Autonomous Driving Technology
This paper proposes a Hidden Markov Model (HMM) based driver perception- decision- manipulation behavior model to simulate the car- following behaviors. The HMM model is used to describe driving intention and simulate the driver's perception process, that is, to obtain the desired vehicle spacing. The prediction module is developed to predict the vehicle trajectory responding to the traffic conditions and driver's psychological status. The prediction module represents driver's decision- making process. The optimization module simulates driver's control actions and adjusts the predicted vehicle spacing to meet the expected vehicle spacing. Driver's perception- decisioncontrol behavior is then simulated through a rolling process of the three proposed sub-modules. The natural driving data were used for empirical analysis and the results indicate the average error of the model is 1.47%, which reflects the effectiveness and accuracy of the model. This paper provides a new perspective for the theoretical research and application of driving behavior modeling.
The traditional autonomous driving vehicles designed the conflict resolution algorithm based on the assumption of the right of way. However, the right of way was not clear in most cases of the mixed traffic of manual and autonomous driving, which will bring great trouble to the decision-making of autonomous vehicles. This paper proposed a conflict resolution optimization method for autonomous vehicles at intersections. The multiobjective optimal control theory was used to plan the speed for the conflicting vehicles, so as to achieve the purpose of cooperative driving. Finally, the simulation experiments of cooperative and non- cooperative conflict resolution were carried out. The results show that the conflict resolution of multi-vehicle cooperative driving can optimize the speed trajectory of vehicles to improve the overall driving efficiency, and the performance of various stakeholders is balanced relatively. Compared with non-cooperative conflict resolution, the time of conflict resolution is shortened, and the average delay per vehicle is reduced by 1~2 s at the intersection, and the average reduction is approximately 5%. The research results can provide a reference for an autonomous cooperative driving when the conflict occurred.
This study proposes a Vehicle Headlights Intention (VHI) recognition model to improve the communication between the Connected and Automated Vehicles (CAV) and Human- driven Vehicles (HV). The VHI recognition model is consist of light perception module, optical data processing module, and VHI recognition module. The light perception module is able to locate and track the HV that sent a light signal through the RedGreen-Blue (RGB) and Hue-Saturation-Value (HSV) color space, the Kanade-Lucas-Tomasi Tracking (KLT), and the vehicle matching algorithm. The optical data processing module calculates the optical radiant flux using optical channel gain algorithm. The VHI recognition module identifies the number of headlight flashing and vehicle driving status based on the Double- layer Hidden Markov Model(DHMM). The experimental results from three typical VHI scenarios indicate that the average accuracy of vehicle headlights perception is 96.8%, and the error of positioning and tracking is within 1 degree. The 1-second VHI recognition rate reaches 96.6%, which enables the driving intention recognition of CAVs and provides the basis for the automated driving decision of the CAV in mixed traffic flow.
With the coexistence shared autonomous vehicles(SAVs) and traditional vehicles, this paper studies how the SAV company optimizes its operating strategies with regards to different operational objectives and the influences on the commuters' travel mode choices. Provided that a certain number of solo commuters drive the traditional vehicles on the highway, and the other commuters without a car make travel mode choices between SAV and transit. This paper optimizes the operating strategies (i.e., the fare and capacity for SAVs) with the objectives changing from the total system cost or the net system benefit to the profit of the SAV company under the fixed demand and the elastic demand, respectively. The equilibrium mode-split flow, the optimal number of SAVs, the total system cost or the net system benefit, the profit for the company for SAVs, and other indicators are obtained. The equilibrium results are verified by a numerical example, and it is found that the monopoly SAV company always charges a higher fare and provides a smaller capacity. At the state of system optimum, the company for SAVs can't produce the positive profit and can only operate with the subsidy from the government.
To study the automatic control of high-speed trains with time-varying exterior disturbances and state saturation, this paper proposes an adaptive iterative learning control algorithm. Based on Lyapunov function, the control law and the parameter of updating law are deduced by considering the state error during the operating process. Then the Lyapunov-like composite energy function is established. The differential negative definiteness and robustness of the proposed function are verified. The proposed adaptive iterative learning control algorithm has been applied to computational simulation and real case study to verify the tracking performance. The results show that the proposed algorithm improves tracking accuracy and convergence speed. It was able to accurately track the desired profile with less iterative times than before.
Shared autonomous vehicles (SAV) are the products of combining autonomous vehicles with shared economy and could provide a new travel mode for people. To explore travelers' choice preferences between SAV with the concern of ride- sharing and private car or public transit, the SAV choice preference survey was implemented, and the potential user characteristics for SAV with the concern of ride-sharing were analyzed. Based on the valid data obtained from the survey, the K-Means clustering method was used to classify historical travel modes, and the characteristics of character and attitude were classified using the factor analysis. In addition, two mixed Logit models in which the parameters of the explanatory variables were subject to different distributions were established for people with and without private cars, respectively, and the results of parameter calibration were compared and analyzed. The research results show that the characteristics of travel modes have extremely significant effects on travelers' mode choice behaviors; the characteristics of character and attitude are significant factors which affect travelers' choice for SAV with the concern of ride-sharing, and their significance is obviously higher than the significance of socio-economic attributes, such as gender, age and so on.
Fuel consumption has a direct relationship with energy conservation and vehicle exhaust emissions. This paper explores the impacts of automated vehicles on the fuel consumption. The manual driving platoon and automated driving platoon were considered as the objective in numerical simulations, which were performed in the environment of traffic oscillations. Additionally, parameter sensitivity analyses of vehicle number, initial speeds, and vehicle-to-vehicle communication delay of automated vehicles were also conducted in simulations. Then, the vehicle-specific power-based evaluation model of fuel consumption was used to calculate the reduction of average fuel consumption rate by automated driving compared with manual driving. Meanwhile, from the perspective of traffic flow stability, the intrinsic relevance between fuel consumption reduction and stability state transition was evaluated. The results show that reduction magnitudes of fuel consumption by automated vehicles have relation with initial speeds of vehicular platoon. Moreover, there is qualitative influence relationship between fuel consumption reduction and traffic flow stability. This means the stable vehicular flow has benefits in significantly improving the reduction magnitude of fuel consumption, which can be used for providing theoretical reference for fuel control strategy, under the background of large-scale automated vehicles.
The development of autonomous driving technology has made it possible to replace traditional manned vehicles with Shared Autonomous Vehicles(SAV) in the future. The SAV' s fleet size problem is studied in the case of using SAV to meet all motorized travel demands of residents. The cell phone signaling data of 3 million users in Shanghai was used, and the motorized travel demands were extracted from it. The impact of actual road conditions in Shanghai was considered. A graph theory model based on the vehicle-sharing network was established to convert the minimum fleet size problem into the minimum path cover problem of directed acyclic graphs, which was solved by the Hopcroft-Karp algorithm. 128 000 SAVs are needed to meet the motorized travel demands of 3 million cell phone users. The impact of maximum scheduling time limit, service area limitation and traffic congestion on fleet size are also studied. Providing a reference for determining the fleet size of SAVs and corresponding infrastructure planning at the city level after the popularization of the autopilot technology.
As a part of future traffic, automatic truck queue is considered as one of earliest automatic driving scenes. To deeply probe into the possible characteristics and causes of mixed traffic system composed of ordinary vehicles and automatic driving trucks, this paper respectively establishes cellular automation (CA) models suitable for describing the driving behavior, and adopts the method of numerical simulation to explore the evolution process of traffic flow state. Research finding that the participation of automatic driving trucks is “a double-edged sword” under the environment of double lane. Little impact is exerted on ordinary vehicles when traffic flow shows low density and automatic driving trucks take up a small proportion; ordinary vehicles face harsh lane changing conditions when traffic flow shows high density and automatic driving trucks take up a high proportion, which leads to low lane changing frequency and failure to obtain a higher speed, thus affecting the traffic efficiency of the whole road system.
With the development of vehicle technology, more and more autonomous vehicles appear on street, which will greatly impact on road traffic. This paper improves the NaSch cellular automata model by taking into account the Gipps safe distance algorithm. The traffic flow mixed by manual and autonomous vehicles are studied using numerical simulation method, and several new conclusions are drawn. First, the highway capacity can be dramatically increased, up to twice of the original capacity value, by adjusting the reaction time of the autonomous driving vehicle. Second, the influence of the reaction time on the highway traffic capacity can be ignored, when the value of the reaction time is reduced to 0.5s. Third, the proportion of the autonomous vehicles in traffic has significant impact on the road capacity and traffic congestion. When the autonomous vehicles is 80%, the highway capacity will be twice of the capacity of the traffic flow consisting of only manual vehicles and the traffic congestion can be reduced up to 50%. Fourth, in the fully autonomous driving traffic flow, increasing the autonomous driving reaction time can reduce the traffic congestion. Especially, when the density is in the range of 30~60 veh/km, the congestion can be reduced 20%, which can be used as an important strategy of traffic congestion mitigation.
Aim to improve the efficiency and flexibility of vehicle active collision avoidance system, a novel path planning methodology for vehicle collision avoidance is proposed in this paper, in which, dynamic traffic conditions, driver intention and vehicle dynamic constraints are comprehensively considered. The proposed road potential field model takes three advantages when comparing with traditional artificial potential field model. Firstly, several local targets are set up to ensure travelling path avoidance of trapping into local minimum potential. Secondly, motion states of road dynamic obstacles are predicted, combined with the grid algorithm, the traditional repulsion field model is modified to ensure that the vehicle motion along the programmed path can effectively prevent collision accidents in maximum degree. Thirdly, the symmetric polynomials method is applied and shortest travel nodes are calculated to smooth the path for meeting the requirement of vehicle dynamic characteristics. The results show that the proposed method can lead vehicle motion away from local minimum potential position. Compared with traditional artificial potential field model, the maximum collision risk value is 55.1% lower by using the improved model calculated trajectory, and the programmed trajectory can comprehensively satisfy the condition of vehicle dynamic restriction and motion performance, and the design results are reasonable.
