by the ego vehicle to each traffic vehicle, thereby improving the interpretability of planning outcomes. Furthermore, for improved safety in planning, the model is trained on both positive and negative expert demonstrations. The experimental results show that the proposed method enhances model safety while maintaining imitation levels compared to conventional methods. Additionally, the attention allocation results closely align with those of human drivers, indicating the model’s ability to elucidate the importance of each traffic vehicle for decision-making and planning, thereby improving interpretability. Therefore, the proposed method not only ensures high levels of imitation and safety but also enhances interpretability by providing accurate attention allocation results for decision-making and planning. Note to Practitioners—This paper presents a method for enhancing the planning of autonomous vehicles by making it more interpretable and safer. Using Transformer-based deep reinforcement learning, the approach improves clarity by showing how the vehicle prioritizes other traffic participants and learning from both positive and negative examples. This not only enhances safety and decision accuracy but also provides insights into the vehicle’s reasoning process, which is crucial for debugging and increasing user trust. Future work could focus on adapting this method for even more complex driving scenarios.

Autonomous driving decision-making ability is a key technology for intelligent vehicles. Traditional rule-based autonomous driving decision-making methods set corresponding rules in relatively simple single scenarios, which are not suitable for complex and diverse multi-source heterogeneous driving scenarios. The Learning methods based on Spatio-Temporal Graph Neural Networks (ST-GNNs), unlike recurrent networks or convolution networks, can extract the complexity of spatial relationships and temporal dependencies and output driving actions for a period of time in the future. However, most ST-GNNs used for autonomous driving decision-making only consider vehicle nodes, ignoring the heterogeneous node attributes that record rich scenario semantic information, which have not been fully utilized to guide better graph structure learning. But these semantic information are equally crucial for autonomous driving decision-making. To couple with these problems, we propose an Spatio-Temporal Heterogeneous graph Attention Network, namely ST-HANet, to first construct a dynamic heterogeneous graphs to represent a multi-source heterogeneous driving scenarios. Furthermore, for spatial dimension, via mining heterogeneous graph information through heterogeneous graph attention network, and fuse the mined relationship representations with ego vehicle node information through cross attention mechanisms. Next, for temporal dimension, capturing the temporal dependencies of the sequences through the temporal attention mechanism. Moreover, we conducted open-loop and closed-loop testing of the model based on the CARLA simulator, and compared it with state-of-the-art methods. The experimental results demonstrate that this method has higher accuracy and better decision-making performance than existing methods.

Human-like autonomous driving can enhance user acceptance and integration within traffic. In light of this, this paper presents a planning method for the human-like longitudinal trajectory in car-following scenarios with offline sample-based maximum entropy deep inverse reinforcement learning (DIRL). Specifically, the proposed method doesn’t mimic human driving behavior directly. Instead, it uses naturalistic driving data to learn the internal reward function that leads to these driving behaviors. To enhance the capacity for fitting the human reward function, DIRL leverages deep neural networks to replace linear functions used by traditional IRL. However, the long-tail effect of naturalistic driving data makes it challenging for DIRL to capture the reward function in edge scenarios. A simulated dataset covering edge scenarios is collected by employing feature-based inverse reinforcement learning to deal with this challenge. Furthermore, this paper trains a long-term actor network guided by DIRL’s reward network. The long-term actor network significantly reduces the computation cost by three orders of magnitude compared to the reward network-based method, while also avoiding system oscillation in contrast to the traditional one-step actor network. The simulation experiments confirm that the planning results from the proposed method are closer to human drivers’ behavior than the baseline. And, the hardware-in-the-loop experiment results affirm the proposed method’s effectiveness and good real-time performance.Note to Practitioners—Human-like autonomous driving can enhance user acceptance and integration within traffic. This paper proposes a planning method for human-like car-following driving behaviors using offline sample-based deep inverse reinforcement learning (DIRL). Instead of mimicking human trajectories directly, the method learns the internal reward function leading to these driving behaviors. To address challenges posed by naturalistic driving data’s long-tail effect, a simulated dataset covering edge scenarios is collected by employing feature-based inverse reinforcement learning. Additionally, this paper trains a long-term actor network guided by DIRL’s reward network to reduce the computation cost. The simulation experiments confirm that the planning results from the proposed method are closer to human drivers’ behavior than the baseline. And, the hardware-in-the-loop experiment results affirm the proposed method’s effectiveness and good real-time performance.

Improving the performance of autonomous vehicles in adverse weather conditions is vital for the commercialization of such automated systems. Existing synthetic datasets for developing rain-tolerant vision are of limited value. To address this deficiency, a closed environment capable of simulating different degrees of rainfall is constructed. And a new Closed Field Rain dataset is collected in 36 testing cycles. Inspired by the idea that human can infer the content of rainy images directly without removing the raindrops. A new single-image deraining method is proposed, that does not require ground truth images . This method incorporates an image content estimation module applied to predict the scene content representation, and a pixel-wise content attention block used to evaluate the significance of each pixel. After that, an encoder-decoder network is applied to complete the image. On the other hand, it is almost impossible to obtain the ground truth of rainy images because of the dynamic characteristics of real traffic environment. Thus, the model is trained by employing PatchGAN, using a patch-based loss. Using common no-reference and feature point metrics as performance indicators, this paper conducts a comprehensive evaluation on both synthetic and real-world datasets including Closed Field Rain dataset. Results show the effectiveness of our model quantitatively and qualitatively.

Traffic modeling is vitally important to the simulation test of autonomous driving, as on-road testing is not only tedious, inefficient, costly, and often unsafe. The fidelity of the traffic model is thus the key to achieving effective and efficient simulation test results. However, high-fidelity traffic modeling remains a challenge due to the complexity of vehicle movement and the dynamic spatiotemporal interactions among vehicles. In this paper, we propose a novel system-based traffic modeling approach that considers traffic as a whole, or a collection of all vehicles’ movement and their interactions involved simultaneously. In addition, a long short-term memory (LSTM) encoding-decoding framework with a multi-head self-attention mechanism is proposed to capture the temporal dependency of vehicle motion and represent the vehicle-vehicle interaction. Furthermore, some traffic data are collected and used for model training purposes. A metric representing one of the traffic attributes, as the criterion for data screening, is proposed to describe the traffic complexity and disorder, or chaos. Experimental results demonstrate that the proposed model has achieved better performance compared with others from the literature.

Vehicle trajectory prediction is a crucial but intricate problem for lateral driving assistance systems because of driver uncertainty. This article presents a probabilistic vehicle-trajectory prediction method based on a dynamic Bayesian network (DBN) model integrating the driver’s intention, maneuvering behavior, and vehicle dynamics. By selecting a most-relevant-feature vector using joint mutual information, we design a Gaussian mixture model- hidden Markov model and employ the model as a node in the DBN to identify the driver’s intention. Then, a reference path is generated using the road information. The uncertainties of drivers are captured in steering- and longitudinal-control using a stochastic driver model and a Markov chain, respectively. A vehicle dynamic model ensures that the predicted vehicle trajectory adheres to the vehicle dynamics, which improves the prediction accuracy. A particle filter is used to recursively estimate the vehicle trajectory, including position coordinates and the lateral distance from the vehicle center of gravity to the road edge. We evaluate the proposed DBN trajectory prediction method in both lane-keeping and lane-changing scenarios based on a dataset collected from a real-time dynamic driving simulator. Results show that the proposed method can achieve accurate long-term trajectory prediction.

The scenario-based test method is the research hotspot of automated vehicle (AV) validation and verification (V&V), and testing with hazardous scenarios is of important means. An Optimization Searching (OS) method for enhanced generation in hazardous scenarios is proposed in this paper to efficiently explore functional boundary scenarios in a huge logical state space. The method is computationally tractable, and its generated experimental parameters are optimized using past test results. The method includes five essential modules. The Exploration and Exploitation module uses theMulti-arm banditmethod to obtain the greatest sum of the$TTC^{\mathbf {-1}}$(Time To Collision). The Parameter Moving Probability Determination module uses an analytic hierarchy process to ensure that influential parameters are more likely to move. The Step Size Determination module is built withLevy-stepto find a greater number of hazardous scenarios. The Memory Function module is used to avoid repeat experiments that can reduce computing efficiency. The Result Analysis module creates a hazard parameter space for subsequent tests. We tested an ACC (Adaptive Cruise Control) algorithm with a specified logical scenario in the virtual environment built by PreScan. The results showed that the OS method can effectively discover the dangerous range with the tested ACC algorithm, and its test speed can reach more than five times that of an exhaustive algorithm without prior knowledge.

With the common use of lidar sensors on intelligent vehicles, the simulation of lidar in autonomous driving is necessary. This study proposes an innovative lidar modeling method, and introduces solutions for the simulation of the lidar detection function and its physical mechanism. The model consists of geometric and physical models, and can simulate both point clouds and targets. The geometric model addresses the spatial relationship between lidar and the environment. The physical model describes how the lidar detection mechanism may influence detection results. Signal attenuation and unwanted raw data caused by raindrops are the main consideration in the physical model. Characteristics of lidar signal attenuation in different weather conditions are modeled, and a simplified lidar equation is derived for use by lidar users, as opposed to designers. Unwanted raw data are simulated in a stochastic model employing the Monte Carlo method, where raindrop size and distance are sampled. The model is calibrated and validated with real lidar data. The application of the proposed lidar model for an autonomous emergency braking system is introduced.

Deep learning techniques have been widely used in autonomous driving community for the purpose of environment perception. Recently, it starts being adopted for learning end-to-end controllers for complex driving scenarios. However, the complexity and nonlinearity of the network architecture limits its interpretability to understand driving scenarios and judge the importance of certain visual regions in sensory scenes. In this paper, based on the convolutional neural network (CNN), we propose two complementary frameworks to automatically determine the most contributive regions of the input scenes, offering intuitive knowledge of how a trained end-to-end autonomous vehicle controller understands driving scenarios. In the first framework, a feature map-based method is proposed by leveraging current progress in CNN visualization, in which the deconvolution approach recovers the feature maps to extract features that contribute most to understand driving scenes. In the second framework, the importance level of regions is ranked using the error map between the labeled and predicted control inputs generated by occluding different parts of input scenes, thus providing a pixel-wise rank of importance. Test data sets with extracted contributive regions are input to the CNN controller. Then, different CNN controllers trained with the new data sets preprocessed using our proposed frameworks are verified via closed-loop tests. Results show that both the features identified from the first framework and the regions identified from the second framework are of crucial importance to scene understanding for the controller and can significantly affect the performance of CNN controllers.

The exact and reliable understanding of driver behavior characteristics has a large potential contribution to the active safety control system, advanced driver assistance system, and intelligent traffic system. This study presents a driver behavior characteristics identification strategy based on bionic intelligent algorithms. First, a driver behavior data acquisition system is established to be used with test subjects of different driving skill levels for driving data acquisition. The preview optimal curvature model, which directly reflects the driver's behavior characteristics based on its parameters, is then chosen as the ideal driver behavior model and is identified through the genetic algorithm, particle swarm optimization, and back propagation neural network. Moreover, this paper discusses the application of driver behavior characteristics identification in an integrated chassis control system (ICC), which integrates active front steering with electronic stability control. Finally, simulations are performed to verify the identification results and the application in the ICC by PanoSim and MATLAB/Simulink software. Based on the results, the proposed identification strategy precisely characterized the driver behavior. As a result, its application in the ICC improved the path-following ability and vehicle stability performance of the drivers.

This paper presents a neural network (NN) based methodology for power demand prediction and a power distribution strategy for battery/supercapacitor hybrid energy storage systems of pure electric vehicles. To develop an efficient prediction model, driving cycles are first grouped and distinguished as three different driving patterns. For each driving pattern, characteristic parameter data that could better featured driving cycles are extracted effectively and used to train NN. The predictive information combined with its error is subsequently used for power distribution. Then, to deal with different dynamics of battery and supercapacitor systems, a frequency splitter is used and its frequency is further optimized by a particle swarm optimization algorithm to minimize the total cost including battery degradation and system energy for each driving pattern. Based on these efforts, a real-time predictive power management control strategy is finally proposed. To verify its effectiveness, simulation has been conducted to compare with the state-of-the-art control strategy under a speed profile composing of five standard driving cycles. Results show that obviously enhanced performance can be achieved by the proposed control strategy.

Distributed cooperative control requires that every participant shares a consistent view of objectives and the world. Information is periodically disseminated over a noisy time-varying network topology so that all the agents asymptotically converge to a common value. However, the strict global consensus is of excessive resource consumption and not mandatory for the majority of coordination tasks. To better satisfy such quantitative requirements of consensus in the practical multi-agent systems, this paper proposes a real time and distance-driven consensus quantification model especially for C-ITS applications. This model encodes agents' spatial location distribution into their mutual consensus quantification through introducing their inter-distance into consensus calculation. Accordingly, this paper proposes a distance-driven-consensus-based power adaptive control method as a practical use case of the quantitative framework of consensus, by which agents can autonomously optimize the transmit power through balancing the desired consensus benefit and power cost according to the real timely predicted local consensus. We perform extensive numerical calculations to investigate the effectiveness and the applicability of the consensus quantification framework and the power adaptive control method. The results show that the model can effectively capture the real time consensus fluctuation as the multi-agent systems evolve and can provide reliable decision basis to cooperative control, in such way to restrict the consensus extent to a target value and to tradeoff between the anticipated consensus level and the paid cost accordingly.

Although it is well known that driver's intention and driving behavior have great influence on the performance of the advanced driver assistance systems (ADAS), little consideration has been taken in the design of the existing systems. To improve the system performance, in particular, the acceptance and adaption of ADAS to human drivers, it is important to understand human drivers' intention and driving behavior that makes the systems more human-like or personalized for forward collision avoidance (FCA) and autonomous emergency braking (AEB). The research presented in this paper proposed a method to recognize driver's intention and driving behavior based on Gaussian Mixture Model (GMM). A typical testing scenarios of longitudinal braking case was created under a real-time driving simulator with both PanoSim-RT® and dSPACE®. The samples with 36 drivers were used for the testing, and the driving data were collected, analyzed and further employed in driving behavior recognition via a Gaussian mixture model. An optimization method was taken in model parameter identification. The parameters were used in the control design of FCA systems. Compared with existing FCA systems, the proposed personalized systems have demonstrated advantages in both performance and human acceptance.

Deep learning-based approaches have been widely used for training controllers for autonomous vehicles due to their powerful ability to approximate nonlinear functions or policies. However, the training process usually requires large labeled data sets and takes a lot of time. In this paper, we analyze the influences of features on the performance of controllers trained using the convolutional neural networks (CNNs), which gives a guideline of feature selection to reduce computation cost. We collect a large set of data using The Open Racing Car Simulator (TORCS) and classify the image features into three categories (sky-related, roadside-related, and road-related features). We then design two experimental frameworks to investigate the importance of each single feature for training a CNN controller. The first framework uses the training data with all three features included to train a controller, which is then tested with data that has one feature removed to evaluate the feature's effects. The second framework is trained with the data that has one feature excluded, while all three features are included in the test data. Different driving scenarios are selected to test and analyze the trained controllers using the two experimental frameworks. The experiment results show that (1) the road-related features are indispensable for training the controller, (2) the roadside-related features are useful to improve the generalizability of the controller to scenarios with complicated roadside information, and (3) the sky-related features have limited contribution to train an end-to-end autonomous vehicle controller.

Selective assembly has been employed to obtain high-precision assemblies of two mating parts. Most studies only consider the case where machines are reliable and the buffer capacity is infinite. However, unreliable machines and finite buffers are commonly observed in many assembly systems, such as battery pack assemblies and powertrain production lines in the automotive industry. This paper studies a selective assembly system with two component machines, two finite buffers, and one assembly machine. Each component can exhibit different quality behaviors. Bernoulli machine reliability models are assumed. Analytical methods based on a two-level decomposition procedure are developed to evaluate the system performance efficiently. Numerical experiments suggest that the iteration always converges and can deliver high estimation accuracy. Extension to larger systems is also discussed.

Terrain information can significantly impact load power demand, and in turn, on battery life and system efficiency of a hybrid energy storage system (ESS) with battery and supercapacitor. Taking terrain information ahead into consideration for proactive power management is one of the most important ways to improve battery life and overall system efficiency. However, since terrain information is typically available from commercial geographic information systems database, it is by nature inaccurate with uncertainties with respect to the requirements of power management. This is often worsening when combining with commercially low-quality global positioning systems. This paper proposes a novel power management strategy to cope with the inaccuracy and uncertainties of the terrain information with the aim to improve battery life, while maintaining overall system performance. First, the impact of terrain inaccuracy on battery life and system efficiency is analyzed based on two different hybrid ESSs with semiactive topologies. Then, a power management control strategy is developed that actively distributes the power between battery and supercapacitor with adaptation to terrain inaccuracy and uncertainties. The objective of the proposed power management control strategy is to minimize the total cost of the system, including the cost for battery life and energy. Finally, simulation is conducted that has verified the effectiveness of the proposed control strategy.

Battery management system has attracted mounting research attention recently, within which cell equalization plays a key role. Although many research and practices have been devoted to developing various structures of cell equalizers, there are still substantial opportunities for performance improvement yet to investigate. In particular, mathematical modeling and systematic analysis of equalizer systems are limited. In this paper, the performance analysis of the modularized global equalizer system for Lithium-ion battery cell equalization is conducted analytically. Specifically, a mathematical model is developed to emulate the equalization dynamics by considering both charging/discharging and energy loss. Analytical formulas are derived to evaluate the performance of the global equalizer. The introduced model is also compared with the state-of-the-art structures in terms of equalization speed and energy loss. Numerical studies show that the modularized global equalization outperforms others by its substantial reduction on energy loss with similar equalization performance and much less equalizers. In addition, a module segmentation guide is provided to facilitate the equalization system design. Lithium-based battery technology offers performance advantages over traditional battery technologies, which makes it promising in application such as automobiles, portable devices, power grid, etc. To ensure the Lithium-ion batteries working efficiently, reliably and safely, battery equalization systems play a critical rule, especially in large volume battery packs. Various equalization structures have been proposed to balance the state of charge within a string of battery cells. In this paper, we first review the state-of-the-art equalization structures in a unified model representation scheme, and then focus on a modularized global equalization structure with much less equalizers. Based on the mathematical models for performance evaluation, the modularized global equalization system outperforms the state-of-the-art structures in terms of energy loss with similar equalization speed. In addition, we provide a module segmentation guide to determine the number of modules in the system design.

Abstract The Notice of Proposed Rulemaking 13‐22 released by Federal Communications Commission unlocks the Dedicated Short Range Communication (DSRC) spectrum for Wi‐Fi availability, which undoubtedly brings unpredictable effects to the new‐emerging vehicular applications and services. To efficiently harmonize the spectrum operation between DSRC and Wi‐Fi networks, several dynamic spectrum‐sharing schemes are already proposed to improve the spectral efficiency over a limited bandwidth situation and as well to satisfy the ever‐increasing demand for bandwidth resource. Different from most previous literature that mainly focused on the performance analysis of cellular‐network‐centric spectrum sharing, we aim to analyze the performance of the mainstream dynamic spectrum‐sharing schemes specially designed for the coexistence of DSRC and Wi‐Fi networks against various combinations of network parameters through a hybrid network model and performance indicators. We employ the Poisson point process to model a hybrid network where DSRC vehicles and Wi‐Fi devices coexist, and introduce the performance indicators of spectrum efficiency and data rate to assess the utility of different spectrum sharing candidates. Through the presented hybrid model and performance indicators, we collect extensive numerical and simulation results to investigate four typical spectrum allocation schemes for DSRC and Wi‐Fi coexistence, that is non‐sharing scheme, original sharing scheme, and Qualcomm's and Cisco's proposals, respectively. The results show that the dynamic spectrum sharing in the 5.9‐GHz band can significantly raise the performance of Wi‐Fi network without excessively degrading the DSRC system, and especially the Cisco's proposal prefers to protect the DSRC profit while the Qualcomm's draft favors Wi‐Fi exclusively. Copyright © 2016 John Wiley & Sons, Ltd.

For a full x-by-wire electric vehicle, an integrated chassis control method is proposed with optimal tire force distribution. The aim is to achieve better performance for vehicle handling and stability. A quadratic programming method is used in the optimal tire force distribution, in which the desired vehicle body forces, including longitudinal and lateral forces and yaw moment, are optimally distributed to all four wheels in both longitudinal and lateral directions. The desired vehicle body forces typically represent driver's intention and can be inferred from driver's brake or throttle pedal position, and/or steering wheel position based on a reference vehicle model. Further, a sliding mode control based feedback control is employed to ensure vehicle stability and control robustness to various disturbances, including model uncertainties. In order to verify the effectiveness of the proposed control method, some simulation has been conducted under different driving scenarios with emergency braking on split-μ road surface and for evasive lane changing with braking during high speed respectively. The simulation results show that the proposed control method has greatly improved vehicle limit handling performance and safety under the severe driving maneuvers.

Steering torque feedback provides drivers with road feel and haptic feedback on vehicle motion, which are necessary for drivers to properly operate vehicles. Therefore, simulating steering torque feedback with sufficient fidelity is of great significance to systems without actual steering systems, such as driving simulators and steer-by-wire systems. In this paper, a steering torque feedback model is developed and a global sensitivity analysis method is used to identify the influence degree of each model parameter on steering feel properties for on-center performance. The analysis results provide useful guidelines on how to configure steering characteristics to obtain accurate and tunable steering feel.

Information exchange is not easily secured in the emergency cases where the normal telecommunication infrastructure might have been collapsed. When vehicles are moving on a highway, communications between the vehicles and the base stations always result in a high delay that causes a vehicle to fail to verify all the messages received from the neighbors in real time. These situations may result in message losses and even security risks. To address these issues, we propose a scheme that combines the technologies of trusted network connect and multi-secret sharing to securing authentication and verification for ad hoc vehicles SAV4AV, in which a new vehicle is permitted to flexibly join in a platoon through collaborating with t existing vehicles and thereby to accomplish identity authentication and integrity verification. We list several possible attacks and provide a detailed security analysis on how to avoid these threats in SAV4AV. Moreover, we perform extensive simulations to investigate the performance of SAV4AV against various network scenarios with respect to time consumption and network throughput. Copyright © 2014 John Wiley & Sons, Ltd.

Battery management system has attracted mounting research attention recently, within which cell equalization plays a key role. Although many research and practices have been devoted to developing system-level structure of cell equalizers, there are still substantial opportunities for performance improvement yet to investigate. This paper proposes a novel architecture for battery cell equalization, referred to as modularized global equalizer. The mathematical model is developed to emulate the equalization dynamics by considering both charging/discharging and energy loss. Analytical formulas are derived to evaluate the performance of the global equalizer. The proposed method is also compared with the state-of-the-art structures in terms of equalization speed and energy loss.

This paper proposes a novel approach of developing a vision-based forward collision warning system (V-FCW) under an integrated platform with both V-FCW algorithm development and field-test data driven system verification. The developed verification platform provides huge amount of video data collected from field testing under various real driving conditions with labeled ground truth and fast search capability. Under this integrated platform, the V-FCW system can be effectively developed, tested and verified under various real driving conditions and scenarios in a lab environment, and in a repeatable, cost effective and even automatic way to achieve robust, reliable and high performance.

This paper proposes an allocation-based control method for four-wheel independently driven and braked electric vehicle by taking into consideration of actuators dynamic characteristics. The dynamic characteristics of both driving and braking actuators are modeled as physical constraints in order to form a constrained optimization problem. A slacking method is introduced and analyzed in determining the body force distribution through the optimization-based control allocation. The simulation is conducted and the results show that the proposed method works as expected and is valid and effective.

Trajectory planning is one of the key and challenging tasks in autonomous driving. This paper proposes a novel method that dynamically plans trajectories, with the aim to achieve quick and safe reaction to the changing driving environment and optimal balance between vehicle performance and driving comfort. With the proposed method, such complex maneuvers can be decomposed into two sub-maneuvers, i.e., lane change and lane keeping, or their combinations, such that the trajectory planning is generalized and simplified, mainly based on lane change maneuvers. A two fold optimization-based method is proposed for stationary trajectory planning as well as dynamic trajectory planning in the presence of a dynamic traffic environment. Simulation is conducted to demonstrate the efficiency and effectiveness of the proposed method.

This paper first reviewed the driver or driving models used in traffic simulation that were created mainly for generating the required mobility. Some commonly used models and the state-of-the-art technologies were discussed. In order to better study the effect of human driver on the transportation system, this paper introduced several other typical driver models and the related technologies that were developed for the purposes of studies on vehicle dynamics and controls during last decades. These models not only perform the tasks in generating the required mobility with better fidelity and flexibility for traffic simulation, but also provide more insight on the physics of drivers' driving nature and characteristics. Further research was suggested in the modeling of drivers' personal driving characteristics, in particular, drivers' habitual driving styles. These can be very valuable and important in the design and study of the modern transportation systems with vehicle-driver-environment taken into consideration.