Automatic driving control method based on driver characteristics and storage medium

Abstract

This application provides an autonomous driving control method and storage medium based on driver characteristics. The autonomous driving control method includes: in response to identifying a vehicle driving scenario, generating scenario data including key indicators, and acquiring time-series data of driver operation of key indicators; a first controller generating a first control strategy for the vehicle based on the scenario data and the time-series data; a second controller generating a second control strategy for the vehicle based on the scenario data and the time-series data; in response to arbitrating the first control strategy and the second control strategy, generating an arbitration strategy for the vehicle; and controlling the driving parameters of the vehicle according to the arbitration strategy. The autonomous driving control method and storage medium based on driver characteristics provided in this application can better achieve human-machine co-driving with the driver, while making more drivers accept the driving system, improving drivers' trust and comfort in the intelligent driving system.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to an autonomous driving control method and storage medium based on driver characteristics. Background Technology

[0002] Current autonomous driving systems are divided into several layers: perception, fusion, decision-making, planning, and control. In areas such as controlling the vehicle along a planned trajectory, most employ optimal control. Current autonomous driving systems can establish autonomous driving models that consider driver characteristics and use different computational methods to analyze how individual driver differences affect the driving model.

[0003] In conceiving and implementing this application, the inventors discovered at least the following problems: current autonomous driving systems cannot adapt to drivers with different styles, and cannot adjust intelligent driving styles in changing road and driving environments, which to some extent reduces the driver's trust in the driver assistance system.

[0004] The preceding description is intended to provide general background information and does not necessarily constitute prior art. Summary of the Invention

[0005] To alleviate the above problems, this application provides an autonomous driving control method and storage medium based on driver characteristics.

[0006] In one aspect, this application provides an automated driving control method based on driver characteristics, comprising:

[0007] In response to identifying vehicle driving scenarios, it generates scenario data including key indicators and acquires time-series data of driver operation of key indicators;

[0008] The first controller generates a first control strategy for the vehicle based on the scenario data and the timing data;

[0009] The second controller generates a second control strategy for the vehicle based on the scenario data and the timing data;

[0010] In response to the arbitration of the first control strategy and the second control strategy, an arbitration strategy for the vehicle is generated;

[0011] The vehicle's driving parameters are controlled according to the arbitration strategy.

[0012] Optionally, the autonomous driving control method includes at least one of the following steps in the process of the first controller generating a first control strategy for the vehicle based on the scene data and the timing data:

[0013] When the vehicle speed is less than a first speed threshold, a kinematic model of the vehicle is established, and the first controller adjusts the key indicators based on the kinematic model to generate a first control strategy for the vehicle.

[0014] When the vehicle speed is greater than the second speed threshold, a dynamic model and a dual-closed-loop control model of the vehicle are established. The first controller adjusts the key indicators based on the dynamic model and the dual-closed-loop control model to generate a first control strategy for the vehicle.

[0015] In response to acquiring personnel information, environmental information, and road information of the vehicle, the first control strategy is adjusted.

[0016] Optionally, the autonomous driving control method, in executing the first controller generating a first control strategy for the vehicle based on the scene data and the timing data, includes:

[0017] Control information is output based on the target path. The control information includes multiple reference trajectory points, vehicle status, environmental information, target information, target speed, and road information.

[0018] Based on the control information, the current vehicle driving scenario is identified, which includes the current position and current speed. The driver's key indicators are obtained, which are selected from at least one of the following: vehicle speed, acceleration, rate of change of acceleration, steering wheel angle, steering wheel speed, and yaw rate: maximum value, minimum value, midpoint value, mean, and variance.

[0019] The step of establishing a kinematic model of the vehicle when the vehicle speed is less than a first speed threshold, and the first controller adjusting the key indicators based on the kinematic model to generate a first control strategy for the vehicle includes:

[0020] Based on the vehicle driving scenario and the key indicators, the kinematic model-based lateral controller generates the target lateral acceleration, lateral acceleration rate of change, target longitudinal acceleration, longitudinal acceleration rate of change, and matrix parameters.

[0021] And / or, the step of establishing a dynamic model and a dual-closed-loop control model for the vehicle when the vehicle speed is greater than a second speed threshold, and the first controller adjusting the key indicators based on the dynamic model and the dual-closed-loop control model to generate a first control strategy for the vehicle includes:

[0022] Based on the vehicle driving scenario and the key indicators, the lateral controller based on the dynamic model generates and sets the target lateral acceleration, lateral acceleration change rate, and matrix parameters.

[0023] The deviation between the target information and the current position is used as the position closed-loop control index, and the deviation between the target speed and the current speed is used as the speed closed-loop control index, so as to control the longitudinal controller to generate the target longitudinal acceleration and the rate of change of longitudinal acceleration.

[0024] Optionally, the autonomous driving control method includes the following steps before executing the step of the second controller generating a second control strategy for the vehicle based on the scenario data:

[0025] Establish a time series data model based on a time-recurrent neural network;

[0026] The time series data model is trained based on the scenario data and the time series data;

[0027] The control parameters are determined based on the output dimension of the time-series data model.

[0028] The second control strategy is generated based on the control parameters.

[0029] Optionally, the autonomous driving control method includes the step of training the time-series data model based on the scene data and the time-series data in the following steps:

[0030] Collect scenario data and time-series data from multiple drivers, and filter out excellent driving behavior data based on excellent driving standards;

[0031] Based on the scene recognition strategy, the excellent driving behavior data is labeled with scenes.

[0032] The data labeled with the scene is classified to train the time series data model for different scenes, and the trained time series data model is tested and verified.

[0033] When the training effect of the time series data model meets the mass production requirements, the time series data model will be deployed to the vehicle.

[0034] Optionally, after executing the step of the second controller generating a second control strategy for the vehicle based on the scene data and the timing data, the autonomous driving control method includes:

[0035] In response to the driver adjusting the second control strategy of the vehicle, the system acquires personalized scenario data of the vehicle's current driving situation.

[0036] The vehicle's second control strategy is adjusted based on the personalized scenario data.

[0037] Optionally, the autonomous driving control method includes the following steps in executing the second control strategy for adjusting the vehicle based on the personalized scenario data:

[0038] The scene data and time-series data in the personalized scene data are cleaned, labeled, and classified in sequence.

[0039] The time series data model is trained based on the classified data, and the trained time series data model is tested and verified.

[0040] The trained time-series data model is deployed to the vehicle, and the second control strategy is adjusted using the time-series data model.

[0041] Optionally, the autonomous driving control method, in performing the step of generating an arbitration strategy for the vehicle in response to arbitration of the first control strategy and the second control strategy, includes at least one of the following:

[0042] When the deviation of at least one control parameter between the first control strategy and the second control strategy is greater than a first threshold, the vehicle drives using the first control strategy as the arbitration strategy.

[0043] When the deviation of at least one control parameter between the first control strategy and the second control strategy is less than a second threshold, the vehicle drives using the second control strategy as the arbitration strategy.

[0044] When switching between the first control strategy and the second control strategy, the vehicle switches smoothly.

[0045] Optionally, the autonomous driving control method includes the following steps in executing the step of controlling the driving parameters of the vehicle according to the arbitration strategy:

[0046] Based on the front wheel steering angle and longitudinal acceleration of the vehicle, at least one of the target driving torque and braking pressure is calculated using an inverse vehicle model.

[0047] On the other hand, this application also provides a storage medium, specifically, the storage medium stores a computer program, which, when executed by a processor, implements the driver-characteristic-based autonomous driving control method as described above.

[0048] As described above, the driver-characteristic-based autonomous driving control method and storage medium provided in this application, based on the generated scene data, consider the impact of driver characteristics on autonomous driving in multiple scenarios and dimensions through the first and second controllers, thereby controlling the vehicle's driving parameters. This enables better human-machine co-driving with the driver, while also allowing more drivers to accept the driving system, thus improving the driver's trust in and comfort with the intelligent driving system. Attached Figure Description

[0049] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0050] Figure 1 This is a flowchart of an embodiment of an autonomous driving control method based on driver characteristics according to this application.

[0051] Figure 2 This is a control flowchart based on driver characteristics according to an embodiment of this application.

[0052] Figure 3 This is a flowchart illustrating the training process of a time-series data model according to an embodiment of this application.

[0053] The realization of the objectives, functional features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and textual descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0054] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0055] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of this application may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0056] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0057] First Embodiment

[0058] On one hand, this application provides an autonomous driving control method based on driver characteristics. Figure 1 This is a flowchart of an embodiment of an autonomous driving control method based on driver characteristics according to this application.

[0059] Please see Figure 1 In one embodiment, the autonomous driving control method includes:

[0060] S10: In response to identifying the vehicle driving scenario, generate scenario data including key indicators, and obtain time-series data of the driver's operation of key indicators.

[0061] For example, the vehicle driving scenario is selected from typical operating conditions such as lane changing, starting, stopping, the vehicle in front cutting in, the vehicle in front cutting out, and entering / exiting ramps. Scenario data may include vehicle status, environmental information, road information, and the status of surrounding vehicles. Optionally, key indicators are selected from controllable driving parameters such as vehicle speed, acceleration, rate of change of acceleration, steering wheel angle, steering wheel speed, and the maximum, minimum, midpoint, mean, and variance of yaw rate. Based on the recorded status of each driving parameter and the corresponding time, the time-series data of the key indicators can be determined.

[0062] S20: The first controller generates the first control strategy for the vehicle based on the scenario data and timing data.

[0063] For example, the first controller can be a rule-based conventional controller or a training-based personalized controller. The first control strategy is a control strategy based on the first controller for the vehicle's target driving trajectory.

[0064] S30: The second controller generates a second control strategy for the vehicle based on scenario data and timing data.

[0065] For example, the second controller can be a training-based personalized controller or a rule-based conventional controller. The second control strategy is a control strategy based on the second controller for the vehicle's target driving trajectory.

[0066] S40: In response to the arbitration of the first control strategy and the second control strategy, generate an arbitration strategy for the vehicle.

[0067] For example, the first control strategy and the second control strategy are in a parallel output relationship. In order to avoid abnormal control results, the first control strategy and the second control strategy are arbitrated and verified to generate the most suitable arbitration strategy for the vehicle.

[0068] S50: Controls vehicle driving parameters according to arbitration strategy.

[0069] In this embodiment, the driver-characteristic-based autonomous driving control method considers the impact of driver characteristics on autonomous driving in multiple scenarios and dimensions through the first and second controllers based on the generated scene data, thereby comprehensively controlling the vehicle's driving parameters. This enables better human-machine co-driving with the driver, while also allowing more drivers to accept the driving system, thus improving the driver's trust in and comfort with the intelligent driving system.

[0070] In one embodiment, the autonomous driving control method, in performing step S20: the first controller generates a first control strategy for the vehicle based on scene data and timing data, includes at least one of the following:

[0071] S21: When the vehicle speed is less than the first speed threshold, a kinematic model of the vehicle is established, and the first controller adjusts key indicators based on the kinematic model to generate the first control strategy of the vehicle.

[0072] For example, when the vehicle is traveling at low speed, the first controller employs the vehicle's kinematic model. Optionally, this application does not limit the size of the first speed threshold; for example, the first speed threshold can be 30 kph.

[0073] S22: When the vehicle speed is greater than the second speed threshold, a dynamic model and a dual closed-loop control model of the vehicle are established. The first controller adjusts key indicators based on the dynamic model and the dual closed-loop control model to generate the first control strategy of the vehicle.

[0074] For example, when the vehicle is traveling at medium to high speeds, the first controller employs a dynamic model and a dual-closed-loop control model. Optionally, this application does not limit the size of the second speed threshold; for example, the second speed threshold can be 30 kph.

[0075] S23: In response to obtaining personnel information, environmental information and road information of the vehicle, adjust the first control strategy.

[0076] For example, passenger information may include: determining whether there are passengers in the front passenger seat and rear seats based on whether seat belts are worn; when there are passengers, especially those prone to motion sickness, adjusting the first control strategy, such as reducing the rate of change of lateral and longitudinal acceleration and the maximum and minimum values ​​of lateral and longitudinal acceleration, to improve ride comfort and reduce following and tracking trajectory sensitivity. Environmental information may include: determining whether the vehicle is traveling in rainy, foggy, or snowy weather based on wipers, headlights, temperature sensors, etc.; especially when the vehicle is in rainy or snowy weather, adjusting the first control strategy, such as increasing the following distance, limiting the maximum speed during driving, and reducing the yaw rate during trajectory tracking control. Road information may include: analyzing the driver's familiarity with the area the vehicle is traveling in based on background data and determining whether the vehicle is traveling on "high-altitude, high-velocity, and high-speed" roads, urban roads, or rural roads based on high-precision maps or navigation information; adjusting the first control strategy accordingly, such as adopting conservative driving behavior when the vehicle is in an unfamiliar area, adopting normal driving behavior when the vehicle is in a familiar area, and increasing control sensitivity, deceleration response sensitivity, and maximum braking deceleration when the vehicle is traveling on rural roads.

[0077] In this embodiment, the first controller can generate a first control strategy by establishing a mature algorithm model, or it can adjust the first control strategy based on multi-dimensional information such as vehicle, environment, and road, which can better realize the interaction between the driver and the vehicle system and improve the driver's trust in the intelligent system and comfort.

[0078] In one embodiment, the autonomous driving control method, in executing S20: the first controller generates a first control strategy for the vehicle based on scene data and timing data, including:

[0079] S24: Output control information based on the target path. The control information includes multiple reference trajectory points, vehicle status, environmental information, target information, target speed, and road information.

[0080] S25: Based on the control information, identify the current vehicle driving scenario, which includes the current position and current speed, and obtain the driver's key indicators. The key indicators are selected from at least one of the following: vehicle speed, acceleration, rate of change of acceleration, steering wheel angle, steering wheel speed, and yaw rate, including maximum value, minimum value, midpoint value, mean, and variance.

[0081] When traveling, users often plan their destination route in advance. Based on the route, key control information that needs attention along the way can be extracted. This extracted control information allows for the pre-retrieval of relevant trajectory points, speed limits, road conditions, weather conditions, vegetation cover, landmarks, etc., enabling better identification and positioning of the current vehicle driving scenario. For example, the first controller optimizes the solution based on the aforementioned control information and key indicators to obtain a first control strategy tailored to the driver's characteristics.

[0082] S21: When the vehicle speed is less than a first speed threshold, the steps of establishing a kinematic model of the vehicle, adjusting key indicators based on the kinematic model, and generating a first control strategy for the vehicle include:

[0083] S210: Based on the vehicle driving scenario and key indicators, control the kinematic model-based lateral controller to generate the target lateral acceleration, lateral acceleration rate of change, target longitudinal acceleration, longitudinal acceleration rate of change, and matrix parameters.

[0084] For example, when the vehicle is traveling at low speed, the lateral controller optimizes the target lateral acceleration, lateral acceleration rate of change, target longitudinal acceleration, and longitudinal acceleration rate of change among the key indicators based on the kinematic model, and sets the matrix parameters according to the key indicators.

[0085] Optionally, the lateral controller algorithm based on the kinematic model can be: the state variables are [x, y, φ], which are the x and y position coordinate deviations and the heading angle deviation, respectively.

[0086]

[0087] The state vector and control vector are respectively:

[0088]

[0089]

[0090] k represents time k, u(k) represents the control output vector at time k, and T s V is the duration of a single-step execution. r The reference speed is φ, the reference heading angle is L, the wheelbase is δ, and the average steering angle of the front wheels is δ.

[0091] And / or, S22: When the vehicle speed is greater than the second speed threshold, the steps of establishing a vehicle dynamics model and a dual-closed-loop control model, and the first controller adjusting key indicators based on the dynamics model and the dual-closed-loop control model to generate the vehicle's first control strategy include:

[0092] S220: Based on the vehicle driving scenario and key indicators, the lateral controller based on the dynamic model generates and sets the target lateral acceleration, lateral acceleration change rate and matrix parameters;

[0093] S221: The deviation between the target information and the current position is used as the position closed-loop control index, and the deviation between the target speed and the current speed is used as the speed closed-loop control index, so as to control the longitudinal controller to generate the target longitudinal acceleration and the rate of change of longitudinal acceleration.

[0094] For example, when the vehicle is traveling at high speed, the lateral controller optimizes the target lateral acceleration and rate of change of lateral acceleration based on the dynamic model, and sets the matrix parameters according to the key indicators. The longitudinal controller optimizes the target longitudinal acceleration and rate of change of longitudinal acceleration based on the dual closed-loop control model.

[0095] Optionally, the algorithm for the lateral controller based on the dynamic model can be: the state variables are These represent the lateral deviation of the planned trajectory point, the rate of change of the lateral deviation, the deviation from the planned trajectory point's heading wheel, and the rate of change of the heading angle deviation; State space vector: Where, φ des The theoretical yaw rate is determined by the road radius R, k1 and k2 are the front and rear axle lateral stiffness respectively, m is the vehicle curb weight, and v is the theoretical yaw rate. x Let I be the longitudinal speed of the vehicle, and a and b be the distances between the front and rear axles and the center of gravity, respectively. z Let be the moment of inertia of the vehicle about the z-axis.

[0096] The expanded state transition matrix is ​​shown below:

[0097]

[0098] Where, e1 = e y e2 = e φ .

[0099] Discretize the above model:

[0100] Bilinear discretization is performed on term A, and Euler discretization is performed on terms B1 and B2.

[0101]

[0102] Where X represents the system state variable, k represents time k, and T s The sampling step size is represented by 0.01 seconds, and u(k) is the control output vector.

[0103] A is a 4x4 matrix in the state transition matrix:

[0104] B1 is a 4*1 dimension matrix in the state transition matrix:

[0105] B2 is a 4*1 dimension matrix in the state transition matrix:

[0106] In one embodiment, the autonomous driving control method includes the following steps before executing S30: the second controller generates a second control strategy for the vehicle based on scenario data:

[0107] S31: Establish a time series data model based on a time-recurrent neural network.

[0108] For example, a time-recurrent neural network can be a long short-term memory network (LSTM).

[0109] S32: Train the time series data model based on scene data and time series data.

[0110] For example, scene data and time series data are the input dimensions of the time series data model, and may include: vehicle status, surrounding vehicle status, environmental information, road information, and relative timestamps.

[0111] S33: Determine control parameters based on the output dimension of the time series data model.

[0112] For example, the output dimension may include the vehicle's target steering wheel angle and desired acceleration.

[0113] S34: Generate a second control strategy based on the control parameters.

[0114] In this embodiment, the second controller generates a second control strategy based on the time-series data model, thereby controlling the vehicle's movement. By continuously training and learning from excellent driving behavior data, the time-series data model becomes increasingly intelligent and user-friendly, generating control strategies that conform to the driving habits of most drivers.

[0115] In one embodiment, the autonomous driving control method includes the following steps in executing S32: training a time-series data model based on scene data and time-series data:

[0116] S320: Collects scenario data and time-series data from multiple drivers, and filters out excellent driving behavior data based on excellent driving standards.

[0117] Optionally, this application does not limit the criteria for excellent driving, and can select excellent driving behavior data based on the safety of the driver's driving.

[0118] S321: Based on scene recognition strategy, scene labeling is performed on excellent driving behavior data.

[0119] For example, based on the identified scenarios, high-quality driving behavior data is labeled with scenarios, and low-quality data segments that are not labeled are deleted.

[0120] S322: Classify the scene-annotated data to train the time series data model for different scenarios, and test and verify the trained time series data model.

[0121] For example, the labeled data is classified into various scenarios. Due to special circumstances or sensor failure, the data in the corresponding scenario may not be accurate. It is necessary to manually check and perform detailed scene labeling, and add other necessary labels such as weather and road type to the labeled data. At the same time, the time series data model is continuously trained and learned based on the above data, and the trained time series data model is tested and verified.

[0122] S323: When the training effect of the time series data model meets the mass production requirements, deploy the time series data model to the vehicle.

[0123] In this embodiment, the second controller continuously trains and learns excellent driving behavior data based on the time-series data model through driving data under excellent driving standards. This makes the time-series data model increasingly standardized to conform to the driving habits of most drivers, thereby forming a common driving mode suitable for the vast majority of people.

[0124] In one embodiment, the autonomous driving control method includes, after performing step S30: the second controller generates a second control strategy for the vehicle based on scene data and timing data, the following:

[0125] S35: In response to the driver's adjustment of the vehicle's second control strategy, acquire personalized scenario data of the vehicle's current driving situation;

[0126] S36: Adjust the vehicle's secondary control strategy based on individual scenario data.

[0127] In this embodiment, when the driver is dissatisfied with the second control strategy and intervenes, the vehicle's second control strategy is adjusted according to the driver's personalized scenario data, making the autonomous driving more closely resemble the driver's driving behavior. It can also automatically adjust the driving style as driving time increases, forming a personalized driving mode and improving comfort and safety.

[0128] In one embodiment, the autonomous driving control method includes the following steps in executing S36: adjusting the vehicle's second control strategy based on individual scenario data:

[0129] S360: Cleans, labels, and classifies the scene data and time-series data in the personalized scene data in sequence;

[0130] S361: Train the time series data model based on the classified data, and test and verify the trained time series data model;

[0131] S362: Deploy the trained time-series data model to the vehicle and adjust the second control strategy using the time-series data model.

[0132] In this embodiment, the second controller trains personalized scenario data based on the time-series data model according to the driver's personalized driving data training process, which enables the time-series data model to become more and more in line with the driving habits of individual drivers.

[0133] In one embodiment, the autonomous driving control method, in performing S40: generating an arbitration strategy for the vehicle in response to arbitration of a first control strategy and a second control strategy, includes at least one of the following:

[0134] S41: When the deviation of at least one control parameter of the first control strategy and the second control strategy is greater than the first threshold, the vehicle drives using the first control strategy as the arbitration strategy.

[0135] For example, multiple control parameters output by the first control strategy and the second control strategy are mutually arbitrated and verified. When the deviation of at least one control parameter is greater than a first threshold, the vehicle is controlled to drive according to the first control strategy. Optionally, this application does not limit the size of the first threshold.

[0136] S42: When the deviation of at least one control parameter of the first control strategy and the second control strategy is less than the second threshold, the vehicle drives using the second control strategy as the arbitration strategy.

[0137] The traditional control model, individual control model, and common control model have a parallel output relationship. To avoid outliers in certain scenarios due to data bias in the trained individual and common control models, the outputs of the two offline-learned controllers need to be cross-validated with the output of the traditional controller. If the outputs of both exceed a certain range, the offline-learned model is considered to have an abnormal output, and the output of the traditional control model can be used as the actual control data output. For example, multiple control parameters output by the first and second control strategies are cross-arbitrated and validated. When the deviation of at least one control parameter is less than a second threshold, the vehicle is controlled to drive according to the second control strategy. Optionally, this application does not limit the size of the second threshold.

[0138] S43: When switching between the first control strategy and the second control strategy, the vehicle switches smoothly.

[0139] Optionally, to avoid jitter during the switching between the two control models, a ramp-smoothing process can be performed when switching outputs. For example, the vehicle switches between the first and second control strategies in a smooth manner, which can prevent jitter caused by abrupt changes during model switching.

[0140] Subsequently, the vehicle smoothing process may include: an algorithm scheduling period of Ts, the maximum absolute value of the acceleration change rate jerk denoted as jerk_max, a smoothing transition time of Time3, and when a sudden change occurs in the controller output, i.e., the difference between the outputs of two periods exceeds a threshold, the controller outputs the target acceleration Ax_Ctrl1 at time t0 and the target acceleration Ax_Ctrl2 at time (t0+1). The smoothing logic then follows:

[0141] The final output at time (t0+2) is Ax_CtrlRes1 =

[0142] Ax_Ctrl1+min(jerk_max*Ts, (Ax_Ctrl2-Ax_Ctrl1) / Time3);

[0143] The final output at time (t0+3) is Ax_CtrlRes2 =

[0144] Ax_CtrlRes1+min(jerk_max*Ts, (Ax_Ctrl2-Ax_CtrlRes1) / (Time3-Ts));

[0145] The final output at time (t0+4) is Ax_CtrlRes3 =

[0146] Ax_CtrlRes2+min(jerk_max*Ts, (Ax_Ctrl2-Ax_CtrlRes2) / (Time3-2*Ts));

[0147] This continues until Ax_CtrlRes = Ax_Ctrl2, where Ax_Ctrl2 changes over time. Similarly, other control parameters are smoothed using the same calculation method described above.

[0148] In this embodiment, when the difference between at least one control parameter is greater than a first threshold, the driver-personalized control output based on the time-series data model is deemed unreasonable, and the rule-based first controller calculation result is used as the final control strategy output, which improves the stability of autonomous driving. When the difference between at least one control parameter is less than a second threshold, the driver-personalized control output based on the time-series data model is deemed reasonable, and the second controller calculation result is used as the final control strategy output, which improves the intelligence of autonomous driving and better matches the driver's individual driving behavior. Optionally, when the first threshold is greater than the second threshold, if the difference between at least one control parameter is within the common interval between the two thresholds, control can be performed by a third mode, or control modes can compete or arbitrate to determine a control mode. For example, if driving at low speed most of the time and occasionally reaching the high-speed threshold for a short period, then the low-speed mode is temporarily used. Similarly, if driving at high speed most of the time and occasionally braking to the low-speed threshold for a short period, then the high-speed mode is temporarily used. Optionally, when the first threshold is less than the second threshold, and the difference of at least one control parameter is within the blank interval between the two thresholds, control can be implemented using a third mode, or a combination of low-speed and high-speed modes can be used, or a mode can be arbitrated to determine control. For example, if driving at low speed most of the time, and occasionally briefly exceeding the low-speed threshold but not reaching the high-speed threshold, then control can temporarily revert to low-speed mode. Similarly, if driving at high speed most of the time, and occasionally briefly braking to below the high-speed threshold but not reaching the low-speed threshold, then control can temporarily revert to high-speed mode.

[0149] In one embodiment, the autonomous driving control method includes the following steps in executing S50: controlling the vehicle's driving parameters according to an arbitration strategy:

[0150] S51: Based on the vehicle's front wheel steering angle and longitudinal acceleration, calculate at least one of the target driving torque and braking pressure using an inverse vehicle model.

[0151] The vehicle model generates vehicle condition information such as lateral and longitudinal acceleration and the rate of change of acceleration based on information related to driving force and braking force, and the target vehicle speed. In this embodiment, adjusting the braking strategy based on the vehicle condition feedback information, the front wheel angle combined with the driving speed can solve for data such as lateral offset acceleration. Changes in the front wheel angle can lead to data such as the rate of change of lateral offset acceleration. Furthermore, the longitudinal acceleration can determine data such as the rate of change of longitudinal acceleration. Through reverse engineering calculations of the vehicle model, control data such as target driving torque and braking pressure can be further solved in reverse, thereby generating a control strategy for corresponding vehicle control.

[0152] Second Embodiment

[0153] On the other hand, this application also provides a storage medium, specifically, a computer program stored on the storage medium, which, when executed by a processor, implements the driver-characteristic-based autonomous driving control method as described above.

[0154] Third Embodiment

[0155] In one embodiment of the intelligent driving controller, typical operating condition recognition strategies such as lane changing, starting, following and stopping, cutting in front of another vehicle, and entering and exiting ramps are formulated in the intelligent driving controller; the scene recognition module is mainly responsible for the following strategies:

[0156] Autonomous vehicle lane change recognition: When the vehicle speed is greater than V1_min, based on the four lane line parameters output by the perception system (C0: distance between the vehicle and the lane line), when C0 is less than Dist1_min and the absolute value of the steering wheel angle is greater than StrAng_min, turn on the turn signal (optional) to determine that this is an autonomous vehicle lane change trigger scenario;

[0157] Follow-up start recognition: When the vehicle speed changes from stationary (0kph) to non-stationary (speed > V2_min), and there is vehicle target information in front of the vehicle in the lane, and the distance between the vehicle and the vehicle is less than Dist2_min, and the speed of the vehicle in front is accelerating, it is determined that this is a follow-up start scenario.

[0158] Smooth following recognition: When the vehicle speed is greater than V3_min, there is a vehicle target in the lane, and the speed change between the vehicle in front and the vehicle speed is less than the threshold DeltV1_min, and the duration exceeds the threshold Time1_min, it is determined that this is a steady following driving scenario.

[0159] Constant speed driving recognition: If the vehicle speed is greater than V4_min and there are no vehicles within the Dist3_min range in front of the vehicle in the lane for a duration exceeding the threshold Time2_min, the vehicle is determined to be in a steady-state driving scenario.

[0160] Stop-and-go recognition: When the vehicle's speed is decreasing and there are vehicles gradually decelerating within the Dist4_min range ahead in this lane, the vehicle's speed drops to 0, and this is judged as a stop-and-go scenario;

[0161] Forward vehicle cut-in recognition: When the vehicle speed is greater than the threshold V5_min, there is a vehicle target on the adjacent left or right side, and the target distance is less than Dist5_min, when the following target changes from the target ID of the current lane to the target ID of the vehicle cutting into the adjacent lane, it is judged as a forward vehicle cut-in scenario.

[0162] Forward vehicle cut-out recognition: When the vehicle speed is greater than the threshold V6_min, there is a vehicle in front of the vehicle within the Dist6_min range in the vehicle's lane. When the target in front of the vehicle's lane changes to a target in an adjacent lane, the vehicle in front is judged to be cutting out.

[0163] Ramp identification: Based on the geographical information output by the high-precision map, as well as the vehicle's heading or navigation information, determine whether the current vehicle is on an ramp.

[0164] Off-ramp identification: Based on the geographical information output by the high-precision map, as well as the vehicle's heading or navigation information, determine whether the current vehicle is in an off-ramp scenario.

[0165] During the driver's operation, identify the above typical scenarios and extract key indicators related to the driver. Key indicators include vehicle speed, acceleration, jerk (rate of change of acceleration), steering wheel angle, steering wheel speed, and yaw rate.

[0166] For typical scenario data, the system can determine whether to re-upload data for that scenario based on the data already collected in the cloud. If so, the scenario data is packaged and uploaded to the cloud for storage in the database for training purposes; otherwise, it is not uploaded to the cloud. Specifically, the strategy for determining whether to upload data is necessary is as follows: if the cloud stores sufficient data for a particular scenario and the data metric shows no significant change, data does not need to be uploaded to the cloud. Otherwise, the data upload operation is performed.

[0167] Figure 2 This is a control flowchart based on driver characteristics according to an embodiment of this application.

[0168] Please refer to Figure 2 In this embodiment, driver characteristic learning includes two parts: online learning and offline learning.

[0169] Online learning involves learning statistical indicators representing the driver at the intelligent driving controller, which then automatically recalibrates the controller's relevant indicators. This represents shallow learning of driver characteristics. Since driving behavior changes over time, online learning methods learn the driver's real-time driving style to keep pace with dynamically changing driving behaviors. As the driver operates the vehicle, key indicators are updated online, and model calibration parameters are corrected in real time to obtain different control styles.

[0170] Offline learning is a form of deep learning based on driver characteristics; it is divided into personalized driving behavior learning and common driving behavior learning. Personalized driver learning involves training only on the time-series data uploaded by the driver, and the resulting control model is a personalized controller. On the other hand, time-series data uploaded by drivers with good driving habits is used for training, and the resulting controller model is a common controller. The model learned offline will be sent to the vehicle-side intelligent driving controller to participate in vehicle trajectory tracking control.

[0171] It identifies multi-dimensional information such as vehicle, environment, and road conditions to adjust driving style indicators. For example, it adjusts control parameters such as PID and MPC to adapt to different driving behaviors.

[0172] 1. Vehicle information includes: determining whether there are passengers in the front passenger seat and the back seat based on whether they are wearing seat belts; when there are passengers, especially those prone to motion sickness, minimizing the maximum and minimum values ​​of jerk and lateral and longitudinal acceleration as much as possible under the premise of trajectory tracking and following safety, improving ride comfort, and reducing the sensitivity of following or tracking trajectory.

[0173] 2. Environmental information includes: determining whether the vehicle is driving in rainy, foggy, or snowy weather based on wipers, headlights, temperature sensors, etc. In particular, when driving in rainy or snowy weather, increasing the following distance when following other vehicles, limiting the maximum driving speed during driving, and reducing the yaw rate during trajectory tracking control.

[0174] 3. Road information may include:

[0175] A. Based on backend data analysis, when the vehicle enters an unfamiliar area for the driver, a conservative driving behavior is adopted; when the vehicle enters a familiar area for the driver, the driving style is adjusted to normal mode.

[0176] B. Based on high-precision maps or navigation information, determine whether the vehicle is traveling on "high-speed" roads, urban roads, or rural roads. Improve control sensitivity on rural roads, increase response sensitivity during deceleration, and increase maximum braking deceleration.

[0177] In this embodiment, for the planned path, when controlling the vehicle to travel along the target trajectory, the desired steering wheel angle and acceleration are calculated based on the indicators under different driving styles.

[0178] The traditional controller, personalized controller, and common controller have a parallel output relationship. To avoid the trained model from outputting abnormal values ​​in certain scenarios due to limited data, the output results of the two controllers in offline learning need to be cross-validated with the output result of the traditional controller. When the output results of both exceed a certain range (which can be calibrated), the output of the offline learning controller is considered abnormal. At this time, the calculation result of the traditional controller is used as the expected output. When the offline learning controller switches between the two controllers, a ramp smoothing process is performed when switching outputs to avoid jitter.

[0179] The control methods for each model are explained below.

[0180] In model-based control methods, vehicle control primarily employs mature algorithms such as PID and MPC. Lateral control utilizes the MPC control method. At lower vehicle speeds, such as less than 30 kph, kinematic models can be used to control the lateral and longitudinal driving parameters. The state variables [X, Y, Φ] represent the X and Y position coordinate deviations and the heading angle deviation, respectively.

[0181] Optionally, the lateral controller algorithm based on the kinematic model can be: the state variables are [x, y, φ], which are the x and y position coordinate deviations and the heading angle deviation, respectively.

[0182]

[0183] The state vector and control vector are respectively:

[0184]

[0185]

[0186] T s V is the duration of a single-step execution. r The reference speed is φ, the reference heading angle is L, the wheelbase is δ, and the average steering angle of the front wheels is δ.

[0187] At higher vehicle speeds, such as greater than 30 kph, a 2DOF dynamic model can be used to control lateral movement.

[0188] State variables are These represent the lateral deviation of the planned trajectory point, the rate of change of the lateral deviation, the deviation from the planned trajectory point's heading wheel, and the rate of change of the heading angle deviation; State space vector: Where, φ des The theoretical yaw rate is determined by the road radius R, k1 and k2 are the front and rear axle lateral stiffness respectively, m is the vehicle curb weight, and v is the theoretical yaw rate. x Let I be the longitudinal speed of the vehicle, and a and b be the distances between the front and rear axles and the center of gravity, respectively. z Let be the moment of inertia of the vehicle about the z-axis.

[0189] The expanded matrix is ​​shown below:

[0190]

[0191] Discretize the above model:

[0192] Bilinear discretization is performed on term A, and Euler discretization is performed on terms B1 and B2.

[0193]

[0194] At medium to high vehicle speeds, longitudinal control can employ dual closed-loop control, with the control variables being longitudinal distance deviation and speed deviation, respectively. A control flow based on a traditional model may include at least one of the following:

[0195] 1. The intelligent driving controller outputs a series of reference trajectory points S1, S2, ..., Sn, and outputs data such as vehicle status, environmental information, target information, and road information.

[0196] 2. The DCCS controller identifies the current vehicle driving scenario based on the upper-level control output information.

[0197] 3. The driver characteristic self-identification module identifies key driver indicators online, such as acceleration, jerk (rate of change of acceleration), steering wheel angle, steering wheel speed, vehicle speed, and yaw rate, including their maximum, minimum, midpoint, mean, and variance.

[0198] 4. After identifying the above indicators, output the results to the arbitration module and the lateral controller (based on kinematic MPC controller and dynamic MPC controller).

[0199] 5. When optimizing the solution, the lateral controller sets limits on the lateral and longitudinal accelerations and rates of change based on the driver's characteristic indicators; when setting the Q and R matrices, the matrix parameters are set based on the driver's characteristic parameters.

[0200] 6. The longitudinal controller uses the deviation between the target position and the current position as the position closed-loop control (outer loop), and the target speed and the current vehicle speed as the speed closed-loop control (inner loop), and outputs the target longitudinal acceleration to the arbitration module.

[0201] 7. The arbitration module receives the characteristic parameters and the output results of the lateral and longitudinal controllers from the driver characteristic parameter identification module, and performs mutual verification between the traditional controller and the data-trained model. When the output deviation between the two exceeds a certain threshold, it is determined that the output of the driver characteristic controller is unreasonable, and the calculation result of the traditional controller is used as the final lateral control output.

[0202] 8. When the longitudinal controller is a dual closed-loop PID controller, the output target acceleration and acceleration change rate are limited according to the identification results of the driver parameter identification module, and the final target acceleration is output.

[0203] 9. The final front wheel steering angle and longitudinal acceleration are used to calculate the target driving torque or braking pressure based on the inverse vehicle model.

[0204] 10. Smooth the output results to prevent sudden changes in vehicle operation that could cause jitter.

[0205] Figure 3 This is a flowchart illustrating the training process of a time-series data model according to an embodiment of this application.

[0206] Please see Figure 3 In one embodiment, the training control method based on a time-series data model includes:

[0207] The model's input dimensions include: information such as vehicle targets in nine recognition scenes around the vehicle body, basic vehicle information such as vehicle speed, acceleration, jerk, steering wheel angle, and steering wheel speed, environmental information such as navigation, weather, and road conditions, and relative timestamps.

[0208] The model's output dimensions include: steering wheel angle and expected acceleration.

[0209] Training methods include, but are not limited to, LSTM.

[0210] The trained model is then deployed to the vehicle's intelligent driving controller.

[0211] When the vehicle is actually controlled by a personalized control model, and the driver is dissatisfied with the control behavior and intervenes, the data is uploaded to the cloud to correct and train the model, making the model increasingly intelligent and human-like.

[0212] The personalized model training process is as follows:

[0213] During the driver's operation of the vehicle, typical scenario data is identified and uploaded according to the upload strategy;

[0214] Since preliminary labels are already present when the data is uploaded (e.g., in the case of data in a stop-and-go scenario, there will be a label indicating the scenario type of the data segment), the data is then automatically classified and cleaned in the cloud to remove low-quality data segments.

[0215] Because the scene recognition strategy based on rules on the vehicle side may not be accurate due to special circumstances or sensor failure, manual verification and detailed scene annotation are required to add ID and other necessary labels (such as weather, road type, etc.) to the data.

[0216] The finely labeled data is categorized and stored for training and learning in different scenarios;

[0217] The initial data selection included: information on 9 targets around the vehicle, basic vehicle information such as vehicle speed, acceleration, jerk, steering wheel angle, and steering wheel speed, environmental information such as navigation, weather, and road conditions, as well as relative timestamps, as model inputs;

[0218] The trained model is tested and verified. When the model training effect meets the mass production requirements, the model is upgraded online and distributed to the vehicle-side intelligent driving controller.

[0219] During the process of the trained model controlling the vehicle, if the driver is dissatisfied with the current control effect and intervenes, the data segment of the scene at this time is recorded, uploaded to the cloud, and the above process is repeated.

[0220] The training process for the common model is as follows:

[0221] After multiple vehicles upload typical scenario data to the cloud, excellent driving behaviors are selected as target training data.

[0222] Since preliminary labels are already present when the data is uploaded (e.g., in the case of data in a stop-and-go scenario, there will be a label indicating the scenario type of the data segment), the data is then automatically classified and cleaned in the cloud to remove low-quality data segments.

[0223] Because the scene recognition strategy based on rules on the vehicle side may not be accurate due to special circumstances or sensor failure, manual verification and detailed scene annotation are required to add ID and other necessary labels (such as weather, road type, etc.) to the data.

[0224] The finely labeled data is categorized and stored for training and learning in different scenarios;

[0225] The initial data selection included: information on nine vehicle targets around the vehicle, basic vehicle information such as vehicle speed, acceleration, jerk, steering wheel angle, and steering wheel speed, environmental information such as navigation, weather, and road conditions, as well as relative timestamps, which were used as model inputs.

[0226] The trained model is tested and verified. When the model training effect meets the mass production requirements, the model is upgraded online and distributed to the vehicle-side intelligent driving controller.

[0227] During the process of the trained model controlling the vehicle, if the driver is dissatisfied with the current control effect and intervenes, the data segment of the scene at this time is recorded, uploaded to the cloud, and the above process is repeated.

[0228] As described above, the driver-characteristic-based autonomous driving control method and storage medium provided in this application improve the public's acceptance of assisted driving functions; during autonomous driving, the driving behavior is closer to the driver, resulting in higher comfort and safety; the driving style takes into account passengers prone to motion sickness and drives more smoothly; the driving style becomes more conservative considering weather or road conditions; when the driver frequently intervenes in the second control strategy, the second controller will recommend excellent driving control methods and gradually improve the driver's driving skills.

[0229] It should be noted that step designations such as S10 and S20 are used in this application for the purpose of more clearly and concisely describing the corresponding content, and do not constitute a substantial limitation on the order. In specific implementation, those skilled in the art may execute S20 first and then S10, etc., but these should all be within the protection scope of this application.

[0230] In the embodiments of the storage medium provided in this application, all the technical features of any of the above method embodiments may be included. The extended and explanatory content of the specification is basically the same as that of the embodiments of the above methods, and will not be repeated here.

[0231] This application also provides a computer program product, which includes computer program code. When the computer program code is run on a computer, it causes the computer to perform the methods described in the various possible implementations above.

[0232] This application also provides a chip, including a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that a device with the chip installed performs the methods described in the various possible implementations above.

[0233] It is understood that the above scenarios are merely examples and do not constitute a limitation on the application scenarios of the technical solutions provided in the embodiments of this application. The technical solutions of this application can also be applied to other scenarios. For example, as those skilled in the art will know, with the evolution of system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0234] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0235] The steps in the method of this application embodiment can be adjusted, combined, or deleted according to actual needs.

[0236] The units in the device of this application embodiment can be merged, divided, and deleted according to actual needs.

[0237] In this application, the same or similar terms, concepts, technical solutions and / or application scenario descriptions are generally described in detail only when they appear for the first time. When they appear again, they are generally not repeated for the sake of brevity. When understanding the technical solutions and other contents of this application, the same or similar terms, concepts, technical solutions and / or application scenario descriptions that are not described in detail later can be referred to their previous relevant detailed descriptions.

[0238] In this application, the descriptions of the various embodiments have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0239] The technical features of the present application can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present application.

[0240] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. An automated driving control method based on driver characteristics, characterized in that, include: In response to identifying vehicle driving scenarios, it generates scenario data including key indicators and acquires time-series data of driver operation of key indicators; The first controller generates a first control strategy for the vehicle based on the scenario data and the timing data; The second controller generates a second control strategy for the vehicle based on the scenario data and the timing data; In response to the arbitration of the first control strategy and the second control strategy, an arbitration strategy for the vehicle is generated; The vehicle's driving parameters are controlled according to the arbitration strategy; The step of the first controller generating the first control strategy for the vehicle based on the scene data and the timing data includes at least one of the following: When the vehicle speed is less than a first speed threshold, a kinematic model of the vehicle is established, and the first controller adjusts the key indicators based on the kinematic model to generate a first control strategy for the vehicle. When the vehicle speed is greater than the second speed threshold, a dynamic model and a dual-closed-loop control model of the vehicle are established. The first controller adjusts the key indicators based on the dynamic model and the dual-closed-loop control model to generate a first control strategy for the vehicle. In response to acquiring personnel information, environmental information, and road information of the vehicle, the first control strategy is adjusted.

2. The automatic driving control method as described in claim 1, characterized in that, The first controller generates a first control strategy for the vehicle based on the scene data and the timing data, including: Control information is output based on the target path. The control information includes multiple reference trajectory points, vehicle status, environmental information, target information, target speed, and road information. Based on the control information, the current vehicle driving scenario is identified, which includes the current position and current speed. The driver's key indicators are obtained, which are selected from at least one of the following: vehicle speed, acceleration, rate of change of acceleration, steering wheel angle, steering wheel speed, and yaw rate: maximum value, minimum value, midpoint value, mean, and variance. The step of establishing a kinematic model of the vehicle when the vehicle speed is less than a first speed threshold, and the first controller adjusting the key indicators based on the kinematic model to generate a first control strategy for the vehicle includes: Based on the vehicle driving scenario and the key indicators, the kinematic model-based lateral controller generates the target lateral acceleration, lateral acceleration rate of change, target longitudinal acceleration, longitudinal acceleration rate of change, and matrix parameters. And / or, the step of establishing a dynamic model and a dual-closed-loop control model for the vehicle when the vehicle speed is greater than a second speed threshold, and the first controller adjusting the key indicators based on the dynamic model and the dual-closed-loop control model to generate a first control strategy for the vehicle includes: Based on the vehicle driving scenario and the key indicators, the lateral controller based on the dynamic model generates and sets the target lateral acceleration, lateral acceleration change rate, and matrix parameters. The deviation between the target information and the current position is used as the position closed-loop control index, and the deviation between the target speed and the current speed is used as the speed closed-loop control index, so as to control the longitudinal controller to generate the target longitudinal acceleration and the rate of change of longitudinal acceleration.

3. The automatic driving control method as described in claim 1, characterized in that, Before the step of the second controller generating the second control strategy for the vehicle based on the scenario data, the following steps are included: Establish a time series data model based on a time-recurrent neural network; The time series data model is trained based on the scenario data and the time series data; The control parameters are determined based on the output dimension of the time-series data model. The second control strategy is generated based on the control parameters.

4. The automatic driving control method as described in claim 3, characterized in that, The step of training the time series data model based on the scene data and the time series data includes: Collect scenario data and time-series data from multiple drivers, and filter out excellent driving behavior data based on excellent driving standards; Based on the scene recognition strategy, the excellent driving behavior data is labeled with scenes. The data labeled with the scene is classified to train the time series data model for different scenes, and the trained time series data model is tested and verified. When the training effect of the time series data model meets the mass production requirements, the time series data model will be deployed to the vehicle.

5. The automatic driving control method as described in claim 4, characterized in that, After the step of the second controller generating the second control strategy for the vehicle based on the scene data and the timing data, the following steps are included: In response to the driver adjusting the second control strategy of the vehicle, the system acquires personalized scenario data of the vehicle's current driving situation. The vehicle's second control strategy is adjusted based on the personalized scenario data.

6. The automatic driving control method as described in claim 5, characterized in that, The step of adjusting the second control strategy of the vehicle based on the personalized scenario data includes: The scene data and time-series data in the personalized scene data are cleaned, labeled, and classified in sequence. The time series data model is trained based on the classified data, and the trained time series data model is tested and verified. The trained time-series data model is deployed to the vehicle, and the second control strategy is adjusted using the time-series data model.

7. The automatic driving control method according to any one of claims 1-6, characterized in that, The step of generating an arbitration strategy for the vehicle in response to arbitration of the first control strategy and the second control strategy includes at least one of the following: When the deviation of at least one control parameter between the first control strategy and the second control strategy is greater than a first threshold, the vehicle drives using the first control strategy as the arbitration strategy. When the deviation of at least one control parameter between the first control strategy and the second control strategy is less than a second threshold, the vehicle drives using the second control strategy as the arbitration strategy. When switching between the first control strategy and the second control strategy, the vehicle switches smoothly.

8. The automatic driving control method according to any one of claims 1-6, characterized in that, The step of controlling the vehicle's driving parameters according to the arbitration strategy includes: Based on the front wheel steering angle and longitudinal acceleration of the vehicle, at least one of the target driving torque and braking pressure is calculated using an inverse vehicle model.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the driver-characteristic-based autonomous driving control method as described in any one of claims 1-8.

Images