1. Vehicle On-board Multi-target Tracking and Motion Prediction Method for Vulnerable Road User

• Research Background

Vulnerable road users, such as pedestrians, riders or other cyclists, are common and diverse in actual traffic scenarios, yet they are susceptible and need to be protected. The detection is the first step to protect them. It needs to further track the positions of the same target between consecutive frames, form a motion trajectory and achieve multi-target tracking, so as to predict their motion intentions subsequently.

• Research Achievements

Aiming at the trajectory prediction of vulnerable road users (VRU, including pedestrians and riders) around self-driving vehicles in autonomous driving environment, an in-depth recurrent neural network trajectory prediction method integrating multiple trajectory prediction factors is proposed, taking into account the motion cues such as the positions and shapes of targets between consecutive frames and the deep convolutional feature of local rectangular frame.

In the multi-target trajectory prediction database of vulnerable road users established by Tsinghua-Daimler Joint Research Center for Sustainable Transportation, different variants of recurrent neural network (RNN / LSTM / GRU), target trajectory predictors and network parameter optimization strategies are comprehensively evaluated, and the effectiveness of the proposed VRU multi-target trajectory prediction method is verified.

• Leading Research Member: XIONG Hui
• Representative Papers

1. Xiong H, Flohr F. B., Wang S, Wang, B, Wang, J, and Li, K. Recurrent Neural Network Architectures for Vulnerable Road User Trajectory Prediction[C].2019 IEEE Intelligent Vehicles Symposium (IV),IEEE, 2019: 171-178.
2. Huang B, Xiong H, Wang J, Xu Q, Li X, and Li K. Detection-level fusion for multi-object perception in dense traffic environment[C].2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI),IEEE, 2017: 411-416.

2. Multi-target Tracking Based on Roadside Camera 

• Research Background

Accurate and robust multi-target tracking method is the premise of cognition and decision-making for intelligent and connected vehicles, and plays an important role in the automatic driving perception technology. And the roadside camera has the prominent advantages of low cost, rich sensing information and wide field of vision, etc. Therefore, the multi-target tracking method based on roadside camera is of great significance for the realization of vehicle-road coordinated perception and high-level automatic driving.

3. Estimation Method of Tire-road Adhesion Coefficient Based on Current On-board Sensors

• Research Background

Tire-road adhesion coefficient is a considerable parameter for describing the current tire-road condition, and also an important prior knowledge of automatic driving system decision-making and control algorithm. It is of great significance to the safety and comfort of automatic driving system. Besides, for the connected traffic system, the estimation results of the adhesion coefficient can be shared, providing early warning for the driver and automatic driving system, thereby greatly enhancing the safety of the whole traffic system.

• Research Achievements

Based on 2D-LuGre tire model and Unscented Kalman Filtering, the real-time estimation of tire-road adhesion coefficient under three driving conditions of braking, acceleration and steering is realized. The observable condition of the established tire system is proved, and the theoretical boundary of dynamic-based estimation method of tire-road adhesion coefficient is put forward. Through the co-simulation of CarSim-MATLAB and Simulink, the effectiveness of the proposed method is preliminarily verified.

• Leading Research Member: LIN Xuewu

1. Research on Intelligent Vehicle Environment Perception Technology Based on Multi-sensor Fusion

• Research Background

Environmental perception is essential in the field of autonomous driving. Due to the advantages and disadvantages of different sensors, multi-sensor fusion has become a hot and challenging research topic in environmental perception. The research on multi-sensor fusion perception method will provide technical support for the construction of intelligent transportation and the development of automatic driving, and will play a significant and positive role in advancing the intelligent development of automobile industry.

• Research Achievements

In order to tackle the difficulty of moving vehicle detection caused by the incomplete target contour in remote sparse point cloud, a moving vehicle detection algorithm based on PE-CPD is proposed. In this approach, the moving targets in the scenario are filtered by the adaptive threshold which changes with the distance of the target and the resolution of the virtual scanning mapping angle. Then, the pose of the target is estimated by using the point cloud registration information combined with the process of motion state calibration and model selection. And it is introduced into the process of moving vehicle detection. Experimental results show that the proposed method performs well in moving vehicle detection, especially in the remote sparse point cloud scenario.

To solve the problem of inaccurate ground segmentation in the case of undulating ground, occlusion of obstacles and sparse point cloud, a point cloud ground modeling algorithm based on hybrid regression technology is initiated. RLWR and gradient filter are used to establish the ground seed model along the scanning radial direction, to obtain the overall ground seed skeleton. Based on this, GPR is used to conduct the ground autonomous modeling along the scanning circumferential direction, and the ground circumferential continuity is used to make up for the modeling error caused by the lack of radial point cloud. The method uses the whole plane continuity to predict the height of the ground point cloud, which improves the robustness of the algorithm to the environment.

At the same time, a pedestrian detection algorithm based on single template matching is proposed. KDE pedestrian clustering algorithm based on hierarchical segmentation and multi-level fusion is used to solve the problem of under-segmentation of neighboring pedestrians. Secondly, the projection map of potential pedestrian clustering is constructed by main plane projection, from which the LARK feature matrix reflecting the target contour is extracted, and the similarity with the template is calculated to determine the target type. The algorithm avoids the need for a large number of training data, and can effectively distinguish pedestrians from other similar objects in the scenario.

• Leading Research Member: LIU Kaiqi
• Representative Papers

1. Liu K, Wang W, Tharmarasa R, et al. Dynamic vehicle detection with sparse point clouds based on PE-CPD[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 20(5): 1964-1977.
2. Liu K, Wang W, Wang J. Pedestrian Detection with Lidar Point Clouds Based on Single Template Matching[J]. Electronics, 2019, 8(7): 780.
3. Liu K, Wang W, Tharmarasa R, et al. Ground surface filtering of 3D point clouds based on hybrid regression technique[J]. IEEE Access, 2019, 7: 23270-23284.
4. Liu K, Wang J. Fast dynamic vehicle detection in road scenarios based on pose estimation with Convex-Hull model[J]. Sensors, 2019, 19(14): 3136.
5. Liu K, Wang W. Pedestrian detection on the slope using multi-layer laser scanner[C]//2017 20th International Conference on Information Fusion (Fusion). IEEE, 2017: 1-7.
6. Liu K, Wang W, Sun Z. Recognition of SAR image based on combined templates[C]//2013 IEEE International Conference on Imaging Systems and Techniques (IST). IEEE, 2013: 284-287.

2. Optimal Sensor Selection and Deployment Planning Based on Multi-objective Optimization

• Research Background

Due to the advantages of high computing efficiency and wide sensing range, Vehicle-Road-Cloud collaborative fusion sensing is an inevitable trend in the development of intelligent connected vehicle. Traditional researches on sensor deployment optimization often only optimize for a single index and variable and lack a reasonable road network model. Therefore, the abstract modeling of multi-sensor perception systems and urban roads, together with sensor-deployment-multi-objective optimization algorithms for perception system, have positive and important significance for the improvement of intelligent connected vehicle perception systems and the full use of resources.

• Leading Research Member: LU Zhengye

3. Target Positioning and Tracking Based on Multi-sensor Information Fusion

• Research Background

Intelligent connected vehicles rely on the data of sensors. In practical applications, different sensors have different working principles and application characteristics, which cannot be accurately applied in some scenarios. Therefore, multi-sensor fusion has become a consensus. The application of sensors in various information environments, to resist various types of noise interference, accurately identify targets, fuse asynchronous heterogeneous information , and obtain invariant features, is the focus of multi-sensor fusion research with great significance.

• Research Achievements

In order to study how to reduce noise interference in multi-sensor fusion perception, improve the accuracy of positioning and tracking of targets after multi-sensor fusion, we need to analyze the source of the noise,improve the fusion algorithm for the noise of the sensor's original data and the noise caused by the unstable reference of the vehicle, and then reduce the positioning and tracking error by comparing the improved algorithm with the unimproved algorithm.

• Leading Research Member: ZHU Shihao

Autonomous vehicles play a positive role in traffic safety, traffic efficiency, environmental protection and energy-saving. With the development of intelligent vehicles, how to interact with these dynamic, random and diversified traffic participants is a potential challenge to realize high-level automatic driving in mixed traffic. Therefore, research on human-like decision-making system of autonomous driving is necessary. However, there remains some challenges as follows: 1) how to identify the intentions of surrounding traffic participants, instead of passively accelerating or braking according to the current state of the surrounding vehicles; 2) how to quantify the driving risks caused by different traffic factors in complex environment; 3) how to plan specific driving strategies and improve the decision-making ability in complex environments.

1. Comprehensive Assessment of Driving Risks Based on Driver-vehicle-road Characteristics

• Research Achievements

Traffic accident statistics have shown the necessity of risk assessment when driving in a dynamic traffic environment. If the risk associated with different traffic elements (i.e., road, environment and vehicles) could be evaluated accurately, potential accidents could be significantly avoided or mitigated. Therefore, in view of the problem that it is difficult to assess the situation of complex road traffic environment, a unified driving environment situation assessment method of subjective/objective risks under the combined action of driver-vehicle-road is proposed, by referring to the equivalent force and field theory.

2. Behavioral Probabilistic Situation Awareness Based on Intention Identification

• Research Achievements

Understanding the dynamic characteristics of surrounding vehicles and estimating the potential risk of mix traffic is of great help to safe and efficient autonomous driving. However, the existing risk assessment methods are insufficient in detecting dangerous situations in advance and tackling the uncertainty of mixed traffic. The research is to evaluate the risk of dynamic driving in consideration of the intention of surrounding vehicles, and output the probabilistic risk map to support the anthropomorphic decision-making of vehicles. Specifically, the intention recognition module can identify the intention possibility of surrounding vehicles, and the human-vehicle-road coupling model can reflect the coupling relationship among drivers, vehicles and roads by analyzing the interactions among them. Finally, a potential risk graph based on multi-vehicle interaction is formed and output to the autonomous decision module for real-time behavior decision under security constraints.

3. Risk Assessment Based on Vehicle Intention and Trajectory Prediction

• Research Background

The traditional risk assessment method only considers the current driving conditions and environmental factors, and lacks consideration of the possible change of surrounding moving objects’ behaviors. If we can accurately identify the intentions of surrounding moving objects and predict their possible trajectories, then the risk assessment model based on the prediction information can more accurately represent the distribution of risk, thereby providing a better support for the decision after the risk assessment.

• Research Achievements

In the acquisition process of prediction information, the method of deep learning is adopted. Compared with the traditional method of directly using vehicle position information and speed information as training data, the research first uses field model to describe this information and obtain a matrix that can reflect the whole information as input. This method improves the accuracy of prediction and adaptability to the scenario. Finally, a risk assessment model is established for the environment based on predicted information.

Leading Research Member: HUANG Heye ，TU Maoran

Representative Papers：

1. Heye Huang, Yang Li, Xunjia Zheng, Jianqiang Wang*, Qing Xu, Sifa Zheng. Objective and Subjective Analysis to Quantify Influence Factors of Driving Risk. In IEEE Intelligent Transportation Systems Conference (ITSC), 2019.
2. Heye Huang, Wenjun Liu, Xunjia Zheng, Qing Xu, Jianqiang Wang*.  Path Planning for Vehicle Obstacle Avoidance Based on Collaborative Perception. In International Conference on Green Intelligent Transportation Systems and Safety (GITSS), 2019.
3. Xunjia Zheng, Heye Huang, Jianqiang Wang*, Xiaocong Zhao, Qing Xu*. Behavioral decision-making model of the intelligent vehicle based on driving risk assessment. Computer-aided Civil and Infrastructure Engineering, 2019.
4. Jianqiang Wang, Xunjia Zheng, Heye Huang. Least Action Principle Followed by Driver’s Decision-making Mechanism [J]. China Journal of Highway and Transport, 2019

Autonomous vehicles play a positive role in traffic safety, traffic efficiency, environmental protection and energy-saving. With the development of intelligent vehicles, how to interact with these dynamic, random and diversified traffic participants is a potential challenge to realize high-level automatic driving in mixed traffic. Therefore, research on human-like decision-making system of autonomous driving is necessary. However, there remains some challenges as follows: 1) how to identify the intentions of surrounding traffic participants, instead of passively accelerating or braking according to the current state of the surrounding vehicles; 2) how to quantify the driving risks caused by different traffic factors in complex environment; 3) how to plan specific driving strategies and improve the decision-making ability in complex environments.

1. Decision-making and Motion Planning for Autonomous Vehicles

• Research Achievements

In view of the problem that the existing intelligent decision-making system is difficult to adapt to the drivers in complex traffic environment and the integration of various driving assistance functions, an anthropomorphic decision-making strategy for the decision-making mechanism of self-learning is proposed based on the concept of "learner, simulator and transcendent". Through the analysis of the relationship of various influence factors of risk assessment, together with the function relationship among traffic elements and characteristics of mechanical driving system, the optimal solution to the action of the mechanical system is found enabling intelligent vehicles to adapt to the human personality without limitations in specific scenarios. A unified intelligent driving decision-making model is established by combining the physical mechanism of the driver's decision-making with the artificial intelligence method, and personalized driving decision-making and multi-objective collaborative control are realized based on the driver's feature recognition.

Leading Research Member: HUANG Heye

Representative Papers:

1. Heye Huang, Yang Li, Xunjia Zheng, Jianqiang Wang*, Qing Xu, Sifa Zheng. Objective and Subjective Analysis to Quantify Influence Factors of Driving Risk. In IEEE Intelligent Transportation Systems Conference (ITSC), 2019.
2. Heye Huang, Wenjun Liu, Xunjia Zheng, Qing Xu, Jianqiang Wang*.  Path Planning for Vehicle Obstacle Avoidance Based on Collaborative Perception. In International Conference on Green Intelligent Transportation Systems and Safety (GITSS), 2019.
3. Xunjia Zheng, Heye Huang, Jianqiang Wang*, Xiaocong Zhao, Qing Xu*. Behavioral decision-making model of the intelligent vehicle based on driving risk assessment. Computer-aided Civil and Infrastructure Engineering, 2019.
4. Jianqiang Wang, Xunjia Zheng, Heye Huang. Least Action Principle Followed by Driver’s Decision-making Mechanism [J]. China Journal of Highway and Transport, 2019.

2. Proactive Decision-Making based on Prediction and Interaction of Traffic Participants in Dynamic Environment

• Research Background

Automated driving technologies have been successfully applied to multiple scenarios. However, while facing the challenges of urban conditions, traditional decision-making methods fall short in guaranteeing safety and efficiency in dynamic environment with other road users. This research targets at proactive decision-making in dynamic environment while considering the prediction and interaction with other traffic participants. It will contribute to the promotion of automated vehicle's intelligence as well as its application in urban area.

• Research Achievements

Based on risk assessment with Safety-field and receding horizon planner, a proactive motion planner is firstly proposed considering intentions of other road users. Comparing with other proactive methods, this planner is more considerate in terms of the uncertainty of intention-awareness as well as other risk factors, generating a more human-like and safe trajectory. Current research continues to consider the interaction among agents. A novel motion planning method is under development based on game theory, and inverse reinforcement learning is also applied to study human’s behavior from naturalistic driving data.

With the algorithms getting more and more complicated, on-board processor develops higher requirement for computing power, energy consumption and volume. The research designs algorithm based on FPGA, taking its advantage of parallel computation, and tests its performance based on on-board application with PCI-e designed on Altera Stratix V.

Leading Research Member: CUI Mingyang

1. Analysis of Intelligent and Connected Vehicle Control under Communication Delay and Packet Drop

• Research Background

V2X technology helps overcome the performance constraints of intelligent vehicle in such aspects as perception, decision and control, which enhances driving safety, efficiency, comfort and economy, etc. Yet, inherent communication delay and packet drop impair the safety and performance of connected vehicle control. Understanding the influences of delay and packet drop on the lateral and longitudinal control of ICV is significant for the implementation of ICV.

• Research Achievements

The influence of delay and packet dropout on connected path-following control and vehicle-following control was studied by the joint-simulation of Simulink and CarSim, and results are as follows. Lateral error is in nonlinear positive correlation with delay and dropout rate. There are thresholds of delay, dropout rate and path frequency beyond which lateral error grows quickly, and within thresholds, lateral errors are not significant. With delays, steady-state speed errors are almost zero, and steady-state distance errors are almost linear. Error in dynamic process is in positive correlation with delay. With dropouts, steady-state distance error grows quickly with speed error after a certain dropout rate, and occasional dropouts in dynamic process lead to chatter of speed.

Field tests is conducted with the help of ICV test bed and LTE-V prototype, and results are as follows. The influence of randomness on control error is comparable to that of delay and packet drop. Large delay and dropout rate deteriorates lateral control performance by reducing response and leading to low frequency oscillation. Tracking error is not in strict positive correlation with delay or dropout rate due to their randomness. Timing of delay or dropout can impose greater influences on performance than their severity, when delay or dropout rate is large.

Leading Research Member: CHANG Xueyang

2.  ICV Control Method Based on Bounded Countermeasure Information

• Research Background

In the future, the control system of intelligent and connected vehicles may suffer from the intrusion of countermeasure information in the internet of vehicles environment, affecting the safety of the vital vehicle control. Therefore, it is of practical significance to study the vehicle control method adapted to the countermeasure information to ensure the safety of vehicles. On the one hand, the traditional control method is not sufficient to guarantee the vehicle control safety in this case. On the other hand, there are still some problems in current researches regarding the control method towards countermeasure information, such as difficult solution under complex time-varying constraints, insufficient control performance optimization, and loose combination with vehicle control, all of which still need to be further studied.

• Research Achievements

In order to study the potential influence of uncertain bounded countermeasure information on vehicle state, the state accessibility of vehicles in this case is analyzed. By using the state feedback control law, the deviation between the actual trajectory and the reference trajectory can be decoupled from the selection of the reference trajectory, and the reachability analysis of the differential reachability set formed by the deviation can be made. Furthermore, by optimizing dynamic feedback control law, it can not only reduce and reshape the differential reachability set, but also leave more room for the control limit used for reference trajectory generation, which alleviates the trade-off between the existing feedback matrix and the control limit to a certain extent.

In order to ensure the safety of vehicle control under bounded countermeasure information, based on Reachability Analysis and with the help of Satisfiability Modulo Theories, the vehicle safety verification is carried out. Under the framework of Receding Horizon Control, an optimal control method suitable for vehicle control in infinite horizon is initiated. On the one hand, the reference trajectory optimal control strategy is obtained, which not only ensures the safety, but also takes into account the control performance. On the other hand, it can improve the applicability of the vehicle control method in more complex traffic scenarios. The simulation results of vehicle-following scenario and lane-changing scenario verify the effectiveness of the control method.

Leading Research Member: LIU Yicong

1. Intersection Signals and Vehicle Coordinated Control Method in Mixed Traffic Environment

• Research Background

As a common spot in cities, intersection is an important factor affecting road traffic efficiency and vehicle fuel economy. When traffic congestion occurs, intersection signals will lead to frequent start-stops and idles of vehicles, resulting in the decline of vehicle fuel economy. The signal light control method with fixed time distribution is difficult to solve the traffic congestion problem. The appearance of intelligent and connected vehicles (ICVs) provides the possibility for further traffic optimization. On the one hand, it can obtain such traffic environment information as signal timing to optimize the driving track of its own; on the other hand, it can acquire information of other vehicles around, and improve the regional traffic efficiency through self-control. However, the popularity of ICVs still needs decades, and the existing research on vehicle coordinated control algorithm for the coexistence of human driving vehicles (HDVs) and connected automatic driving vehicles fails to fully consider the driver's driving information. Based on the mixed traffic environment, this research proposes the queue formation method of ICVs and hierarchical optimization control method of signal lights, which can optimize the traffic efficiency and fuel economy in accordance with different market permeability.

• Research Achievements

In consideration of the driving situation of HDVs, a hybrid queue formation method under mixed traffic is proposed. By combining the ICVs and HDVs into a "1 + N" queue mode, ICVs can actively guide the following HDVs through the intersection. By means of system analysis, the necessary and sufficient conditions to ensure the stability and controllability of the hybrid queue are proved, which provides a theoretical basis for the control of the hybrid queue.

By studying the vehicle-following model of HDVs, a method of determining the optimal following distance and the maximum number of passable vehicles under fixed timing is proposed. For the process of a fixed length queue entering an intersection, an optimal control model of hybrid vehicle queue is presented. The optimal control based on pseudo spectral method is used for trajectory planning, and thus the optimization of fuel economy can be realized.

Drivers need to be guided by signal lights when driving through intersections. Therefore, aiming at the proposed mixed traffic queue model, a reservation signal control method under mixed traffic is designed. This research extends the common "first come, first served" priority allocation model of single ICV, designs a comprehensive priority model allocation method in accordance with queue length in mixed traffic, realizes the dynamic allocation of signal light time under eight phases, and comprehensively optimizes the traffic efficiency of the overall intersection. A simulation platform based on microscopic traffic flow model is built to verify the effectiveness of the hybrid queue algorithm.

Leading Research Member: CHEN Chaoyi

2. Research on Multi-vehicle Coordinated Control System under Unreliable Communication

• Research Background

In recent years, with the development of communication technology, the related connected technology has brought more possibilities for intelligent vehicles to break through the bottlenecks existing in a number of sections including perception, decision-making and control, and improve the performance of the current intelligent vehicles in the aspects of safety, efficiency, comfort and energy saving. However, in the application process of intelligent connected technology, some inherent defects or restrictions of the existing communication technology inevitably affect the application scope and performance, such as network delay, packet loss, quantization error, time-varying topology, channel attenuation and bandwidth limitation. Some factors even seriously threaten the safety of the application in some cases. Therefore, it is necessary to study vehicle control methods and theories under unreliable communication factors, to set the theoretical foundation for safe, reliable and high performance vehicle control under existing communication restrictions, to provide the design ideas for degraded controller in the case of some sudden communication failures, and to improve the safety of intelligent connected vehicle system.

• Research Achievements

The original research in the laboratory considered the Markov Linear Jump System with logarithmic quantitative feedback, derived Riccati inequality in random sense, and calculated the jump control parameters corresponding to different modes using the method of solving linear matrix inequality. Based on the above methods, the study describes the network log quantization feedback control system with finite jump delay by matrix augmentation. The control parameters of stabilization in different delay states are obtained. The control amount of current time stabilization is calculated by using the control amount of the cache limited step historical time. In the actual control, the corresponding control parameters are switched according to the communication delay at the current time, in order to stabilize the system. Through the simulation test of the transverse and longitudinal control scenarios of the ICVs and the comparison with the conventional controller control performance, a feasible controller design idea for ICV control with jump delay is provided and verified in the study.

Leading Research Member: PAN Ji'an

Representative Paper:

1. Pan J, Xu Q, Li K, et al. Controller Design for V2X Application Under Unreliable Feedback Channel[C].2019 IEEE Intelligent Transportation Systems Conference (ITSC). IEEE, 2019: 2496-2502.

3. Common Decision Making and Formation Control Method for Intelligent and Connected Vehicle Groups in Multiple Traffic Scenarios

• Research Background

Traditional control methods for Intelligent and Connected Vehicles (ICVs) usually focus on simple and single traffic scenarios, which makes it difficult to be applied in the real world or switched among different scenarios. This project aims to propose a common formation control method for ICVs that can cover multiple traffic scenarios and improve traffic efficiency and fuel economy.

• Research Achievements

The formation control method for ICVs is proposed based on the concept of coordinated assignment. In the scenario of intelligent highway that contains on-ramps and off-ramps, the geometric topology of the formations is generated according to the number of lanes and vehicles. The Hungarian Algorithm is adopted to solve the vehicle-to-goal assignment problem. Bezier curves are used to generate trajectories for vehicles. The proposed method extends the traffic scenarios of coordinated control of ICVs while improving traffic efficiency, driving safety and fuel economy.

The coordinated control method for ICVs in multi-lane intersections without signal lights is proposed. The conflict model for traffic flow movements in multi-lane intersections is built and the depth-first spanning searching algorithm is chosen to calculate conflict-free passing sequence for vehicle groups. The vehicle regrouping method is proposed to reform sub-formations where vehicles have the same turning expectations. This method can cover the multi-lane intersection scenario and improve traffic efficiency and fuel economy compared with the existing methods.

Leading Research Member: CAI Mengchi

Representative Paper:

1. Cai M, Xu Q, Li K, et al. Multi-lane Formation Assignment and Control for Connected Vehicles[C]//2019 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2019: 1968-1973.

4. Performance Study and Optimal Controller Design in Mixed Traffic System

• Research Background

The traditional technology of ICV is mostly aimed at the condition that all vehicles are operated automatically. In the actual process of promoting ICV technology, there is bound to be a long-term transitional stage, that is, the mixed traffic condition that ICV and HDV coexist. It is of great significance to study the hybrid transportation system, and to design ICV technology for this environment, so as to understand the influence mechanism of ICV on the transportation system, and enhance the positive role of ICV in the transition stage.

• Research Achievements

In order to study the influence mechanism of ICV on the traffic system, the stability, controllability, stabilizability and accessibility of the closed-ring hybrid traffic system are analyzed theoretically. The stabilizability of the ring hybrid traffic system composed of one ICV and several HDVs with homogeneous / heterogeneous dynamic characteristics is proved under certain conditions. The corresponding solutions are obtained through ICV's active influence on improving the upper bound of traffic flow speed. And the internal factors and development potential of ICV in reducing traffic disturbance and improving traffic efficiency under low permeability is revealed.

In order to handle with ICV’s control strategy design under the condition of the limited communication ability in mixed traffic, this research constructs an optimal controller under structural constraints, and uses the principle of sparsity invariance to make the convex relaxation of the original problem, so as to obtain the sub-optimal ICV control strategy. In addition, the traditional ACC, CACC and other strategies focus on the performance of ICV itself. This strategy directly optimizes the overall traffic system by controlling ICV, so that ICV can significantly improve the traffic performance under low permeability.

Leading Research Member: WANG Jiawei

Representative Papers:

1. Zheng Y, Wang J, Li K. Smoothing traffic flow via control of autonomous vehicles[J]. IEEE Internet of Things Journal, 2020.
2. Wang J, Zheng Y, Xu Q, et al. Controllability Analysis and Optimal Controller Synthesis of Mixed Traffic Systems[C]//2019 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2019: 1041-1047.

1. Soft Control Transition Mechanism for Man-Machine Shared Driving

• Research Background

Human error is the main cause of traffic accidents. With the development of intelligent driving technology, the on-board driving system has become the second driving agent which can help human drivers in vehicle control, correctingoccasional misbehaviors of human drivers and thus improving driving safety. In the man-machine shared driving mode, the driving control transition mechanism plays a decisive role in coordinating the man-machine relationship. Hence, to fully utilize the current intelligent driving system for the improvement of driving safety, study on driving control transition could be an impetus at this stage.

• Research Achievements

The proper mode selection as well as the driving authority allocation in the man-machine shared driving highly depends on the reliable identification of the driver's state. To analyze the driver's state in man-machine shared driving, the driving risk field is used as the theoretical framework, and the model structure of the electrostatic field in physics is used to provide a scalable and transferable electric-like risk field model to measure driving risk. Through analyzing the way human drivers respond to the on-road risk, parameters of risk-response characteristics are extracted, allowing the construction of the personalized longitudinal driver model, namely the risk-response driver model (RRDM).

Leading Research Member: ZHAO Xiaocong

Representative Paper：

1. He R, Zhao X, Wang J. Modeling of Driving Risk Response under Human-Vehicle-Road Interaction [J]. China Journal of Highway and Transport, 2020.

1.  Design and Implementation of Cloud Control Simulation Platform

• Research background

The traditional simulation platform of ICV can meet the simulation requirements under the condition of small scale and simple working conditions. However, with the increase of scales and the uncertainty of communication conditions, a new simulation platform is required to be developed to solve the above problems. With the increase of permeability, it is of great significance for the simulation of large-scale and complex conditions in the future.

• Research Achievements

The 1:1 virtual scene is developed, the available vehicle model is built, the controller is designed, and the tracking function is realized. At the level of human-computer interaction, the HOLOLENS device based on Mixed Reality is applied, which enriches the interface form and content.

Leading Research Member: YANG Chunying