Vehicle control method and device, electronic device, vehicle and storage medium

Abstract

The application relates to a vehicle control method and device, an electronic device, a vehicle and a storage medium, wherein the method comprises the following steps: in response to detecting that a steering system of a vehicle fails, acquiring state information of the steering system and a virtual lane in front of the vehicle; based on the state information and the virtual lane, determining a risk value function corresponding to a first region; the first region is a predicted collision avoidance region; according to the risk value function and an operating parameter of the vehicle, determining a target acceleration of the vehicle; and controlling the vehicle based on the target acceleration. According to the application, the first region is analyzed by combining the state information of the steering system and the virtual lane in front of the vehicle, the risk value function corresponding to the first region is determined, the target acceleration is dynamically adjusted by introducing the risk value function, the safety redundancy demand under different working conditions can be adapted, and the flexibility and safety of vehicle control are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile transmission, in particular to a vehicle control method and device, an electronic device, a vehicle and a storage medium. BACKGROUND

[0002] The vehicle can control the vehicle to turn through a steering system. In the related art, the steering system can include internal power supply, circuit, motor, steering controller, vehicle speed sensor, rotation angle sensor and other modules. In the case of failure of the internal modules of the steering system, the steering system will have steering failure, which will cause the vehicle to be unable to travel normally or cause safety hazards in the process of vehicle travel, and there are problems of low safety and poor flexibility. SUMMARY

[0003] The present application provides a vehicle control method and device, an electronic device, a vehicle and a storage medium, which improves the flexibility and safety of vehicle control.

[0004] The present application provides a vehicle control method, applied to an electric drive controller of a vehicle, which comprises:

[0005] In response to detecting that the steering system of the vehicle fails, state information of the steering system and a virtual lane in front of the vehicle are obtained;

[0006] Based on the state information and the virtual lane, a risk value function corresponding to a first region is determined; the first region is a predicted collision avoidance region;

[0007] According to the risk value function and the running parameters of the vehicle, a target acceleration of the vehicle is determined;

[0008] The vehicle is controlled based on the target acceleration.

[0009] According to the above technical means, first, in response to detecting that the steering system of the vehicle fails, the state information of the steering system and the virtual lane in front of the vehicle are obtained; then, based on the state information and the virtual lane, the risk value function corresponding to the first region is determined; then, according to the risk value function and the running parameters of the vehicle, the target acceleration of the vehicle is determined; finally, the vehicle is controlled based on the target acceleration. In this way, on the one hand, by combining the health state of the steering system with the collision avoidance region division, the safety redundancy demand under different working conditions can be adapted; on the other hand, by introducing the risk value function to dynamically adjust the target acceleration, the safety and comfort of the longitudinal control are improved, which is more suitable for unstructured road and low algorithm power platform application scenarios compared with the prior art, reduces the hardware cost and improves the generalization ability of the algorithm.

[0010] In some embodiments, based on the state information and the virtual lane, a risk value function corresponding to the first region is determined, including: collision avoidance path planning is performed based on the state information and the virtual lane to determine a target collision avoidance path; the first region is determined according to the target collision avoidance path and a current vehicle speed of the vehicle; a distribution of road boundary points in the first region is counted; and the risk value function corresponding to the first region is constructed based on the distribution.

[0011] According to the above technical means, by performing collision avoidance path planning based on the state information and the virtual lane, determining the first region in combination with the current vehicle speed, counting the distribution of the road boundary points, and finally constructing the risk value function, the driving environment of the vehicle on the non-standard road can be accurately modeled, so that the longitudinal speed planning strategy can be dynamically adjusted, the collision risk can be effectively reduced, and the safety performance and reliability of the electric drive controller can be improved.

[0012] In some embodiments, the distribution includes a scatter point density of the road boundary points in the first region, a first parameter corresponding to the scatter point density, and a spatial gradient; the first parameter is used to represent a trend of change of the scatter point density in a first time period; and the risk value function corresponding to the first region is constructed based on the distribution, including: weight values corresponding to the scatter point density, the first parameter, and the spatial gradient are respectively determined; and the scatter point density, the first parameter, and the spatial gradient are weighted and summed to obtain the risk value function corresponding to the first region.

[0013] According to the above technical means, in the first step, the road boundary point distribution map is formed by obtaining the environment perception data; in the second step, key feature indicators such as the scatter point density, the first parameter, and the spatial gradient are extracted; in the third step, appropriate weights are assigned to the above key feature indicators to ensure that the relative importance of different factors in risk assessment is reflected; and finally, the risk value function is generated by weighted summation as the basic input of subsequent path planning and control strategy. The above implementation process realizes the transformation from the original perception data to the abstract risk model, and enhances the adaptability and decision accuracy of the vehicle in the complex and unstructured road environment.

[0014] In some embodiments, the collision avoidance path planning based on the state information and the virtual lane to determine the target collision avoidance path includes: judging whether the steering system is abnormal based on the state information; in the case that the steering system is normal, determining the target collision avoidance path based on the virtual lane; and in the case that the steering system is abnormal, determining the target collision avoidance path based on the wheelbase, the steering ratio, and the current steering wheel angle of the vehicle.

[0015] According to the above technical means, by introducing the state information judgment mechanism, the virtual lane modeling, and the path planning strategy based on the vehicle geometric parameters, the embodiment can flexibly switch the path planning mode under different working conditions, thereby significantly improving the safety and reliability of the electric drive controller in complex and abnormal scenarios.

[0016] In some embodiments, the first area is determined according to the target collision avoidance path and the current vehicle speed, including: extending a preset width on the left and right sides of the target collision avoidance path respectively to obtain the width of the first area; determining the length of the first area based on the current vehicle speed and the preview time.

[0017] According to the above technical means, by symmetrically expanding a preset width on both sides of the target collision avoidance path, and calculating the length of the first area based on the current vehicle speed and the preview time, a safe collision avoidance area with horizontal and vertical dimensions is constructed. The safe collision avoidance area constructed in the above manner can discover potential collision risks earlier, thereby improving the accuracy of obstacle identification of the vehicle in a non-standard road environment.

[0018] In some embodiments, the target acceleration of the vehicle is determined according to the risk value function and the operating parameters of the vehicle, including: calculating a candidate acceleration according to a pre-set control strategy and the operating parameters of the vehicle; and performing dynamic gain adjustment on the candidate acceleration by using the risk value function to obtain the target acceleration.

[0019] According to the above technical means, by introducing a control strategy to generate a candidate acceleration, and combining the risk value function for dynamic gain adjustment, more flexible and intelligent longitudinal speed planning can be achieved. By introducing a control strategy to generate a candidate acceleration, and combining the risk value function for dynamic gain adjustment, the robustness and adaptability of the electric drive controller for different working conditions can be improved, which is beneficial to enhancing the overall driving safety and ride comfort.

[0020] In some embodiments, the candidate acceleration is calculated according to a pre-set control strategy and the operating parameters of the vehicle, including: determining a first acceleration based on a first control strategy and first operating parameters of the vehicle; the first control strategy is used for active speed limiting control of the vehicle; determining a second acceleration based on a second control strategy and second operating parameters of the vehicle; the second control strategy is used for speed limiting control of the vehicle based on road boundary points; and determining the selected acceleration according to the first acceleration and the second acceleration.

[0021] According to the above technical means, two different control strategies are introduced, i.e., active speed limiting control based on virtual lanes and speed limiting control based on Free Space boundary points, realizing hierarchical decision of the longitudinal speed of the vehicle. The application of this method enables the electric drive controller to cope with the complexity of non-standard roads and the risk of lateral control failure at the same time, thereby improving the overall robustness of the vehicle control method.

[0022] In some embodiments, the first operating parameter comprises a current vehicle speed of the vehicle, a maximum lateral acceleration supported by the vehicle, a road adhesion coefficient, and a current slope; determining the first acceleration based on the first control strategy and the first operating parameter of the vehicle comprises: determining a curvature radius of a road ahead of the vehicle based on a preview distance of the vehicle; determining a first speed according to the curvature radius and the maximum lateral acceleration; determining a second speed according to the curvature radius, the road adhesion coefficient, and the current slope; determining a minimum value between the first speed and the second speed as a target speed; and calculating the first acceleration based on the target speed and the current vehicle speed.

[0023] According to the above technical means, a complete vehicle longitudinal control process is formed: first, the behavior boundary of the vehicle is established by collecting the operating parameters of the vehicle and the road environment; second, the road curvature information is obtained based on the preview technology, and the theoretical speed limit is calculated in combination with the performance parameters of the vehicle; then, the more conservative safe speed is generated by considering factors such as road adhesion conditions and slope; finally, the most stringent speed is selected as the target speed by comparison, and the appropriate longitudinal acceleration is calculated accordingly, completing the entire control process. This hierarchical progressive processing method not only guarantees the clarity of the control logic, but also enhances the ability of the electric drive controller to cope with complex working conditions.

[0024] In some embodiments, the second operating parameter comprises initial and final states of the vehicle in a second time period; determining the second acceleration based on the second control strategy and the second operating parameter of the vehicle comprises: generating a quintic polynomial trajectory equation according to the initial and final states to obtain a candidate trajectory; calculating a cost function for each candidate trajectory; and determining the second acceleration according to the cost function and the corresponding dynamic constraint conditions of the vehicle.

[0025] According to the above technical means, by introducing the initial and final states and combining the quintic polynomial trajectory equation, candidate trajectories that meet the actual driving requirements can be generated, improving the accuracy and flexibility of trajectory planning, thereby better coping with complex road conditions and significantly enhancing the safety and comfort of longitudinal control.

[0026] In some embodiments, obtaining the state information of the steering system and the virtual lane ahead of the vehicle comprises: obtaining road geometric features ahead of the vehicle through a front-view camera of the vehicle; extracting road boundary points based on the road geometric features; detecting point cloud data corresponding to obstacles ahead of the vehicle through a millimeter wave radar; and fitting a virtual lane ahead of the vehicle according to the road boundary points and the point cloud data.

[0027] According to the above technical means, by combining the data acquisition modes of the front-view camera and the millimeter wave radar, and generating a virtual lane, the accuracy and stability of obstacle identification on a non-standard road can be improved, which helps to optimize the division logic of the longitudinal collision avoidance area, and helps the electric drive controller to realize more intelligent and flexible longitudinal speed planning to adapt to more complex traffic scenarios.

[0028] The embodiment of the application provides a vehicle control device, which is applied to an electric drive controller of a vehicle and comprises:

[0029] An acquisition unit is configured to acquire state information of a steering system and a virtual lane in front of the vehicle in response to detection of a failure of the steering system of the vehicle.

[0030] A first determination unit is configured to determine a risk value function corresponding to a first area based on the state information and the virtual lane; the first area is a predicted collision avoidance area.

[0031] A second determination unit is configured to determine a target acceleration of the vehicle according to the risk value function and an operating parameter of the vehicle.

[0032] A control unit is configured to control the vehicle based on the target acceleration.

[0033] The embodiment of the application provides a vehicle control system, which comprises a steering system and an electric drive controller of a vehicle.

[0034] The steering system of the vehicle is configured to send a steering failure signal to the electric drive controller in a case where a failure is identified.

[0035] The electric drive controller is configured to acquire state information of the steering system and a virtual lane in front of the vehicle in response to the steering failure signal; determine a risk value function corresponding to a first area based on the state information and the virtual lane; the first area is a predicted collision avoidance area; determine a target acceleration of the vehicle according to the risk value function and an operating parameter of the vehicle; and control the vehicle based on the target acceleration.

[0036] The embodiment of the application provides an electronic device, which comprises a processor and a memory; the memory stores a computer program capable of running on the processor; and the processor implements the steps in the above method when executing the computer program.

[0037] The embodiment of the application provides a vehicle, which comprises the above electronic device.

[0038] The embodiment of the application provides a computer readable storage medium, which stores a computer program; and the computer program is executed by a processor to implement the steps in the above method.

[0039] The embodiment of the application provides a computer program product, comprising a computer program or instructions, which, when executed by a processor, implements the steps in the above method.

[0040] According to the above technical means, first, in response to detecting that the steering system of the vehicle fails, the state information of the steering system and the virtual lane in front of the vehicle are acquired; then, based on the state information and the virtual lane, the risk value function corresponding to the first region is determined; then, according to the risk value function and the running parameter of the vehicle, the target acceleration of the vehicle is determined; finally, the vehicle is controlled based on the target acceleration. In this way, on the one hand, by combining the lateral control health state with the longitudinal collision avoidance region division, the safety redundancy demand under different working conditions can be adapted, especially when the lateral control fails, effective braking can still be performed to avoid collision risks; on the other hand, by introducing the risk value function to dynamically adjust the target acceleration, the safety and comfort of the longitudinal control are improved, which is more suitable for unstructured road and low-computing-power platform application scenarios compared with the prior art, reduces the hardware cost and improves the generalization ability of the algorithm.

[0041] The embodiment of the application provides a vehicle control method and an electric drive controller.

[0042] The beneficial effects of the application are as follows:

[0043] (1) The application acquires the state information of the steering system of the vehicle and the virtual lane in front of the vehicle, calculates the target acceleration based on the risk value function and the running parameter of the vehicle, and controls the vehicle based on the target acceleration calculated based on the risk value function and the running parameter of the vehicle, which can provide an effective longitudinal redundancy mechanism in the case of lateral control failure, thereby adapting to various different working conditions, reducing collision risks, and improving the flexibility and safety of vehicle control.

[0044] (2) The application plans a collision avoidance path based on the state information of the steering system and the virtual lane, determines the first region in combination with the current vehicle speed, and statistically constructs the risk value function by counting the distribution of road boundary points, which can realize accurate modeling of the vehicle driving environment on a non-standard road, thereby dynamically adjusting the longitudinal speed planning strategy and effectively reducing collision risks.

[0045] (3) The application dynamically adjusts the risk value function based on the scatter point density of the road boundary points in the first region, the change trend of the scatter point density with time, and the spatial gradient, which can realize more intelligent collision avoidance control in a non-standard road environment, improve the generalization ability of the vehicle control method, and enhance the adaptability and decision accuracy of the vehicle control method in a complex unstructured road environment.

[0046] (4) The application generates a candidate acceleration by introducing a control strategy, and generates a target acceleration by combining the candidate acceleration with a risk value function, which can fully consider the vehicle limit conditions under various different working conditions, realize the dynamic balance between safety and efficiency, effectively cope with the uncertainty caused by the special-shaped obstacles on the non-standard road, and enhance the overall driving safety.

[0047] (5) The application expands the preset width symmetrically on both sides of the target collision avoidance path, and calculates the length of the first area based on the current vehicle speed and the preview time, so as to construct a safe collision avoidance area with horizontal and vertical dimensions, which is beneficial to improve the accuracy of obstacle identification of the vehicle on the non-standard road environment.

[0048] (6) The application can more comprehensively reflect the traffic environment in front of the vehicle by fusing multi-sensor data to fit a virtual lane, provide more accurate path information for longitudinal speed planning, and can improve the accuracy and stability of obstacle identification on the non-standard road. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A flowchart of a vehicle control method provided by an embodiment of the application is shown in the figure;

[0050] Figure 2 A component structure diagram of a vehicle control device provided by an embodiment of the application is shown in the figure;

[0051] Figure 3 A component structure diagram of a vehicle control system provided by an embodiment of the application is shown in the figure;

[0052] Figure 4 A hardware entity diagram of an electronic device provided by an embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0053] The embodiments of the application will be described below with reference to the accompanying drawings and preferred embodiments, and those skilled in the art can easily understand other advantages and effects of the application from the disclosure in the specification. The application can also be implemented or applied by different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the application. It should be understood that the preferred embodiments are only for illustration of the application, and are not intended to limit the protection scope of the application.

[0054] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the application in a schematic manner, and only the components related to the application are shown in the diagrams, not the number, shape and size of the components when actually implemented. The actual implementation of each component may be arbitrarily changed in type, number and proportion, and the layout pattern of the components may be more complex.

[0055] In the following description, reference is made to "some embodiments", which describe only a subset of all possible embodiments, and which can be understood as "some but not all" possible embodiments, and as such, reference to "some embodiments" is not necessarily made to the same subset of embodiments, and may or can not be combinable with other embodiments.

[0056] In the following description, the terms "first", "second", "third", etc. are used only to distinguish similar objects, and do not represent a specific order or sequence for the objects, and it can be understood that the "first", "second", "third" can be interchanged with a specific order or sequence as allowed, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application is for the purpose of describing embodiments of this application only, and is not intended to be limiting of this application.

[0058] The vehicle can control the vehicle to turn through a steering system. In the related art, the current vehicle-mounted steering system is generally an electric power steering system. The electric power steering system controls the current through a steering controller, which can provide the most ideal current to the assist motor during the driver's steering process, so as to control the assist motor to provide the best assist. When the internal power supply, circuit, assist motor, steering controller, vehicle speed sensor and other internal modules of the electric power steering system fail, the electric power steering system will lose the assist function, resulting in failure of the lateral control of the vehicle, and further causing the vehicle to be unable to travel or to have safety hazards during travel.

[0059] In the related art, there is a lack of effective means for longitudinal speed planning in non-standard road scenarios, especially when the lateral control fails, which cannot provide effective redundancy mechanism. For example, the longitudinal control algorithm widely used in structured roads is difficult to handle irregular obstacles, and has poor adaptability to low algorithm power platforms. In addition, the traditional method often assumes that the lateral control is always effective, ignoring the abnormal situation of lateral control that may occur in actual application, resulting in a greater collision risk in emergency situations.

[0060] To solve the above problems, the application provides a vehicle control method, which is applied to an electric drive controller of a vehicle. The method detects a steering system failure, obtains state information of the steering system and a virtual lane in front of the vehicle, determines a predicted collision avoidance area and a corresponding risk value function based on the state information and the virtual lane, calculates a target acceleration in combination with vehicle operating parameters, and controls the vehicle according to the target acceleration. By introducing a lateral control health degree evaluation mechanism, the application can adaptively adjust the collision avoidance area and the risk evaluation model under different control states, thereby realizing more efficient and safer longitudinal speed planning. The method is not only suitable for complex non-standard roads, but also can realize high-precision collision avoidance control on a low-computing-power platform.

[0061] The vehicle control method provided by the embodiments of the application can be executed by an electric drive controller of a vehicle. In the following, the technical solutions in the embodiments of the application will be clearly and completely described with reference to the drawings in the embodiments of the application.

[0062] Figure 1 is an optional flowchart of the vehicle control method provided by the embodiments of the application, as shown in Figure 1 The method comprises the following steps 101 to 104.

[0063] Step 101, in response to detecting that a steering system of a vehicle fails, state information of the steering system and a virtual lane in front of the vehicle are obtained.

[0064] In actual application, when the steering system of the vehicle is abnormal, such as an electric power steering module failure, a steering angle tracking error exceeding a set threshold, etc., the electric drive controller reads the state signal of the lateral control module through the vehicle-mounted bus system (such as a CAN bus) to determine whether the lateral control system is in a normal or abnormal state. At the same time, the perception module (such as a front-view camera and a millimeter wave radar) continuously collects the front road environment information and fuses to generate a Free Space scatter point set. Based on these scatter point data, the electric drive controller fits the free space boundary curves on the left and right sides of the vehicle, and then constructs the virtual lane in front of the vehicle as a reference path.

[0065] The Free Space scatter point set refers to a passable area in front of the vehicle, which is a space set composed of road boundary points or obstacle information extracted from sensor perception data, and is used to represent the available path range of the road. The Free Space scatter point set can be generated by fusion of multiple source sensors such as front-view cameras and millimeter wave radars of the vehicle.

[0066] Virtual lane refers to a curve fitted based on Free Space scatter points, representing a reference path on which the vehicle can safely travel. The virtual lane can be expressed by a cubic polynomial, which is constructed based on the road boundary information provided by the perception module and dynamically adjusted according to the lateral control health state.

[0067] Exemplarily, the virtual lane in front of the vehicle constructed by the electric drive controller is usually expressed by a cubic polynomial, and the generation process of the virtual lane in front of the vehicle constructed by the electric drive controller needs to consider the influence of the health state of the lateral control system, for example, in the case of failure of the lateral control system, the virtual lane in front of the vehicle constructed by the electric drive controller can be generated based on the predicted path rather than the current path to ensure the accuracy of the collision avoidance area.

[0068] At step 102, based on the state information and the virtual lane in front of the vehicle, a risk value function corresponding to the first area is determined; the first area is the predicted collision avoidance area.

[0069] It should be noted that the risk value function in the embodiments of the present application refers to a mathematical expression for quantifying the potential collision risk of the collision avoidance area, which is used to quantify the potential collision risk in the collision avoidance area in front of the vehicle.

[0070] In some embodiments, the calculation of the risk value function depends on the Free Space scatter point density , the trend of the scatter point density over time , and the spatial gradient and other factors. Specifically, the scatter point density is defined as the ratio of the number of scatter points N to the area S in the first area, that is = N / S. reflects the movement trend of the obstacle, for example, the case of an animal crossing the road, and the spatial gradient characterizes the change of the distribution of scatter points in space, which helps to distinguish static interference (such as shrubs) from real obstacles.

[0071] In some embodiments, the shape and size of the first area are dynamically adjusted according to the lateral control health state. The lateral control health state is used to evaluate the running state of the vehicle lateral control system. The present application judges whether the lateral control is in a normal working state by monitoring the parameters such as steering actuator fault diagnosis information, steering angle tracking error, heading error and lateral error. Different lateral control health states will affect the division logic and risk assessment method of the longitudinal collision avoidance area.

[0072] Exemplarily, if the lateral control is normal, the statistical area is extended to a certain width left and right along the reference path; if the lateral control is abnormal, the statistical area is extended along the predicted path. The embodiments of the present application can divide the low-risk area (p ), early warning zone (ρ ≤ ρ ≤ ), brake trigger zone (ρ ), and construct a risk value function to achieve accurate assessment of collision risk in different scenarios.

[0073] Step 103, determine the target acceleration of the vehicle according to the risk value function and the operating parameters of the vehicle.

[0074] The determination of the target acceleration needs to consider the risk value function, the current state of the vehicle (such as speed, acceleration), road curvature, and dynamic constraint conditions.

[0075] Illustratively, different ways can be used to determine the target acceleration of the vehicle for different scenarios. For example, in the complex road active speed limit control scenario, the electric drive controller first calculates the front preview curvature radius R according to the fitting result of the virtual lane in front of the vehicle, and sets the target speed upper limit combined with the allowed maximum comfortable lateral acceleration. Then, a PID controller is used to calculate the target acceleration according to the deviation between the current speed and the target speed. In the Free Space scatter point deceleration control scenario, the electric drive controller uses a quintic polynomial trajectory optimization method, defines the initial and final states and designs a cost function, and selects the optimal solution from multiple candidate braking schemes as the target acceleration. Regardless of the scenario, the final target acceleration will be dynamically adjusted according to the real-time updated risk value function to ensure optimal longitudinal control effect under different risk levels.

[0076] Step 104, control the vehicle based on the target acceleration.

[0077] The electric drive controller adjusts the motor output torque based on the final target acceleration, thereby achieving precise control of the longitudinal motion of the vehicle.

[0078] Further, to improve control accuracy and response speed, a receding horizon update mechanism can also be introduced, that is, periodically reacquire environmental perception data and vehicle state information, and recalculate the optimal control acceleration. In addition, the electric drive controller can also impose physical limit constraints on the target acceleration, such as maximum acceleration and maximum deceleration limits, to prevent the vehicle from losing control or passengers from being uncomfortable due to the over-activation of the control instructions from the electric drive controller.

[0079] The vehicle control method provided in the embodiments of the present application can provide an effective longitudinal redundancy mechanism in the case of failure of lateral control, thereby significantly reducing the risk of collision and improving the safety and comfort of driving.

[0080] In some embodiments, the step 101 of acquiring the state information of the steering system and the virtual lane in front of the vehicle comprises steps S11 to S14, wherein:

[0081] In step S11, the road geometry in front of the vehicle is acquired by a front-view camera of the vehicle.

[0082] The front-view camera refers to an imaging device installed at the front of the vehicle for capturing real-time image information of the road in front. The front-view camera usually has a high resolution and a wide-angle lens, and can identify irregular obstacles (such as animals, scattered objects, etc.), road edges and irregular terrain features on unstructured roads. The road geometry refers to the road contour information collected by the front-view camera, including but not limited to road boundary lines, road surface textures, obstacle positions, etc. The road geometry can be used to construct the passable area boundary in front of the vehicle, thereby providing basic input for subsequent path planning.

[0083] The embodiments of the present application can effectively identify irregular obstacles on non-standard roads by acquiring road geometry through the front-view camera, which can improve the comprehensiveness and accuracy of vehicle environmental perception, and is conducive to improving the safety of longitudinal collision avoidance planning.

[0084] In step S12, the road boundary points are extracted based on the road geometry.

[0085] In some embodiments, the road geometry can include a set of road-related features extracted from the image collected by the front-view camera through image processing algorithms (such as edge detection, semantic segmentation). By analyzing and extracting these features, visual road edges and visual Free Space points can be obtained. Based on the visual road edges and the visual Free Space points, the road boundary points can be screened out.

[0086] Exemplarily, in the process of extracting the road boundary points, the RANSAC algorithm can be used to remove outliers, and the least squares method can be used to curve fit the inlier set, and finally a discrete point set representing the road boundary is generated. The discrete point set is used as basic data for modeling the virtual lane, which is used to support the electric drive controller to understand the shape and width of the current driving road.

[0087] The embodiment of the present application can more accurately describe the actual geometric shape of the road by extracting the road boundary points, and the road boundary point extraction provides accurate data support for the generation of virtual lanes, thereby improving the reliability of path planning.

[0088] In step S13, the point cloud data corresponding to the obstacle in front of the vehicle is detected by the millimeter wave radar.

[0089] The millimeter wave radar is a high-frequency electromagnetic wave sensor that can emit and receive reflected signals to detect the distance, speed, and angle information of the object in front of the vehicle. The point cloud data refers to a data set composed of multiple target points in a three-dimensional space detected by the radar, and each point contains distance, azimuth, and height information. The point cloud data can complement the deficiencies of visual sensors in low light, rainy and snowy weather, or complex occlusion conditions, and enhance the recognition ability of obstacles.

[0090] In the embodiment of the present application, the point cloud data is obtained by the millimeter wave radar, which can provide stable and reliable obstacle information in harsh environments or complex scenes. Using this obstacle information can improve the robustness and safety of the longitudinal collision avoidance algorithm.

[0091] In step S14, the virtual lane in front of the vehicle is fitted according to the road boundary points and the point cloud data.

[0092] The virtual lane refers to a reference path fitted according to the actual road boundary points and the obstacle point cloud data, and is usually represented in the form of a cubic polynomial. The virtual lane generation process in the embodiment of the present application combines the advantages of visual and radar data. Specifically, a reference line is first determined (preferably using the radar guardrail line), then the Free Space points with a transverse distance less than a predetermined threshold from the reference line are selected, then the Random Sample Consensus (RANSAC) algorithm is used to remove outliers to obtain an inner point set; the inner point set is curve-fitted, and finally the virtual lane center line that the vehicle can travel on is generated. The virtual lane can reflect the topological structure of the current road and provide key inputs for subsequent longitudinal control strategies.

[0093] For example, in the embodiment of the present application, the least squares method can be used to fit a cubic polynomial curve of the Free Space points on the left and right sides of the vehicle as a virtual lane.

[0094] For all inner points ( ), the matrix , the vector , and the vector are constructed:

[0095]

[0096] The least square solution is .

[0097] The fitted cubic polynomial curve can be expressed as:

[0098]

[0099] Further, the fitting result can be verified. Specifically, the mean square error (MSE) after fitting is calculated:

[0100]

[0101] wherein, .

[0102] If the left and right lines are both greater than the set threshold , the safety braking of constant deceleration is performed and the takeover is reported.

[0103] For the fitting result of at least one of the left and right lines satisfying , the center line is calculated as a reference path:

[0104]

[0105] wherein, is a preset lane width.

[0106] The embodiments of the present application can more comprehensively reflect the traffic environment in front of the vehicle by fusing multi-sensor data to fit a virtual lane, provide more accurate path information for longitudinal speed planning, and can improve the accuracy and stability of obstacle identification on non-standard roads, thereby realizing efficient and safe collision avoidance control.

[0107] In some embodiments, step 102 includes steps S21-S24, wherein:

[0108] Step S21, collision avoidance path planning based on state information and virtual lane to determine a target collision avoidance path.

[0109] wherein, the collision avoidance path planning refers to calculating a safe and feasible path to avoid colliding with the front obstacle according to the real-time perceived road environment information (such as Free Space scatter points, obstacle position, vehicle state) and the topological structure of the virtual lane during vehicle driving. The vehicle usually considers factors such as vehicle dynamics constraints, braking ability, and lateral control health degree when performing collision avoidance path planning, to ensure that the vehicle can still make a reasonable response even in a complex non-standard road environment. The target collision avoidance path is the result obtained by collision avoidance path planning, which is an optimal path that meets safety and comfort, and is used to guide subsequent speed planning and control strategy formulation.

[0110] In some embodiments, different lateral control health states correspond to different target collision avoidance paths. For example, if the lateral control is normal, the target collision avoidance path can be determined based on the reference path to which the virtual lane is fitted; if the lateral control is abnormal, a predicted path can be determined based on the current steering wheel angle of the vehicle, and the target collision avoidance path can be determined based on the predicted path.

[0111] Embodiments of the present application can identify potential dangers in advance and generate response plans on unstructured roads by planning collision avoidance paths based on the state information of the steering system and the virtual lane, thereby improving the robustness and safety of the electric drive controller.

[0112] In step S22, a first region is determined according to the target collision avoidance path and the current speed of the vehicle.

[0113] The first region refers to a spatial range in the direction of travel of the vehicle, which is previewed according to the target collision avoidance path and the current speed. The electric drive controller can evaluate whether there are high-risk factors in front of the vehicle based on the first region. The length of the first region is usually determined by the current speed and the preview time, and the width is set according to the boundary of the target collision avoidance path. The preview time refers to a time window used to predict the driving path of the vehicle in a future period of time in longitudinal speed planning. The length of the preview time determines the response capability of the vehicle to the front environment, which is usually related to the current speed.

[0114] In some embodiments, determining the first region according to the target collision avoidance path and the current speed of the vehicle includes extending a preset width on the left and right sides of the target collision avoidance path to obtain the width of the first region, and determining the length of the first region based on the current speed of the vehicle and the preview time.

[0115] The preset width refers to a safety boundary range extended on both sides of the target collision avoidance path. For example, the size of the preset width can be pre-calibrated, or the parameter can be dynamically adjusted according to the current road environment, to define the lateral safety space of the vehicle when making longitudinal collision avoidance decisions. The electric drive controller can set the value of the preset width as a fixed value, or adaptively adjust it according to road types, obstacle density, vehicle speed, and other factors. For example, in a low-speed scenario, a smaller preset width can be set; while in a high-speed or complex non-normative road, a larger preset width can be set to enhance collision avoidance safety.

[0116] It can be understood that there is a logical correlation between the preset width and the target collision avoidance path. The target collision avoidance path provides a reference trajectory for longitudinal control, while the preset width determines the range of lateral activity allowed around the target collision avoidance path, thereby forming a two-dimensional collision avoidance area. By symmetrically expanding the preset width on both sides of the target collision avoidance path, it can be ensured that the vehicle still has sufficient redundant space to avoid collision in the case of abnormal lateral control.

[0117] In actual application, when the vehicle is driving on an unstructured road (such as a rural road or a construction section), the electric drive controller can calculate a virtual lane center line as the target collision avoidance path based on the Free Space point set provided by the perception module, and expand a certain width on both sides of the target collision avoidance path to construct a rectangular or trapezoidal first area for subsequent longitudinal collision avoidance planning and judgment, thereby helping to improve the vehicle's ability to respond to irregular obstacles (such as suddenly appearing animals or scattered objects).

[0118] The setting of the preview time usually takes into account the maximum braking capability of the vehicle, the response delay of the perception system, and the uncertainty factors of the road. For example, in high-speed scenarios, the preview time may be set to 2-3 seconds; while in low-speed or complex environments, the preview time may be shorter to respond faster.

[0119] It can be understood that the current speed of the vehicle and the preview time jointly determine the longitudinal coverage range of the first area. Specifically, the electric drive controller multiplies the current speed by the preview time to obtain the distance that the vehicle can travel in the future time, i.e., the length of the first area. The length of the first area reflects the range of the area in front of the system concerned when making collision avoidance judgments, and through this information, timely deceleration or braking measures can be taken before a collision risk may occur.

[0120] In actual implementation, assuming that the current speed of the vehicle is 60 km / h (about 16.7 m / s) and the preview time is 2 seconds, the length of the first area is 33.4 meters. The electric drive controller will continuously monitor the change of the Free Space point density within the length of the first area, and combine with the lateral control health information to judge whether to enter the warning zone or the braking trigger zone. This method based on the dynamic adjustment mechanism of vehicle speed and preview time makes the collision avoidance area more suitable for actual driving situations, improving the real-time performance and robustness of the system.

[0121] In the embodiments of the present application, the preset width is symmetrically expanded on both sides of the target collision avoidance path, and the length of the first region is calculated based on the current vehicle speed and the preview time, so as to construct a safe collision avoidance region with horizontal and vertical dimensions, thereby improving the accuracy of obstacle identification of the system in a non-standard road environment, so that potential collision risks can be discovered earlier, and more safe and comfortable longitudinal collision avoidance control can be realized.

[0122] In step S23, the distribution of the road boundary points in the first region is counted.

[0123] The road boundary point refers to a point representing the road boundary in the Free Space scatter point set detected in the first region. The road boundary point reflects the geometric characteristics of the road edge, and is of great significance for judging the available space of the road. The distribution of the road boundary points refers to the density, arrangement rule and change trend of the road boundary points in space, and can be used to evaluate the passability of the road and the density of potential obstacles. For example, the distribution of the road boundary points in a low-density area may represent an empty road section, and the distribution of the road boundary points in a high-density area may imply the existence of more obstacles or irregular terrain.

[0124] By counting the distribution of the road boundary points in the first region, the electric drive controller can quantitatively analyze the complexity of the road, and based on the analysis result, implement more refined risk assessment.

[0125] Further, the region type corresponding to the first region can be determined based on the scatter point density of the road boundary points in the first region and a preset density threshold, wherein the region type can include a low-risk area (ρ ), a warning area (ρ ≤ρ≤ ), and a brake triggering area (ρ ). is the scatter point density of the road boundary points in the first region.

[0126] It can be understood that for the low-risk area, no processing is required, i.e., no subsequent process is performed; for the warning area, an alarm information can be generated first without triggering brake control; and for the brake triggering area, brake control can be triggered according to the vehicle control method provided in the embodiments of the present application.

[0127] In step S24, a risk value function corresponding to the first region is constructed based on the distribution of the road boundary points.

[0128] The risk value function is a mathematical expression that comprehensively reflects the safety level of the first region, and is usually calculated by weighting parameters such as the distribution density, change rate, and gradient of the road boundary points. The risk value function can dynamically adjust the weight coefficients to adapt to different driving scenarios. For example, when the lateral control is normal, more attention can be paid to the density of static obstacles; when the lateral control is abnormal, more attention can be paid to the motion trend of the obstacles. The higher the risk value, the more dangerous the first region reflected by the risk value function, and the vehicle should adopt a more conservative control strategy.

[0129] By constructing the risk value function, the electric drive controller can quantitatively evaluate the safety status of the first region, and adjust the longitudinal speed planning strategy according to the quantitative evaluation result of the safety status of the first region, to realize adaptive collision avoidance control, thereby improving the overall driving safety and ride comfort.

[0130] It should be noted that the steps S21 to S24 in the embodiments of the present application are closely related. First, the collision avoidance path planning provides a basic path framework for subsequent region division; then, the first region determined in combination with the current vehicle speed is applied to the basic path provided by the collision avoidance path planning to realize local refinement and focusing of the basic path; subsequently, the road boundary point data in the first region determined in combination with the current vehicle speed is collected and statistically analyzed to further extract environmental features; finally, the risk value function is constructed based on the extracted environmental features, and the function is used to complete the quantitative evaluation of the safety level of the environment. The whole process forms a closed-loop feedback mechanism, which enables the electric drive controller to dynamically adapt to complex and variable non-standard road environments, thereby improving the intelligent level and safety performance of the electric drive controller.

[0131] In the embodiments of the present application, the collision avoidance path planning is performed based on the state information and the virtual lane, the first region is determined in combination with the current vehicle speed, the distribution of the road boundary points is counted, and finally the risk value function is constructed, which realizes accurate modeling of the driving environment of the vehicle on the non-standard road, so as to dynamically adjust the longitudinal speed planning strategy, thereby effectively reducing the collision risk and improving the safety performance and reliability of the electric drive controller.

[0132] In some embodiments, step S21: collision avoidance path planning based on state information and virtual lane to determine a target collision avoidance path, comprises steps S211-S213, wherein:

[0133] Step S211: determining whether the steering system is abnormal based on the state information.

[0134] The state information of the steering system can include the lateral control health state of the steering system, and specifically can include the lateral control normal state and the lateral control abnormal state.

[0135] It can be understood that the steering system abnormality refers to a state that the steering system cannot perform the steering operation according to the expected trajectory, which may be caused by a hardware failure, a software error or external interference. In the embodiments of the present application, whether the lateral control is normal can be evaluated based on the fault diagnosis information of the steering gear and the error data of the lateral control module.

[0136] Specifically, whether the lateral control is in a healthy state can be determined by detecting the steering angle tracking error, the heading angle tracking error and the lateral position tracking error of the vehicle. The steering angle tracking error, the heading angle tracking error and the lateral position tracking error can be expressed as:

[0137] Steering angle tracking error

[0138] Heading angle tracking error

[0139] Lateral position tracking error

[0140] If If the above conditions are met, it can be determined that the lateral control is normal, that is, the steering system is in a normal state.

[0141] If any of the above errors exceeds the set threshold, it is determined that the result of the state evaluation is that the steering system is in a steering abnormal state.

[0142] In the electric drive controller, after determining whether the steering system is in a steering abnormal state based on the state information, if the result of the determination is that the steering system is in a steering abnormal state, the vehicle may deviate from the preset path, increasing the risk of collision. Therefore, in this case, the electric drive controller must switch to a safer longitudinal control strategy to ensure driving safety.

[0143] In addition, in the process of determining whether the steering system is in a steering abnormal state based on the state information, the state information of the steering system is usually provided by an electronic control unit, and the state information provided by the electronic control unit includes but is not limited to the current steering wheel angle, the steering assist output, the steering motor current and the like. These data provided by the electronic control unit can be used to further verify the process of determining whether the steering system is in a steering abnormal state based on the state information.

[0144] By monitoring the state information used in the step of determining whether the steering system is in a steering abnormal state based on the state information in real time, and determining whether the result of determining whether the steering system is in a steering abnormal state is a steering abnormal state, the controller can timely identify the risk of lateral control failure, thereby triggering a corresponding longitudinal collision avoidance path planning strategy, and improving the safety and robustness of the vehicle in abnormal scenarios.

[0145] Step S212, in the case that the result of determining whether the steering system is abnormal is normal, determining the target collision avoidance path based on the virtual lane.

[0146] When the electric drive controller determines whether the steering system is abnormal based on the state information of the steering system and obtains a result that the steering system is normal, the collision avoidance path can be planned according to the pre-constructed virtual lane information. For example, the electric drive controller generates a virtual lane by fusing the data sensed by multiple sensors (such as a camera and a millimeter wave radar), and the virtual lane represents the boundary line of the drivable area of the vehicle. The system dynamically constructs the virtual boundary of the road by fitting the Free Space scatter set, as the basis for path planning.

[0147] The construction method of the virtual lane includes cubic polynomial fitting of the left and right Free Space points to form two boundary curves, and calculation of the center line of the virtual lane as a reference path. The construction method of the virtual lane can effectively deal with the special-shaped obstacles on unstructured roads, and improve the adaptability and accuracy of path planning.

[0148] Step S213, in the case that the result of determining whether the steering system is abnormal is abnormal, determining the target collision avoidance path based on the wheelbase, the steering ratio and the current steering wheel angle of the vehicle.

[0149] When the electric drive controller determines whether the steering system is abnormal based on the state information of the steering system and obtains a result that the steering system is abnormal, the path planning will no longer rely on the virtual lane, but will predict the motion trajectory of the vehicle according to the geometric parameters (such as the wheelbase and the steering ratio) of the vehicle itself and the current steering wheel angle, and generate a collision avoidance path according to the predicted motion trajectory of the vehicle. The method of determining the target collision avoidance path based on the wheelbase, the steering ratio and the current steering wheel angle of the vehicle is suitable for the case that the lateral control fails but the longitudinal control is still available.

[0150] The wheelbase refers to the distance between the front and rear wheels of the vehicle, which determines the minimum radius of the vehicle when turning. The steering ratio is the proportional relationship between the steering wheel angle and the actual turning angle of the front wheel, which affects the steering sensitivity of the vehicle. The steering wheel angle is the steering instruction input by the driver or automatically controlled by the system, which reflects the turning intention of the vehicle.

[0151] In the case of abnormality of the steering system, the electric drive controller assumes that the vehicle will travel along the circular arc trajectory corresponding to the current steering wheel angle, and the electric drive controller plans the collision avoidance path based on the circular arc trajectory corresponding to the current steering wheel angle.

[0152] For example, in order to simplify the calculation, in the case of abnormality of the steering system, the target collision avoidance path is a predicted path, in which case the vehicle is considered to travel along a circular arc with a constant radius, and the radius of the circular arc is wherein is a turning ratio, is a steering wheel turning angle, is a wheelbase of the vehicle.

[0153] The way of determining the target collision avoidance path based on the wheelbase, the turning ratio and the current steering wheel turning angle of the vehicle, although not as accurate as the way of determining the target collision avoidance path based on the virtual lane, can provide effective longitudinal redundancy protection in emergency situations.

[0154] When the electric drive controller determines that the steering system is in an abnormal steering state based on the state information of the steering system, it can quickly switch to a path planning mode based on the parameters of the vehicle itself to ensure that even if the lateral control fails, the vehicle can continue to perform basic safe collision avoidance operations, thereby reducing the probability of accidents and enhancing the fault tolerance capability of the vehicle in extreme situations.

[0155] The embodiments of the present application can flexibly switch the path planning mode under different working conditions by introducing a state information judgment mechanism, virtual lane modeling and path planning strategy based on vehicle geometric parameters, thereby significantly improving the safety and reliability of the electric drive controller in complex and abnormal scenarios.

[0156] In some embodiments, step S24 constructs a risk value function corresponding to the first region based on the distribution of the road boundary points, including: respectively determining a scatter point density of the road boundary points in the first region, a first parameter corresponding to the scatter point density, and a weight value corresponding to each of the spatial gradients; and performing weighted summation on the scatter point density, the first parameter and the spatial gradients to obtain the risk value function corresponding to the first region. The first parameter is used to represent the trend of the scatter point density in the first time period.

[0157] It should be noted that the scatter point density refers to the number of road boundary points per unit area in the first region, indicating the degree of density of the road boundary information. The scatter point density can reflect the complexity of the road environment, such as a high risk of a large number of obstacles and sparse distribution of road boundary points. The first parameter is used to represent the trend of the scatter point density in the first time period, for example, whether the scatter point density is rapidly decreasing or increasing, so as to determine whether a dynamic obstacle (such as an animal crossing the road) appears. The spatial gradient refers to the change rate of the scatter point density between adjacent grids, which can reflect the degree of local boundary change and is used to identify the influence of static interference such as shrubs. The weight value is a different influence factor given to the above three factors to quantify the contribution degree of the weight value to the overall risk value. By setting reasonable weight coefficients, the adaptability and robustness of the algorithm to different scenarios can be improved.

[0158] The three of the scatter point density, the first parameter and the spatial gradient are complementary. The scatter point density provides static spatial distribution information, the first parameter describes the trend of the scatter point density changing over time, and the spatial gradient reflects the mutation in the local space.

[0159] The embodiments of the present application combine the scatter point density, the first parameter and the spatial gradient and assign weights, which can more comprehensively depict the risk state of the current environment.

[0160] In some embodiments, the risk value function can be expressed as:

[0161]

[0162] wherein, is the calculated Free Space scatter point density, is the trend of the scatter point density changing over time, that is, the first parameter in the embodiments of the present application, which can represent the change of the scatter point distribution caused by the obstacle movement (such as an animal crossing the road), is the spatial gradient of the scatter point density, which can represent the spatial distribution of the scatter points (which can effectively suppress the false triggering of static interference such as shrubs). and are the corresponding weight coefficients, respectively.

[0163] In some embodiments, The calculation can be performed by finite difference and sliding average filtering. The density difference of adjacent grids can be calculated by grid division on the statistical area S.

[0164] The risk value function in the embodiments of the present application can comprehensively evaluate the road traffic safety of the first region and serve as an important basis for subsequent longitudinal speed planning. The embodiments of the present application can realize more intelligent collision avoidance control in non-standard road environment by dynamically adjusting the risk value function based on the scatter point density of the road boundary point in the first region, the trend of the scatter point density changing over time and the spatial gradient, improve the generalization ability of the vehicle control method, and enhance the adaptability and decision accuracy in complex unstructured road environment, thereby enhancing the safety and stability of the vehicle driving on the non-standard road.

[0165] In some embodiments, step 103: determining the target acceleration of the vehicle according to the risk value function and the running parameters of the vehicle, includes steps S31-S32, wherein:

[0166] Step S31: calculating the candidate acceleration according to the pre-set control strategy and the running parameters of the vehicle.

[0167] The control strategy refers to a set of decision rules based on preset rules and logic for guiding vehicle behavior. The control strategy can be a rule-based method (such as PID control), a model-based method (such as MPC), or a learning-based method (such as reinforcement learning). The operating parameters include current vehicle speed, acceleration, position, distance to the front obstacle, and other information.

[0168] In the embodiments of the present application, different control strategies can be formulated for different application scenarios. In some embodiments, a corresponding control strategy can be selected based on the current application scenario, so as to calculate a candidate acceleration based on the control strategy and the operating parameters of the vehicle.

[0169] In some embodiments, comfort, safety, efficiency, and other dimensions can be considered comprehensively, and multiple initial accelerations that meet different priorities can be generated for different control strategies in each dimension. It can be understood that each initial acceleration represents a possible vehicle action selection, such as acceleration, deceleration, or maintaining the current speed. The calculation of the initial acceleration is usually affected by factors such as dynamic constraints, safety boundaries, and path curvature. For example, at a curve, the electric drive controller can generate a lower candidate acceleration to ensure lateral stability; while on a straight road, the electric drive controller can generate a higher initial acceleration to improve driving efficiency.

[0170] If there are multiple initial accelerations, an optimal acceleration can be selected from the multiple initial accelerations as the candidate acceleration in the embodiments of the present application.

[0171] Step S32, the candidate acceleration is dynamically gain-adjusted by using the risk value function to obtain a target acceleration.

[0172] The risk value function in the embodiments of the present application is a mathematical expression for comprehensively evaluating the risk degree of the front environment, and the output result of the risk value function reflects the safety level of the current driving environment. The risk value function is usually composed of features such as scatter point density, density change trend, and spatial gradient, and generates the final risk score through weighted summation. For example, when the scatter point density in the free space area is high, it indicates that there are more obstacles, and the risk value increases; when the scatter point density change trend is sharp, it may mean that a moving obstacle is approaching, and the risk value further increases. Exemplarily, the risk value function in the embodiments of the present application can be represented as:

[0173]

[0174] According to the risk value function, the calculated candidate acceleration is dynamically gain-adjusted to obtain a target acceleration. Exemplarily, the target acceleration in the embodiments of the present application can be represented as:

[0175]

[0176] wherein, is a gain coefficient, is a nonlinear adjustment factor.

[0177] It can be understood that when the risk value is high, the gain coefficient of the candidate acceleration can be reduced, so that the vehicle adopts a more conservative acceleration strategy to avoid potential collision risks. Conversely, when the risk value is low, a higher gain coefficient can be allowed, thereby improving driving efficiency and comfort. Dynamic gain adjustment can achieve a dynamic balance between safety and efficiency, which is particularly important in complex scenarios such as non-standard roads and lateral control failure.

[0178] In the embodiments of the present application, by introducing a control strategy to generate a candidate acceleration and combining it with a risk value function for dynamic gain adjustment, a more flexible and intelligent longitudinal speed planning can be achieved. By introducing a control strategy to generate a candidate acceleration and combining it with a risk value function for dynamic gain adjustment, the electric drive controller can achieve a dynamic balance between safety and efficiency, thereby effectively dealing with the uncertainty brought by irregular obstacles on non-standard roads, and enhancing the overall driving safety and ride comfort.

[0179] In some embodiments, the step S31 of calculating a candidate acceleration according to a pre-set control strategy and operating parameters of the vehicle comprises steps S311-S313, wherein:

[0180] Step S311, determining a first acceleration based on a first control strategy and first operating parameters of the vehicle; the first control strategy is used for active speed limit control of the vehicle.

[0181] The first control strategy is an algorithm logic for limiting the speed of the vehicle, and the core goal of the first control strategy is to limit the speed of the vehicle in advance according to the geometric characteristics (such as the radius of curvature) of the road ahead and the requirement of driving comfort in a non-standard road environment, so as to avoid potential collision risks caused by complex road conditions. For example, in a curve or a road section with limited visibility, the electric drive controller calculates a reasonable target speed for speed limit by previewing the virtual lane information within a certain distance, and solves the first acceleration according to the reasonable target speed.

[0182] The electric drive controller actively reduces the vehicle speed based on the first acceleration to adapt to the change of path characteristics under the premise that no emergency obstacle is detected, thereby ensuring that the vehicle is decelerated to a safe range before entering the curve, and ensuring the effectiveness of vehicle control and driving comfort.

[0183] At step S312, a second acceleration is determined based on the second control strategy and the second operating parameter of the vehicle; the second control strategy is used for speed limit control of the vehicle based on the road boundary point.

[0184] The second control strategy is a speed limit method relying on the Free Space scatter point distribution characteristics, and the main purpose of the second control strategy is to analyze the density of the road boundary points in front of the vehicle when the lateral control is abnormal or the path cannot be effectively planned, to determine whether there is a potential collision risk, and to generate a corresponding speed limit instruction according to the result of determining whether there is a potential collision risk. The road boundary point refers to the edge point data of the passable area around the vehicle obtained by multi-sensor (such as camera, radar) fusion perception, and the road boundary points jointly constitute the contour of the drivable space of the vehicle.

[0185] For example, when there are a large number of dense road boundary points in front of the vehicle (indicating that there may be obstacles or irregular road areas), the electric drive controller can determine whether to enter the warning zone or the brake trigger zone by calculating the scatter point density function, and adjust the vehicle speed accordingly to prevent collision.

[0186] In the embodiments of the present application, there is a complementary relationship between the first control strategy and the second control strategy. When the first control strategy cannot provide effective speed limit suggestions due to no obvious curvature in front, the second control strategy can make supplementary decisions according to the distribution of the Free Space scatter points, thereby improving the robustness and environmental adaptability of the vehicle control method.

[0187] At step S313, a candidate acceleration is determined according to the first acceleration and the second acceleration.

[0188] The candidate acceleration is a preliminary control output value obtained after considering the first control strategy and the second control strategy. The calculation method of the candidate acceleration can be to take the minimum value of the first control strategy and the second control strategy, or to take their weighted average, depending on the priority judgment of the electric drive controller to different strategies in the current scene. For example, in the case of dense road boundary points and large curvature in front, the electric drive controller can select a more conservative speed limit scheme as the final candidate acceleration to ensure the safety and redundancy of the vehicle in the longitudinal control process.

[0189] In the embodiments of the present application, by introducing two different control strategies: the active speed limit control based on the virtual lane and the speed limit control based on the Free Space boundary point, hierarchical decision of the vehicle longitudinal speed is realized, so that the electric drive controller can cope with the complexity of non-standard road and the risk of lateral control failure at the same time, thereby improving the overall robustness and safety of the vehicle control method.

[0190] In some embodiments, the first operating parameter includes a current vehicle speed of the vehicle, a maximum lateral acceleration supported by the vehicle, a road surface adhesion coefficient, a current slope, etc.

[0191] The current vehicle speed of the vehicle refers to the actual driving speed of the vehicle at a certain time, which is used to determine whether the vehicle is in a high-speed or low-speed state, and the electric drive controller decides the response strength of longitudinal control according to the information.

[0192] The maximum lateral acceleration supported by the vehicle refers to the maximum lateral acceleration that the vehicle can withstand during turning, which is usually determined by the friction between the tire and the ground and is an important factor limiting the turning speed.

[0193] The road surface adhesion coefficient represents the friction performance between the tire and the road, and the coefficient affects the braking distance and steering stability of the vehicle.

[0194] The current slope refers to the inclination angle of the road section where the vehicle is located, which affects the power output and braking torque distribution of the vehicle.

[0195] The above operating parameters jointly determine the motion capability boundary of the vehicle in a specific environment and provide basic input for subsequent acceleration calculation. For example, on a wet road, even if the vehicle has a high lateral acceleration capability, the actual safe turning speed still needs to be reduced due to the decrease in the ground adhesion coefficient.

[0196] Step S311: determining a first acceleration based on the first control strategy and the first operating parameter of the vehicle, including steps 3111 to 3115, wherein:

[0197] Step 3111, determining the curvature radius of the road in front of the vehicle based on the preview distance of the vehicle.

[0198] The preview distance refers to the length that the vehicle observes forward when performing path prediction, which is used to perceive the change of the road in front, such as a curve, an intersection, etc. Through the preview distance, the geometric information of the road in front can be obtained, and the curvature radius of the road can be fitted. The curvature radius is a value used to describe the degree of road curvature, and when the value is small, it indicates that the curve is more acute, and the vehicle should slow down to ensure the safety of driving.

[0199] The embodiments of the present application utilize the preview technology to realize active perception of unstructured roads, so that the vehicle can adjust the speed in time before the curve to avoid loss of control due to sudden entry into a sharp curve.

[0200] It can be understood that there is a direct relationship between the preview distance and the curvature radius. The longer the preview distance, the more road information can be obtained, the higher the fitting accuracy of the curvature radius, and the higher the accuracy of subsequent path planning.

[0201] For example, assuming the preview distance of the vehicle is The radius of curvature of the road in front of the vehicle can be represented as:

[0202]

[0203] Step 3112, determining a first speed according to the radius of curvature and the maximum lateral acceleration supported by the vehicle;

[0204] The first speed is a theoretical speed limit value calculated based on the lateral motion capability of the vehicle.

[0205] For example, the maximum lateral acceleration supported by the vehicle can be represented as:

[0206]

[0207] wherein, is the active speed limit value to be solved, i.e., the first speed, is the preview time of the radius of curvature. The maximum lateral acceleration supported by the vehicle is usually a calibrated value, for example, a typical value is 1.5~2.0 .

[0208] The first speed can be represented as:

[0209]

[0210] The embodiment of the present application can provide a reasonable speed limit for the vehicle before the curve by determining the first speed based on the radius of curvature and the maximum lateral acceleration supported by the vehicle, and ensure that the vehicle can safely pass through the curve without side slipping.

[0211] Step 3113, determining a second speed according to the radius of curvature, the road adhesion coefficient and the current slope.

[0212] The second speed is a safe speed limit obtained after considering the road conditions (adhesion coefficient) and the terrain changes (slope). The calculation method of the second speed can use an empirical model or a physical model to comprehensively evaluate the optimal driving speed of the vehicle in a specific road condition.

[0213] wherein, the road adhesion coefficient can be estimated by camera texture recognition or radar echo, or the slip rate is calculated according to the wheel speed sensor and the IMU vehicle speed, and then estimated by the Burckhardt model.

[0214] The current slope can be calculated according to the longitudinal acceleration provided by the wheel speed differential and the displacement sensor.

[0215] Exemplarily, when there is a slope angle , the effective friction coefficient cos +sin ; the target speed upper limit, i.e., the second speed in the embodiments of the present application, can be expressed as:

[0216]

[0217] Step 3114, determining the minimum value between the first speed and the second speed as the target speed.

[0218] The target speed is the final target value for planning the longitudinal motion of the vehicle, and the smaller one of the two speeds is selected as the target speed to ensure that the most stringent safety requirements can be met under various constraint conditions, thereby effectively preventing the vehicle from driving at a speed higher than the speed limit due to misjudgment of a certain parameter.

[0219] Exemplarily, the target speed in the embodiments of the present application can be expressed as: wherein, is the first speed, is the second speed.

[0220] Step 3115, calculating the first acceleration based on the target speed and the current speed of the vehicle.

[0221] The first acceleration is a longitudinal acceleration calculated according to the difference between the target speed and the current speed of the vehicle, and is used to guide the acceleration or deceleration of the vehicle. The calculation method of the first acceleration can be to use a PID controller or other types of controllers to dynamically adjust the acceleration output according to the error size.

[0222] Exemplarily, the speed error between the target speed and the current speed of the vehicle can be expressed as:

[0223] -

[0224] Then, the first acceleration can be expressed as:

[0225]

[0226] wherein, are the proportional term, the integral term and the differential term of the PID control, respectively.

[0227] The above steps cooperate with each other to form a complete vehicle longitudinal control process. First, the vehicle behavior boundary is established by collecting the vehicle operating parameters and the road environment. Second, the road curvature information is obtained based on the preview technology, and the theoretical speed limit is calculated in combination with the vehicle performance parameters. Then, the safe speed is generated by considering factors such as road adhesion conditions and slope. Finally, the most stringent speed is selected as the target speed by comparison, and the appropriate longitudinal acceleration is calculated accordingly, completing the entire control process. This hierarchical progressive processing method not only ensures the clarity of the control logic, but also enhances the vehicle's ability to cope with complex conditions, improves the vehicle's longitudinal collision avoidance capability on non-standard roads, and enhances driving safety and comfort.

[0228] In some embodiments, the second operating parameter includes the start and end state of the vehicle in the second time period. The start and end state refers to the motion state information of the vehicle at the start time and the end time in a given time period, usually including position, speed, acceleration and the like. For example, in the braking scenario, the start and end state can be defined as the position, speed and acceleration at the target time as the initial state, and the position (such as the stationary point), speed of zero and acceleration of zero at the target time as the target state. By specifying the start and end state of the vehicle in the time interval, an accurate vehicle motion model can be established for subsequent trajectory prediction and optimization.

[0229] In some embodiments, the step S312 of determining the second acceleration based on the second control strategy and the second operating parameter of the vehicle includes steps 3121-3123, wherein:

[0230] Step 3121, generating a quintic polynomial trajectory equation according to the start and end state of the vehicle to obtain a candidate trajectory.

[0231] In the embodiments of the present application, a corresponding quintic polynomial trajectory equation can be generated based on the start and end state of the vehicle.

[0232] Illustratively, first, the braking time upper limit is determined according to the current vehicle speed and the maximum allowable deceleration .

[0233]

[0234] wherein, is a safety factor, is a preset minimum sampling time.

[0235] Then, the sampling interval (i.e., the second time period in the embodiments of the present application) is uniformly divided into N sampling points:

[0236]

[0237] Since the square integral of jerk (evaluating the comfort of acceleration and deceleration) needs to be minimized in both constrained and unconstrained cases, the sixth derivative of the optimization function needs to be zero. For optimization with constraints on initial and final positions, velocity and acceleration, a quintic polynomial is used, i.e.:

[0238]

[0239] where, is the position, is the coefficient of the quintic polynomial, is the corresponding sampling time.

[0240] The velocity and acceleration are respectively:

[0241]

[0242]

[0243] The initial state of the vehicle, i.e. the current state of the vehicle, includes the initial position , the initial velocity , and the initial acceleration .

[0244] For each time sampling , the corresponding final state can be obtained: the final position , the final velocity , and the final acceleration .

[0245] For Free Space scatter braking, it can be considered as a stationary point, and , , where is the distance between the vehicle and the scatter point. is the safety distance, which can be represented by a kinematic safety distance model, i.e.:

[0246]

[0247] where, is the system reaction time, including perception delay + control period; is the set static safety margin distance.

[0248] For each candidate braking time , the above initial and final states are brought into the polynomial and its derivatives to obtain the following equation set:

[0249]

[0250] For ease of programming and calculation, the above equation set is converted into a matrix form. Let

[0251]

[0252]

[0253]

[0254] Thus, the equation set can be expressed as , , are known, then for a certain sampling time the trajectory formed by solving the linear equation set, the corresponding quintic polynomial coefficient is .

[0255] Embodiments of the present application can construct a linear equation set and solve the polynomial coefficients by giving the initial and final states, thereby generating candidate trajectories. The candidate trajectories not only satisfy the basic motion constraints, but also ensure the continuity and comfort of the candidate trajectories, avoiding abrupt acceleration changes.

[0256] Step 3122, respectively calculating the cost function for each candidate trajectory.

[0257] The cost function is a mathematical expression for quantitatively evaluating the performance indicators of each candidate trajectory in multiple dimensions. Common cost items include comfort cost (such as jerk square integral), efficiency cost (such as braking time), termination position deviation cost (deviation from the safety distance), etc. The total cost function is usually a weighted sum of the above single cost, used to measure the overall advantages and disadvantages of the trajectory.

[0258] Exemplarily, for each candidate trajectory (corresponding to each candidate braking time ), the following four costs are calculated:

[0259] Comfort cost (jerk term):

[0260]

[0261] Efficiency cost (braking time):

[0262]

[0263] Termination position deviation cost (position deviation from the safety distance ):

[0264]

[0265]

[0266] Total cost function:

[0267]

[0268] The application can compare different trajectory schemes by calculating the cost function value of each candidate trajectory, and screen out the optimal solution.

[0269] It can be understood that in actual application, the weights of each cost in the cost function can be adjusted according to the specific application scene, so as to adapt to different driving conditions and user preferences. For example, in the emergency braking scene, the safety-related cost term should be given a higher weight; and in the ordinary cruising scene, more attention should be paid to comfort and energy consumption, and the comfort-related cost can be appropriately increased.

[0270] Step 3123, determining the second acceleration according to the cost function and the dynamic constraint condition corresponding to the vehicle.

[0271] The dynamic constraint condition refers to the physical limit that must be followed by the vehicle in the actual running process, including maximum acceleration, minimum deceleration, etc. The dynamic constraint condition reflects the mechanical characteristics of the vehicle and the limit ability of the control system. The embodiments of the application combine the cost function with the dynamic constraint condition, so that the optimal control acceleration, i.e. the second acceleration in the embodiments of the application, can be found under the premise of meeting the physical feasibility.

[0272] Exemplarily, a coarse search method can be used to quickly obtain an approximate optimal solution:

[0273] According to the vehicle dynamics constraint, a candidate acceleration set (coarse solution set) is generated, and the acceleration search calculation is performed:

[0274]

[0275] The acceleration candidate value lower limit and upper limit are respectively. The precision of the acceleration candidate set is Decided according to the hardware computing power.

[0276] For each , the corresponding total cost is calculated.

[0277] Sort in ascending order. For a hardware platform with low computing power, the minimum can be found by binary search. The corresponding is the optimal acceleration ; for a high-computing-power platform, the three with the lowest cost are selected to enter the fine search stage.

[0278] For a high-computing-power platform, sequential quadratic programming (SQP) can be used for fine search:

[0279] a. Set constraints

[0280] b. Set convergence threshold and maximum number of iterations , and guess an initial solution from the 3 coarse solutions.

[0281] c. Convert the objective function into the SQP standard form

[0282]

[0283] d. Solve using the QP solver to get the search direction

[0284] e. Determine the step size by the Armijo criterion :

[0285]

[0286] Exemplarily, the coefficient can be taken as 0.1, the initial step size can be taken as 0.1, and the step size can be attenuated by 0.5 times.

[0287] f. Convergence judgment, if any of the following conditions is met, the iteration is terminated, and the optimal acceleration is obtained , that is, the second acceleration in the embodiment of the application:

[0288]

[0289] The second acceleration finally calculated by the embodiment of the application not only considers the control effect, but also fully respects the physical limit of the vehicle, ensures the robustness and reliability of the longitudinal control system under various working conditions, can be applied to complex scenes such as non-standard roads or lateral control failure, and effectively improves the safety and stability of the longitudinal control.

[0290] Next, the vehicle control method provided by the application will be described in detail in combination with a specific embodiment.

[0291] In the specific implementation process, first, the dynamic generation of the free space and the virtual lane is performed.

[0292] Step 201, acquiring the road geometric features in front of the vehicle through the front-view camera of the vehicle.

[0293] Step 202, extracting the road boundary points based on the road geometric features.

[0294] Step 203, detecting the point cloud data corresponding to the obstacles in front of the vehicle through the millimeter wave radar.

[0295] ​​Step 204: Fit a virtual lane in front of the vehicle based on road boundary points and point cloud data.

[0296] In this embodiment, road geometric features (including roadside boundaries, semantic segmentation of irregular obstacles, etc.) are acquired by a forward-looking camera to obtain visual roadside lines (expressed as cubic polynomials) and visual free space points; millimeter-wave radar is used to detect point cloud data of obstacles in front to obtain radar guardrail lines (expressed as cubic polynomials) and radar free space points.

[0297] In this embodiment, the free space points on both sides of the vehicle are fitted into a curve equation. The guardrail line output by the radar is preferentially selected as the baseline, followed by the forward-looking roadside line. For example, free space points with a lateral distance less than a preset threshold from the baseline are selected, and outliers are removed using the RANSAC algorithm to obtain a set of interior points. The least squares method is then used to fit a cubic polynomial curve of the free space points on the left and right sides of the vehicle using the interior point set, thus creating a virtual lane.

[0298] For example, in the embodiments of this application, the least squares method can be used to fit a set of interior points to a cubic polynomial curve of the free space points on the left and right sides of the vehicle as a virtual lane.

[0299] For all interior points ( ), construct matrix ,vector sum vector :

[0300]

[0301] The least squares solution is .

[0302] The fitted cubic polynomial curve can be represented as:

[0303]

[0304] Furthermore, the fitting results can be validated. Specifically, the mean squared error (MSE) of the fitted sample can be calculated:

[0305]

[0306] in, .

[0307] If the left and right lines All are greater than the set threshold. Perform a safe braking maneuver with a constant deceleration and report to the control unit.

[0308] For at least one of the left and right lines, the following condition must be met. The center line is calculated as a reference path based on the fitting result of the lane marking:

[0309]

[0310] wherein, is a preset lane width.

[0311] Then, a lateral control health state assessment is performed:

[0312] In step 205, it is assessed whether the lateral control is normal based on the fault diagnosis information of the steering gear and the error data of the lateral control module.

[0313] Specifically, it can be determined whether the lateral control is in a healthy state by detecting the steering angle tracking error, the heading angle tracking error and the lateral position tracking error. The steering angle tracking error, the heading angle tracking error and the lateral position tracking error can be expressed as:

[0314] Steering angle tracking error

[0315] Heading angle tracking error

[0316] Lateral position tracking error

[0317] If are simultaneously satisfied, it can be determined that the lateral control is normal, i.e., the steering system is in a normal state.

[0318] If any of the above errors exceeds a set threshold, it is determined that the result of the state assessment is that the steering system is in an abnormal steering state.

[0319] Subsequently, a longitudinal collision avoidance region is divided:

[0320] In step 206, an avoidance path is planned based on the state information and the virtual lane to determine a target avoidance path.

[0321] In some embodiments, for the lateral control health state, the target avoidance path is the reference path ; for the lateral control abnormal state, the target avoidance path is the predicted path, in which case the vehicle is considered to follow a semicircular arc, and the radius of the semicircular arc is , wherein is the steering ratio, is the steering wheel angle, is the wheelbase of the vehicle.

[0322] In step 207, a first region is determined according to the target avoidance path and the current speed of the vehicle.

[0323] In some embodiments, the width of the first region is a region with a width w on the left and right of the target collision avoidance path, and the length of the first region is a distance according to the vehicle speed preview, that is, the length of the first region is the preview distance of the vehicle wherein is the longitudinal speed of the vehicle, is the preview time.

[0324] Step 208, statistics the distribution of the road boundary points in the first region.

[0325] In some embodiments, the scatter point density in the first region (Free Space) is calculated Further, the Free Space can be divided into a low-risk area (ρ ), a warning area (ρ ≤ρ≤ ), a brake triggering area (ρ )

[0326] Step 209, constructing a risk value function corresponding to the first region based on the distribution of the road boundary points.

[0327] In some embodiments, the risk value function can be expressed as:

[0328]

[0329] wherein, is the calculated Free Space scatter point density, is the trend of the change of the scatter point density with time, that is, the first parameter in the embodiments of the present application, which can represent the change of the scatter point distribution caused by the movement of the obstacle (such as an animal crossing the road), is the spatial gradient of the scatter point density, which can represent the spatial distribution of the scatter points (which can effectively suppress the false triggering of static interference such as shrubs). are the corresponding weight coefficients, respectively.

[0330] In some embodiments, can be calculated by finite difference and sliding average filtering. can be calculated by grid division of the statistical area S to calculate the density difference of adjacent grids.

[0331] Next, continue to make a hierarchical speed planning decision:

[0332] Step 210, determining a first acceleration based on the first control strategy and the first running parameter of the vehicle; the first control strategy is used for active speed limiting control of the vehicle.

[0333] In some embodiments, it is assumed that the preview distance of the vehicle is The curvature radius of the road in front of the vehicle can be represented as:

[0334]

[0335] The maximum lateral acceleration supported by the vehicle may be represented as:

[0336]

[0337] wherein, is the active speed limit value to be solved, i.e., the first speed, is the preview time of the curvature radius. The maximum lateral acceleration supported by the vehicle is usually a calibrated value, for example, a typical value is 1.5-2.0 .

[0338] The first speed can be represented as:

[0339]

[0340] When there is a slope angle , the effective friction coefficient cos +sin ; the upper limit of the target speed, i.e., the second speed in the embodiments of the present application, can be represented as:

[0341]

[0342] The target speed is calculated as: wherein, is the first speed, is the second speed.

[0343] The speed error between the target speed and the current speed of the vehicle can be represented as:

[0344] -

[0345] The first acceleration can be represented as:

[0346]

[0347] wherein, are the proportional term, the integral term and the differential term of the PID control, respectively.

[0348] Step 211, determining a second acceleration based on the second control strategy and the second operating parameter of the vehicle; the second control strategy is used for speed limit control of the vehicle based on the road boundary point.

[0349] In some embodiments, the current vehicle speed is first determined according to the current vehicle speed and the maximum allowable deceleration , to determine the upper limit of the braking time:

[0350]

[0351] wherein, is the safety factor, is the preset minimum sampling time.

[0352] The sampling interval (i.e., the second time period in the embodiments of the present application) is then evenly divided into N sampling points:

[0353]

[0354] Since, in the case of constraints or no constraints, to make the square integral of the acceleration jerk (evaluating the comfort of acceleration and deceleration) minimum, the sixth derivative of the optimization function needs to be 0. For the optimization of the initial and final positions, speed and acceleration constraints, a quintic polynomial is used, i.e.:

[0355]

[0356] wherein, is the position, is the coefficient of the quintic polynomial, is the corresponding sampling time.

[0357] The speed and acceleration thereof are respectively:

[0358]

[0359]

[0360] The initial state of the vehicle, i.e., the current state of the vehicle, includes the initial position , the initial speed , and the initial acceleration .

[0361] For each time sampling , the corresponding final state can be obtained, i.e., the final position , the final speed , and the final acceleration .

[0362] For the Free Space scatter point braking, it can be considered as a stationary point, and , , wherein is the distance between the vehicle and the scatter point. is the safety distance, which can be represented by a kinematic safety distance model, i.e.

[0363]

[0364] where, is the system reaction time, including perception delay + control period; is the set static safety margin distance.

[0365] For each candidate braking time , the initial and final states are brought into the polynomial and its derivative, and the following equation set can be obtained:

[0366]

[0367] For ease of programming and calculation, the above equation set is converted into matrix form. Let

[0368]

[0369]

[0370]

[0371] Then, the equation set can be expressed as , , All are known, then for a certain sampling time , the trajectory formed by the linear equation set is solved, and the corresponding quintic polynomial coefficient is .

[0372] For each candidate trajectory (corresponding to each candidate braking time ), the following four costs are calculated:

[0373] Comfort cost (jerk term):

[0374]

[0375] Efficiency cost (braking time):

[0376]

[0377] Termination position deviation cost (deviation from the position of the safety distance ):

[0378]

[0379]

[0380] Total cost function:

[0381]

[0382] The application can compare different trajectory schemes by calculating the cost function value of each candidate trajectory, and screen the optimal solution.

[0383] In some embodiments, a coarse search method can be used to quickly obtain an approximate optimal solution:

[0384] According to the vehicle dynamics constraint, a candidate acceleration set (coarse solution set) is generated, and the acceleration search calculation is performed:

[0385] ,

[0386] The lower and upper limits of the acceleration candidate value are respectively. The accuracy of the acceleration candidate set is determined according to the hardware computing power.

[0387] For each , the corresponding total cost is calculated.

[0388] Sort the in ascending order. For hardware platforms with low computing power, the minimum can be found by binary search. The corresponding is the optimal acceleration ; for high-computing-power platforms, the three with the lowest cost can be selected to enter the fine search stage.

[0389] For high-computing-power platforms, sequential quadratic programming (SQP) can be used for fine search:

[0390] a. Set the constraint condition

[0391] b. Set the convergence threshold and the maximum number of iterations , and guess an initial solution from the three coarse solutions.

[0392] c. Convert the objective function into the standard form of SQP

[0393]

[0394] d. Use the QP solver to solve and get the search direction

[0395] e. Determine the step size by Armijo criterion:

[0396]

[0397] Exemplarily, the coefficient can be taken as , initial step length , attenuated by 0.5 times.

[0398] f. Convergence judgment: if any of the following conditions is met, terminate iteration and obtain the optimal acceleration , i.e., the second acceleration in the embodiment of the application:

[0399]

[0400] Step 212: determining a candidate acceleration according to the first acceleration and the second acceleration.

[0401] In some embodiments, the candidate acceleration can be calculated by taking the minimum value of the first control strategy and the second control strategy, or by taking their weighted average, depending on the priority judgment of the electric drive controller on different strategies in the current scenario. For example, in the case where the road boundary points are dense and the curvature in front is large, the electric drive controller can select a more conservative speed limit scheme as the final candidate acceleration to ensure the safety and redundancy of the vehicle in the longitudinal control process.

[0402] Step 213: dynamically gain adjusting the candidate acceleration by using the risk value function to obtain a target acceleration.

[0403] In some embodiments, the calculated candidate acceleration is dynamically gain adjusted by using the risk value function to obtain a target acceleration. Exemplarily, the target acceleration in the embodiment of the application can be expressed as:

[0404]

[0405] wherein, is a gain coefficient, is a nonlinear adjustment factor.

[0406] Step 214: controlling the vehicle based on the target acceleration.

[0407] Finally, the calculated target acceleration is subjected to physical limit constraint, and the maximum acceleration constraint can be expressed as: . The electric drive controller executes deceleration .

[0408] Further, in the embodiment of the application, the electric drive controller periodically executes the foregoing steps, updates the environmental perception data and control parameters, and calculates the corresponding optimal control deceleration, and controls the vehicle based on the optimal deceleration.

[0409] Referring to Figure 2 , a structural schematic diagram of a vehicle control device provided by the embodiment of the application is shown. As shown inFigure 2 As shown in the figure, the vehicle control device 20 comprises an acquisition unit 21, a first determination unit 22, a second determination unit 23 and a control unit 24, wherein:

[0410] The acquisition unit 21 is configured to, in response to detecting that the steering system of the vehicle fails, acquire state information of the steering system and a virtual lane in front of the vehicle;

[0411] The first determination unit 22 is configured to determine a risk value function corresponding to a first region based on the state information and the virtual lane; the first region is a predicted collision avoidance region;

[0412] The second determination unit 23 is configured to determine a target acceleration of the vehicle according to the risk value function and an operating parameter of the vehicle;

[0413] The control unit 24 is configured to control the vehicle based on the target acceleration.

[0414] In some embodiments, the first determination unit is further configured to: determine a target collision avoidance path based on the state information and the virtual lane for collision avoidance path planning; determine the first region according to the target collision avoidance path and a current vehicle speed of the vehicle; count a distribution of road boundary points in the first region; and construct the risk value function corresponding to the first region based on the distribution.

[0415] In some embodiments, the distribution includes a scatter point density of road boundary points in the first region, a first parameter corresponding to the scatter point density and a spatial gradient; the first parameter is used to represent a variation trend of the scatter point density in a first time period; and the first determination unit is further configured to: respectively determine a weight value corresponding to each of the scatter point density, the first parameter and the spatial gradient; and perform weighted summation on the scatter point density, the first parameter and the spatial gradient to obtain the risk value function corresponding to the first region.

[0416] In some embodiments, the first determination unit is further configured to: determine whether the steering system is abnormal based on the state information; determine the target collision avoidance path based on the virtual lane in a case where the steering system is normal; and determine the target collision avoidance path based on a wheelbase, a steering ratio and a current steering wheel angle of the vehicle in a case where the steering system is abnormal.

[0417] In some embodiments, the first determination unit is further configured to: extend a left side and a right side of the target collision avoidance path by a preset width respectively to obtain a width of the first region; and determine a length of the first region based on a current vehicle speed of the vehicle and a preview time.

[0418] In some embodiments, the second determining unit is further configured to: calculate a candidate acceleration according to a pre-set control strategy and an operating parameter of the vehicle; and perform dynamic gain adjustment on the candidate acceleration by using the risk value function to obtain a target acceleration.

[0419] In some embodiments, the second determining unit is further configured to: determine a first acceleration based on a first control strategy and a first operating parameter of the vehicle; the first control strategy is used for active speed limiting control of the vehicle; determine a second acceleration based on a second control strategy and a second operating parameter of the vehicle; the second control strategy is used for speed limiting control of the vehicle based on road boundary points; and determine the candidate acceleration according to the first acceleration and the second acceleration.

[0420] In some embodiments, the first operating parameter includes a current vehicle speed of the vehicle, a maximum lateral acceleration supported by the vehicle, a road surface adhesion coefficient, and a current slope; and the second determining unit is further configured to: determine a curvature radius of a road in front of the vehicle based on a look-ahead distance of the vehicle; determine a first speed according to the curvature radius and the maximum lateral acceleration; determine a second speed according to the curvature radius, the road surface adhesion coefficient, and the current slope; determine a minimum value between the first speed and the second speed as a target speed; and calculate the first acceleration based on the target speed and the current vehicle speed of the vehicle.

[0421] In some embodiments, the second operating parameter includes initial and final states of the vehicle in a second time period; and the second determining unit is further configured to: generate a quintic polynomial trajectory equation according to the initial and final states to obtain a candidate trajectory; calculate a cost function for each candidate trajectory respectively; and determine the second acceleration according to the cost function and a corresponding dynamic constraint condition of the vehicle.

[0422] In some embodiments, the obtaining unit is further configured to: obtain road geometric features in front of the vehicle by using a front-view camera of the vehicle; extract road boundary points based on the road geometric features; detect point cloud data corresponding to obstacles in front of the vehicle by using a millimeter wave radar; and fit a virtual lane in front of the vehicle according to the road boundary points and the point cloud data.

[0423] Based on the above embodiments, the embodiments of the present application further provide a vehicle control system, Figure 3 A schematic diagram of a composition structure of a vehicle control system provided by the embodiments of the present application is shown in FIG. 3. Figure 3 As shown in FIG. 3, the vehicle control system 30 includes:

[0424] a steering system 31 of the vehicle, configured to send a steering failure signal to the electric drive controller in a case where a failure is identified.

[0425] The electric drive controller 32 is configured to, in response to the steering failure signal, acquire state information of the steering system and a virtual lane in front of the vehicle; determine a risk value function corresponding to a first region based on the state information and the virtual lane; the first region is a predicted collision avoidance region; determine a target acceleration of the vehicle according to the risk value function and an operating parameter of the vehicle; and control the vehicle based on the target acceleration.

[0426] The descriptions of the vehicle control device and the vehicle control system embodiments are similar to the descriptions of the vehicle control method embodiments, and have similar beneficial effects. For technical details not disclosed in the vehicle control device and the vehicle control system embodiments, please refer to the descriptions of the vehicle control method embodiments for understanding.

[0427] It should be noted that, in the embodiments of the present application, if the above method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product in essence or the part that contributes to the related art, which is stored in a storage medium, includes a number of instructions for making an electronic device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory, a magnetic disk or an optical disk, and various program code storage media. Therefore, the embodiments of the present application are not limited to any specific hardware and software combination.

[0428] The present application also provides an electronic device, including a memory and a processor, the memory stores a computer program capable of running on the processor, and the processor executes the computer program to realize the above method.

[0429] The present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the above method. The computer readable storage medium can be transitory or non-transitory.

[0430] The application further provides a computer program product comprising computer programs or instructions, which, when executed by a processor, implement some or all of the steps of the above method. The computer program product can be implemented in particular by hardware, software or a combination thereof. The computer program product can be implemented in particular by hardware, software or a combination thereof. In an optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK) or the like.

[0431] It should be noted that, Figure 4 A hardware entity diagram of an electronic device provided by an embodiment of the application is shown in FIG. 4, which includes a processor 401, a communication interface 402 and a memory 403, wherein: Figure 4 The hardware entity of the electronic device 400 includes a processor 401, a communication interface 402 and a memory 403, wherein:

[0432] The processor 401 generally controls the overall operation of the electronic device 400.

[0433] The communication interface 402 can enable the electronic device 400 to communicate with other terminals or servers through a network.

[0434] The memory 403 is configured to store instructions and applications executable by the processor 401, and can also cache data (e.g., image data, audio data, voice communication data and video communication data) to be processed by the processor 401 and modules in the electronic device 400, which can be implemented by FLASH or RAM. The processor 401, the communication interface 402 and the memory 403 can transmit data through a bus 404.

[0435] Here, the electronic device can be a car machine in a vehicle.

[0436] It should be noted that the above description of the storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of the application, please refer to the description of the method embodiments of the application.

[0437] It should be understood that every feature and combination of features that is described above in relation to one embodiment is applicable to at least one other embodiment, unless specifically stated otherwise. It should also be understood that every embodiment described above can be combined with any other embodiment unless specifically stated otherwise.

[0438] It should be noted that, as used in this application, the terms "comprises" or "comprising," or the like are used in the sense of "including" and not of "consisting only of," such that the process, method, article, or apparatus that includes elements in addition to those listed after such a term in this application are still within the scope of that process, method, article, or apparatus. Where the term "comprises" is used in the form "comprising" or "comprises", it is intended that the process, method, article, or apparatus include at least the recited elements, but not excluding others.

[0439] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of the above-mentioned units is only a logical function division, and actual implementation can have another division manner, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed components can be through some interfaces, indirect coupling or communication connection between devices or units, which can be electrical, mechanical or other forms.

[0440] The units described above as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units; they can be located in one place or distributed on multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0441] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0442] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program executes the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium includes various storage media that can store program codes, such as mobile storage devices, read-only memories, magnetic discs or optical discs.

[0443] Alternatively, the integrated units of the present application can be stored in a computer readable storage medium if they are realized in the form of software function modules and sold or used as independent products. Based on such understanding, the technical solutions of the present application or the parts that make contributions to the related art can be embodied in the form of software products, and the computer software products are stored in a storage medium, including a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the embodiments of the method of the present application. The foregoing storage medium includes various storage media that can store program codes, such as mobile storage devices, ROM, magnetic discs or optical discs, etc.

[0444] The above is only an embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application.

Claims

1. A vehicle control method characterized by, The application relates to an electric drive controller for a vehicle, and the method comprises the following steps: in response to detecting that a steering system of the vehicle fails, acquiring state information of the steering system and a virtual lane in front of the vehicle; based on the state information and the virtual lane, determining a risk value function corresponding to a first region; the first region is a predicted collision avoidance region; determining a target acceleration of the vehicle according to the risk value function and operating parameters of the vehicle; controlling the vehicle based on the target acceleration; wherein the calculation of the risk value function depends on the distribution of road boundary points in the first region; the distribution includes the scatter point density of the road boundary points in the first region, a first parameter corresponding to the scatter point density and a spatial gradient; the first parameter is used for representing the change trend of the scatter point density in a first time period; the target acceleration is dynamically adjusted according to the real-time updated risk value function.

2. The method of claim 1, wherein, The method comprises the following steps: based on the state information and the virtual lane, performing collision avoidance path planning to determine a target collision avoidance path; determining the first region according to the target collision avoidance path and the current speed of the vehicle; statistically analyzing the distribution of road boundary points in the first region; constructing the risk value function corresponding to the first region based on the distribution.

3. The method of claim 2, wherein, The method comprises the following steps: respectively determining the weight values corresponding to the scatter point density, the first parameter and the spatial gradient; performing weighted summation on the scatter point density, the first parameter and the spatial gradient to obtain the risk value function corresponding to the first region.

4. The method of claim 2, wherein, The method comprises the following steps: judging whether the steering system is abnormal based on the state information; in the case that the steering system is normal, determining the target collision avoidance path based on the virtual lane; in the case that the steering system is abnormal, determining the target collision avoidance path based on the wheelbase, the steering ratio and the current steering wheel angle of the vehicle.

5. The method of claim 2, wherein, The method comprises the following steps: extending the left side and the right side of the target collision avoidance path by a preset width to obtain the width of the first region; determining the length of the first region based on the current speed of the vehicle and the preview time.

6. The method of claim 1, wherein, The method comprises the following steps: calculating a candidate acceleration according to a pre-set control strategy and the operating parameters of the vehicle; performing dynamic gain adjustment on the candidate acceleration by using the risk value function to obtain the target acceleration.

7. The method of claim 6, wherein, The method comprises the following steps: determining a first acceleration based on a first control strategy and first operating parameters of the vehicle; the first control strategy is used for actively controlling the speed of the vehicle. determine a second acceleration based on a second control strategy and second operating parameters of the vehicle; the second control strategy is used for speed limiting control of the vehicle based on road boundary points; determine the candidate acceleration according to the first acceleration and the second acceleration.

8. The method of claim 7, wherein, the first operating parameters include a current vehicle speed of the vehicle, a maximum lateral acceleration supported by the vehicle, a road surface adhesion coefficient, and a current slope; determine a first acceleration based on a first control strategy and first operating parameters of the vehicle, including: determine a curvature radius of a road in front of the vehicle based on a preview distance of the vehicle; determine a first speed according to the curvature radius and the maximum lateral acceleration; determine a second speed according to the curvature radius, the road surface adhesion coefficient, and the current slope; determine a target speed as a minimum value of the first speed and the second speed; calculate the first acceleration based on the target speed and a current vehicle speed of the vehicle.

9. The method of claim 7, wherein, the second operating parameters include initial and final states of the vehicle in a second time period; determine a second acceleration based on a second control strategy and second operating parameters of the vehicle, including: generate a quintic polynomial trajectory equation according to the initial and final states to obtain a candidate trajectory; calculate a cost function for each candidate trajectory respectively; determine the second acceleration according to the cost function and corresponding dynamic constraints of the vehicle.

10. The method according to any one of claims 1 to 9, characterized in that, the state information of the steering system and the virtual lane in front of the vehicle include: obtain road geometric features in front of the vehicle through a front-view camera of the vehicle; extract road boundary points based on the road geometric features; detect point cloud data corresponding to obstacles in front of the vehicle through a millimeter wave radar; fit the virtual lane in front of the vehicle according to the road boundary points and the point cloud data.

11. A vehicle control device characterized by comprising: An electric drive controller applied to a vehicle, the method comprising: an acquisition unit configured to, in response to detecting a failure of a steering system of the vehicle, acquire state information of the steering system and a virtual lane in front of the vehicle; a first determination unit configured to determine a risk value function corresponding to a first region based on the state information and the virtual lane; the first region is a predicted collision avoidance region; a second determination unit configured to determine a target acceleration of the vehicle according to the risk value function and operating parameters of the vehicle; a control unit configured to control the vehicle based on the target acceleration; wherein the calculation of the risk value function depends on a distribution of road boundary points in the first region; the distribution includes a scatter point density of the road boundary points in the first region, a first parameter corresponding to the scatter point density, and a spatial gradient; the first parameter is used to represent a change trend of the scatter point density in a first time period; the target acceleration is dynamically adjusted according to the real-time updated risk value function.

12. A vehicle control system comprising a steering system and an electric drive controller of a vehicle, characterized in that: the steering system of the vehicle is configured to send a steering failure signal to the electric drive controller in the case of identifying a failure; The electric drive controller is configured to acquire state information of the steering system and a virtual lane in front of the vehicle in response to the steering failure signal; determine a risk value function corresponding to a first region based on the state information and the virtual lane; the first region is a predicted collision avoidance region; determine a target acceleration of the vehicle according to the risk value function and an operating parameter of the vehicle; control the vehicle based on the target acceleration; wherein the calculation of the risk value function depends on a distribution of road boundary points in the first region; the distribution includes a scatter point density of the road boundary points in the first region, a first parameter corresponding to the scatter point density, and a spatial gradient; the first parameter is used to represent a change trend of the scatter point density in a first time period; the target acceleration is dynamically adjusted according to the real-time updated risk value function.

13. An electronic device, comprising: The processor and the memory, the memory stores a computer program capable of running on the processor, and the processor executes the computer program to realize the steps in the method of any one of claims 1-10.

14. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is executed by a processor to realize the steps in the method of any one of claims 1-10.

15. A computer program product comprising computer programs or instructions, characterized in that, The computer program or instructions are executed by a processor to realize the method of any one of claims 1-10.

Images