Vehicle control method and device and storage medium

Abstract

The invention relates to the field of automatic driving, and discloses a vehicle control method and device and a storage medium, and the method comprises the steps: obtaining the longitudinal motion information of each reference vehicle; wherein the longitudinal motion information comprises a longitudinal distance, a longitudinal speed and a longitudinal acceleration with the current vehicle; inputting the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle into a pre-trained acceleration model to obtain a first acceleration; taking the reference vehicle with the minimum longitudinal distance in the transverse overlapping region corresponding to the current vehicle as a main target vehicle, and determining a second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle; and determining a target acceleration according to the first acceleration and the second acceleration, and controlling the current vehicle based on the target acceleration. According to the technical scheme, the effect of effectively controlling the vehicle under the safety constraint in combination with the driving habit of a driver is achieved.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and in particular to a vehicle control method, device and storage medium. Background Art

[0002] The cruise mode in autonomous driving can simulate human driving habits to control the vehicle. It usually uses a deep learning model to process the motion information of the vehicle and surrounding vehicles to obtain vehicle control information.

[0003] However, in the process of model processing, since human driving habits cannot be completely replicated and there are errors in the calculation process of the model, it will lead to inaccurate vehicle control and even cause driving safety problems.

[0004] In view of this, the present invention is proposed. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a vehicle control method, device and storage medium, which can realize the effect of effectively controlling the vehicle under safety constraints in combination with the driver's driving habits.

[0006] An embodiment of the present invention provides a vehicle control method, the method comprising:

[0007] Acquire longitudinal motion information of each reference vehicle; wherein the longitudinal motion information includes the longitudinal distance between the reference vehicle and the current vehicle, the longitudinal speed and the longitudinal acceleration;

[0008] Inputting the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle into a pre-trained acceleration model to obtain a first acceleration;

[0009] Taking the reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle as the main target vehicle, and determining a second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle;

[0010] A target acceleration is determined according to the first acceleration and the second acceleration, and the current vehicle is controlled based on the target acceleration.

[0011] An embodiment of the present invention provides an electronic device, the electronic device comprising:

[0012] Processor and memory;

[0013] The processor is used to execute the steps of the vehicle control method described in any embodiment by calling the program or instructions stored in the memory.

[0014] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a program or instruction, wherein the program or instruction enables a computer to execute the steps of the vehicle control method described in any embodiment.

[0015] The embodiments of the present invention have the following technical effects:

[0016] By acquiring the longitudinal motion information of each reference vehicle, the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle are input into a pre-trained acceleration model to obtain a first acceleration, so as to predict the acceleration value through a deep learning model, and then, the reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle is taken as the main target vehicle, and the second acceleration is determined according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle, so as to calculate the acceleration value through a kinematic formula, and further, the target acceleration is determined according to the first acceleration and the second acceleration, so as to correct the second acceleration through the first acceleration, and the current vehicle is controlled based on the target acceleration, thereby achieving the effect of combining the driver's driving habits and effectively controlling the vehicle under safety constraints. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 is a flow chart of a vehicle control method provided by an embodiment of the present invention;

[0019] Figure 2 is a flow chart of another vehicle control method provided by an embodiment of the present invention;

[0020] Figure 3 is a structural schematic diagram of a vehicle control system provided by an embodiment of the present invention;

[0021] Figure 4 It is a schematic diagram of a processing flow of a controller of a vehicle control system provided by an embodiment of the present invention;

[0022] Figure 5 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0024] The vehicle control method provided by the embodiment of the present invention is mainly applicable to simulating the driving habits of the driver in the cruise mode and imposing safety constraints on the driver. The vehicle control method provided by the embodiment of the present invention can be executed by an electronic device integrated in the vehicle or independent of the vehicle.

[0025] Figure 1 is a flow chart of a vehicle control method provided by an embodiment of the present invention. Figure 1 , the vehicle control method specifically includes:

[0026] S110. Obtain longitudinal motion information of each reference vehicle.

[0027] The reference vehicle is a vehicle located in front of the current vehicle (including the left front, front and right front) that may affect the driving of the current vehicle. The current vehicle is a vehicle that executes the vehicle control method. The longitudinal motion information is motion-related information in the longitudinal direction of travel. Specifically, the longitudinal motion information includes the longitudinal distance, longitudinal speed and longitudinal acceleration from the current vehicle.

[0028] Specifically, the longitudinal motion information of each reference vehicle corresponding to the current vehicle may be obtained through sensors installed on the current vehicle, such as radars and cameras, for subsequent evaluation and calculation of the target acceleration of the current vehicle.

[0029] Based on the above example, the longitudinal motion information of each reference vehicle can be obtained in the following way:

[0030] In response to receiving the smart cruise signal, the smart camera of the current vehicle is activated;

[0031] Based on the intelligent camera, the longitudinal distance between each reference vehicle and the current vehicle, the longitudinal speed of each reference vehicle and the longitudinal acceleration of each reference vehicle are obtained;

[0032] For each reference vehicle, the longitudinal distance between the reference vehicle and the current vehicle, the longitudinal speed of the reference vehicle, and the longitudinal acceleration of the reference vehicle are used as the longitudinal motion information of the reference vehicle.

[0033] Among them, the smart cruise signal is the signal to enter the smart cruise function in the automatic driving. Under the smart cruise function, the target acceleration of the current vehicle needs to be planned and controlled. The smart camera is a device that combines artificial intelligence technology to analyze and process the content of the shooting to determine the motion information.

[0034] Specifically, when a smart cruise signal is received, it is determined that the smart cruise function of the automatic driving needs to be entered. Therefore, the motion information of other vehicles around the current vehicle needs to be captured, so the smart camera of the current vehicle is started. Then, the longitudinal distance between each reference vehicle and the current vehicle, the longitudinal speed of each reference vehicle, and the longitudinal acceleration of each reference vehicle are obtained through the smart camera. For each reference vehicle, the longitudinal motion information of the reference vehicle needs to be determined. Therefore, the longitudinal distance between the reference vehicle and the current vehicle, the longitudinal speed of the reference vehicle, and the longitudinal acceleration of the reference vehicle are integrated together as the longitudinal motion information of the reference vehicle, thereby obtaining the longitudinal motion information of each reference vehicle respectively.

[0035] S120 , inputting the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle into a pre-trained acceleration model to obtain a first acceleration.

[0036] The acceleration model is a deep learning model, which is used to predict the acceleration of the vehicle by combining the longitudinal motion information of the vehicles in front of the vehicle and the vehicle speed of the vehicle. Optionally, the acceleration model can be a long short-term memory network (Long Short-Term Memory, LSTM), which is a time recurrent neural network that is good at short-term learning, processing and classification of sequential data.

[0037] Specifically, the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle are input into a pre-trained acceleration model. Through analysis and reasoning by the acceleration model, a prediction result for the acceleration of the current vehicle can be output, and the prediction result is used as the first acceleration.

[0038] Based on the above example, the acceleration model can be trained in the following way:

[0039] Obtaining the vehicle speed and artificial acceleration of each reference vehicle, and obtaining longitudinal motion information of the sample vehicles corresponding to each reference vehicle;

[0040] The vehicle speed of each reference vehicle and the longitudinal motion information of the sample vehicle corresponding to each reference vehicle are respectively input into the untrained acceleration model to obtain each output acceleration;

[0041] Based on the preset loss function, the artificial acceleration and output acceleration of each benchmark vehicle are calculated to determine the target loss of the acceleration model that has not been trained;

[0042] If the target loss is less than the preset loss, the untrained acceleration model is used as the pre-trained acceleration model;

[0043] If the target loss is greater than or equal to the preset loss, the model parameters of the untrained acceleration model are adjusted, the untrained acceleration model is updated, and the step of returning to the step of inputting the vehicle speed of each benchmark vehicle and the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle into the untrained acceleration model to obtain each output acceleration is performed until the target loss is less than the preset loss.

[0044] Among them, the reference vehicle is the manually controlled vehicle to be simulated, corresponding to the current vehicle. The artificial acceleration is the acceleration value used by the driver when driving the reference vehicle. The sample vehicles are the vehicles located in front of the reference vehicle (including the left front, the front and the right front) that will affect the driving of the reference vehicle, corresponding to the reference vehicle. The output acceleration is the output result of the acceleration model that has not been trained. The preset loss function is a pre-set function for calculating the error between the output acceleration and the artificial acceleration, such as the root mean square error. The target loss is the error between the output acceleration and the artificial acceleration, and is also the calculation result of the preset loss function. The preset loss is a pre-set loss value for judging whether the acceleration model has been trained. The model parameters are the adjustable parameters in the acceleration model, such as the initial learning rate, the gradient threshold, the learning rate reduction cycle and the learning rate reduction factor.

[0045] Specifically, the vehicle speed and artificial acceleration of each benchmark vehicle are collected and obtained, and the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle is obtained respectively. Then, the vehicle speed of each benchmark vehicle and the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle are used to train the untrained acceleration model. Specifically, the vehicle speed of each benchmark vehicle and the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle are processed separately by the untrained acceleration model to obtain the output acceleration corresponding to each set of data. Furthermore, the artificial acceleration and output acceleration of each benchmark vehicle are comprehensively calculated by the preset loss function, and the calculated result is the target loss of the untrained acceleration model. The target loss is compared with the preset loss to determine whether the acceleration model is trained. If the target loss is less than the preset loss, it means that the error is within an acceptable range, and the untrained acceleration model can be used as a pre-trained acceleration model. If the target loss is greater than or equal to the preset loss, it means that the current acceleration model still does not meet the usage requirements, and the model parameters in the current acceleration model need to be adjusted. Specifically, the model parameters can be adjusted in combination with the target loss, and the untrained acceleration model is updated based on the adjusted model parameters. The step of returning to execute the step of inputting the vehicle speed of each benchmark vehicle and the longitudinal motion information of the sample vehicles corresponding to each benchmark vehicle into the untrained acceleration model to obtain each output acceleration is performed until the target loss is less than the preset loss, and the trained acceleration model is obtained.

[0046] Exemplarily, the acceleration model can be defined, that is, the cycle of the LSTM network model is 15ms. The LSTM network model includes three layers, one input layer, one output layer, and the other is an intermediate layer, and the number of predicted output variables of the intermediate layer is 1 (acceleration). The input is the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle and the vehicle speed of each benchmark vehicle. For training in the intermediate layer, the rapid convergence of data can be achieved by debugging the four model parameters of the initial learning rate, the gradient threshold, the reduced learning rate cycle and the reduced learning rate factor, for example, by setting the number of training times to 20,000 times, the initialization learning rate to 0.002, the gradient to 2, the reduced learning rate cycle to 25, and the reduced learning rate factor to 0.5, so as to shorten the overall training cycle. Whether the training of the acceleration model is successful, the amount can be compared with the acceleration value (artificial acceleration) of the original output of the driver. When the target loss (root mean square error) is less than the preset loss (such as 0.2), the trained acceleration model is obtained.

[0047] It should be noted that the acceleration model can be deployed in two ways: offline or online. The offline method is to filter the collected data and then train it through an offline server to achieve curve fitting. The online method is to collect data in real time, filter valid data according to rules, and further achieve curve fitting based on deep learning.

[0048] S130. Taking a reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle as a main target vehicle, and determining a second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle.

[0049] The lateral overlap area is the area directly in front of the vehicle width corresponding to the current vehicle. The main target vehicle is the one with the smallest longitudinal distance from the current vehicle among the reference vehicles. The second acceleration is an acceleration value calculated and determined based on the kinematic equation.

[0050] Specifically, the longitudinal distances between each reference vehicle and the current vehicle in the lateral overlap region of the current vehicle are compared, and the reference vehicle with the smallest longitudinal distance is used as the main target vehicle. Then, the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle are substituted into the kinematic equation to obtain the second acceleration through calculation.

[0051] Based on the above example, the second acceleration may be determined according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle in the following manner:

[0052] Specifically, the second acceleration may be determined based on the following formula:

[0053]

[0054] Where AX2 is the second acceleration, v t is the longitudinal velocity of the main target vehicle, v e is the vehicle speed of the current vehicle, and s is the longitudinal distance between the main target vehicle and the current vehicle.

[0055] S140: Determine a target acceleration according to the first acceleration and the second acceleration, and control the current vehicle based on the target acceleration.

[0056] Among them, the target acceleration is the acceleration value determined by combining the kinematic equation and the deep learning model.

[0057] Specifically, the first acceleration and the second acceleration are integrated by a pre-defined acceleration integration rule, and the obtained acceleration value is the target acceleration. The acceleration integration rule may be a predetermined function, model, or conditional rule, etc. Further, after the target acceleration is obtained, the current vehicle is controlled according to the target acceleration to perform automatic driving cruising.

[0058] Based on the above example, before determining the target acceleration according to the first acceleration and the second acceleration, the first acceleration may be limited by a preset acceleration range to avoid the target acceleration obtained subsequently being too large or too small, thereby causing safety hazards:

[0059] If the first acceleration is greater than the preset upper acceleration limit, the preset upper acceleration limit is used as the limited first acceleration;

[0060] If the first acceleration is less than the preset acceleration lower limit, the preset acceleration lower limit is used as the limited first acceleration.

[0061] Among them, the preset acceleration upper limit is the upper limit value of the preset acceleration range, and correspondingly, the preset acceleration lower limit is the lower limit value of the preset acceleration range. For example, if the preset acceleration range is [-5, 3], then the preset acceleration upper limit is 3 and the preset acceleration lower limit is -5.

[0062] Specifically, if the first acceleration is greater than the preset acceleration upper limit, it means that the first acceleration is too large and exceeds the preset acceleration range, therefore, the first acceleration is limited, and the preset acceleration upper limit is used as the first acceleration after limitation. If the first acceleration is less than the preset acceleration lower limit, the first acceleration is too small and exceeds the preset acceleration range, therefore, the first acceleration is limited, and the preset acceleration lower limit is used as the first acceleration after limitation.

[0063] Exemplarily, the purpose of the above content is to constrain the first acceleration output by the acceleration model. The safety constraint can be limited according to the physical limit. For example, the preset acceleration range is agreed to be [-5,3]. When the first acceleration value is less than -5, -5 is output. When the first acceleration value is greater than 3, 3 is output.

[0064] Accordingly, the target acceleration is determined according to the first acceleration and the second acceleration in the following manner:

[0065] If the first acceleration is greater than the preset acceleration upper limit or the first acceleration is less than the preset acceleration lower limit, determining the target acceleration according to the limited first acceleration and the second acceleration;

[0066] If the first acceleration is less than or equal to the preset acceleration upper limit and the first acceleration is greater than or equal to the preset acceleration lower limit, the target acceleration is determined according to the first acceleration and the second acceleration.

[0067] Specifically, if the first acceleration is greater than the preset acceleration upper limit or the first acceleration is less than the preset acceleration lower limit, it means that the first acceleration has been restricted, therefore, it is necessary to combine the restricted first acceleration and the second acceleration to determine the target acceleration. If the first acceleration is less than or equal to the preset acceleration upper limit and the first acceleration is greater than or equal to the preset acceleration lower limit, it means that the first acceleration does not need to be restricted and is also within the preset acceleration range, therefore, it is only necessary to combine the first acceleration and the second acceleration to determine the target acceleration.

[0068] The present invention has the following technical effects: by acquiring the longitudinal motion information of each reference vehicle, the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle are input into a pre-trained acceleration model to obtain a first acceleration, so as to predict the acceleration value through a deep learning model, and then, the reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle is used as the main target vehicle, and according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle, the second acceleration is determined, so as to calculate the acceleration value through a kinematic formula, and further, according to the first acceleration and the second acceleration, the target acceleration is determined, so as to correct the second acceleration through the first acceleration, and the current vehicle is controlled based on the target acceleration, thereby achieving the effect of combining the driver's driving habits and effectively controlling the vehicle under safety constraints.

[0069] The above embodiments respectively describe the determination methods of the first acceleration and the second acceleration. In this embodiment, further description may be given for the determination method of the reference vehicle and the method of comprehensively determining the target acceleration using the first acceleration and the second acceleration, so as to improve the safety of vehicle control, such as Figure 2 shown. Figure 2 is a flow chart of another vehicle control method provided by an embodiment of the present invention. The explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here. Figure 2 As shown, the method may specifically include the following steps:

[0070] S210: Take the vehicle ahead of the current vehicle as a candidate vehicle, and obtain the lateral distance between each candidate vehicle and the current vehicle.

[0071] The front vehicle is the vehicle located to the left, right and front of the current vehicle. The candidate vehicle is the vehicle in front of the current vehicle. The lateral distance is the distance in the direction perpendicular to the traveling direction.

[0072] Specifically, the vehicle ahead of the current vehicle is used as an alternative vehicle, and the lateral distance between each alternative vehicle and the current vehicle can be determined by sensors pre-installed on the vehicle.

[0073] S220: For each candidate vehicle, if the lateral distance between the candidate vehicle and the current vehicle is less than or equal to the target width, the candidate vehicle is used as a reference vehicle, and longitudinal motion information of each reference vehicle is obtained.

[0074] The target width is a predetermined width used to determine whether the alternative vehicle will affect the travel of the current vehicle.

[0075] Specifically, for each alternative vehicle, it is determined whether the alternative vehicle can be used as a reference vehicle for subsequent operations, and whether the lateral distance between the alternative vehicle and the current vehicle is less than or equal to the target width. If so, it means that the alternative vehicle will affect the movement of the current vehicle and needs to be included in the scope of consideration. Therefore, the alternative vehicle is used as a reference vehicle and the longitudinal motion information of each reference vehicle is obtained; if not, it means that the alternative vehicle is far away from the current vehicle and does not need to be included in the scope of consideration.

[0076] Exemplarily, the target width is the width of 1.5 lanes, which means that the alternative vehicles within 3 lanes ahead of the current vehicle can be used as reference vehicles.

[0077] S230: Input the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle into a pre-trained acceleration model to obtain a first acceleration.

[0078] S240: taking a reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle as a main target vehicle, and determining a second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle.

[0079] S250: Determine an acceleration difference according to the absolute value of the first acceleration and the absolute value of the second acceleration.

[0080] Specifically, the acceleration difference is determined by the following formula:

[0081] AX′=|AX1|-|AX2|

[0082] Wherein, AX′ is the acceleration difference, AX1 is the first acceleration, and AX2 is the second acceleration.

[0083] S260: If the acceleration difference is greater than the preset acceleration threshold, determine the target acceleration according to the first acceleration, the second acceleration and the preset correction value.

[0084] The preset acceleration threshold is a value used to evaluate whether the difference between the first acceleration and the second acceleration is too large and needs to be arbitrated. The preset correction value is a pre-set value used to compensate the second acceleration value in order to obtain a more appropriate target acceleration. It is understood that the preset correction value is greater than zero, and the preset correction value is less than or equal to the preset acceleration threshold.

[0085] Specifically, if the acceleration difference is greater than the preset acceleration threshold, it means that the first acceleration output by the acceleration model is significantly different from the second acceleration calculated by the kinematic formula. At this time, it is necessary to determine whether to make a positive or negative correction to the second acceleration based on the positive or negative condition of the first acceleration. The specific correction value is the preset correction value.

[0086] Based on the above example, the target acceleration can be determined according to the first acceleration, the second acceleration and the preset correction value in the following manner:

[0087] If the first acceleration is greater than 0, the sum of the second acceleration and the preset correction value is used as the target acceleration;

[0088] If the first acceleration is less than or equal to 0, the difference between the second acceleration and the preset correction value is used as the target acceleration.

[0089] Specifically, if the first acceleration is greater than 0, it means that the second acceleration needs to be corrected in the positive direction, so the sum of the second acceleration and the preset correction value is used as the target acceleration. If the first acceleration is less than or equal to 0, it means that the second acceleration needs to be corrected in the negative direction, so the difference between the second acceleration and the preset correction value is used as the target acceleration.

[0090] S270: If the acceleration difference is less than or equal to the preset acceleration threshold, the second acceleration is used as the target acceleration.

[0091] Specifically, if the acceleration difference is greater than the preset acceleration threshold, it means that the first acceleration output by the acceleration model is slightly different from the second acceleration calculated by the kinematic formula. At this time, in order to avoid using the first acceleration to correct the second acceleration, the second acceleration is used as the target acceleration.

[0092] S280: Control the current vehicle based on the target acceleration.

[0093] Before controlling the current vehicle based on the target acceleration, the target acceleration may also be limited by a preset threshold (for example, an absolute value not exceeding 5, etc.) to ensure that the target acceleration is within the limit of the preset threshold.

[0094] Figure 3 is a schematic diagram of the structure of a vehicle control system provided by an embodiment of the present invention, such as Figure 3As shown, the vehicle control system consists of a perception module, a controller, and an actuator.

[0095] Among them, the perception module is used to identify the longitudinal motion information of the vehicle ahead of the current vehicle's motion trajectory, including the longitudinal distance, lateral distance, longitudinal acceleration, and longitudinal speed from the current vehicle. The perception module can be a smart camera, etc., or a speed sensor, ranging sensor, etc. The controller is used to obtain the target acceleration of the current vehicle based on the information output by the perception module through data screening, combined with the acceleration model trained by the driver's manual driving acceleration, and the kinematic formula. The actuator is used to receive the target acceleration issued by the controller and realize the acceleration and deceleration control of the current vehicle.

[0096] Figure 4 It is a schematic diagram of a processing flow of a controller of a vehicle control system provided by an embodiment of the present invention.

[0097] The data screening module is used to screen the candidate vehicles according to the longitudinal distance, lateral distance, longitudinal acceleration, and longitudinal speed of each candidate vehicle sent by the perception recognition and the current vehicle, combined with the target width and each lateral distance, and eliminate the information with the lateral distance < the target width. The data deep learning processing module is used to fit the acceleration information (first acceleration) AX1 requested by the driver through the deep learning model (acceleration model) of the longitudinal distance, longitudinal speed, longitudinal acceleration of the reference vehicle and the vehicle speed of the current vehicle based on the screened reference vehicle. The safety constraint module is used to constrain the first acceleration through a preset acceleration range. The acceleration control module is used to determine the main target vehicle and determine the second acceleration through the formula: Where AX2 is the second acceleration, v t is the longitudinal velocity of the main target vehicle, v e is the vehicle speed of the current vehicle, s is the longitudinal distance between the main target vehicle and the current vehicle. The acceleration arbitration module is used to determine the target acceleration AX according to the first acceleration AX1 and the second acceleration AX2. out .

[0098] The present invention has the following technical effects: by taking the vehicle in front of the current vehicle as an alternative vehicle and obtaining the lateral distance between each alternative vehicle and the current vehicle, for each alternative vehicle, if the lateral distance between the alternative vehicle and the current vehicle is less than or equal to the target width, the alternative vehicle is used as a reference vehicle to eliminate unnecessary information and improve the subsequent model effect; further, by determining the acceleration difference according to the absolute value of the first acceleration and the absolute value of the second acceleration, if the acceleration difference is greater than the preset acceleration threshold, the target acceleration is determined according to the first acceleration, the second acceleration and the preset correction value; if the acceleration difference is less than or equal to the preset acceleration threshold, the second acceleration is used as the target acceleration, and the second acceleration is corrected in combination with the first acceleration and the second acceleration, thereby ensuring driving safety while simulating human driving.

[0099] Figure 5 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 5 As shown, the electronic device 500 includes one or more processors 501 and a memory 502 .

[0100] The processor 501 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 500 to perform desired functions.

[0101] The memory 502 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 501 may run the program instructions to implement the vehicle control method of any embodiment of the present invention described above and / or other desired functions. Various contents such as initial external parameters, thresholds, etc. may also be stored in the computer-readable storage medium.

[0102] In one example, the electronic device 500 may further include: an input device 503 and an output device 504, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown). The input device 503 may include, for example, a keyboard, a mouse, etc. The output device 504 may output various information to the outside, including early warning information, braking force, etc. The output device 504 may include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, etc.

[0103] Of course, to simplify, Figure 5 Only some of the components related to the present invention in the electronic device 500 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application conditions, the electronic device 500 may also include any other appropriate components.

[0104] In addition to the above methods and devices, an embodiment of the present invention may also be a computer program product, which includes computer program instructions. When the computer program instructions are executed by a processor, the processor executes the steps of the vehicle control method provided by any embodiment of the present invention.

[0105] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the embodiments of the present invention, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0106] In addition, an embodiment of the present invention may also be a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the processor executes the steps of the vehicle control method provided by any embodiment of the present invention.

[0107] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0108] It should be noted that the terms used in the present invention are only for describing specific embodiments, rather than limiting the scope of the present application. As shown in the present specification, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the presence of other identical elements in the process, method or device including the elements.

[0109] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle control method, characterized in that: include: Acquire longitudinal motion information of each reference vehicle; wherein the longitudinal motion information includes the longitudinal distance between the reference vehicle and the current vehicle, the longitudinal speed and the longitudinal acceleration; Inputting the longitudinal motion information of each reference vehicle and the vehicle speed of the current vehicle into a pre-trained acceleration model to obtain a first acceleration; Taking the reference vehicle with the smallest longitudinal distance in the lateral overlapping area corresponding to the current vehicle as the main target vehicle, and determining a second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle; A target acceleration is determined according to the first acceleration and the second acceleration, and the current vehicle is controlled based on the target acceleration.

2. The method according to claim 1, characterized in that Before acquiring the longitudinal motion information of each reference vehicle, the method further includes: Taking the vehicle ahead of the current vehicle as a candidate vehicle, and obtaining the lateral distance between each candidate vehicle and the current vehicle; For each candidate vehicle, if the lateral distance between the candidate vehicle and the current vehicle is less than or equal to the target width, the candidate vehicle is used as a reference vehicle.

3. The method according to claim 1, characterized in that Before determining the target acceleration according to the first acceleration and the second acceleration, the method further includes: If the first acceleration is greater than a preset upper acceleration limit, the preset upper acceleration limit is used as the limited first acceleration; If the first acceleration is less than a preset acceleration lower limit, the preset acceleration lower limit is used as the limited first acceleration; Accordingly, determining the target acceleration according to the first acceleration and the second acceleration includes: If the first acceleration is greater than the preset acceleration upper limit or the first acceleration is less than the preset acceleration lower limit, determining a target acceleration according to the limited first acceleration and the second acceleration; If the first acceleration is less than or equal to the preset acceleration upper limit and the first acceleration is greater than or equal to the preset acceleration lower limit, a target acceleration is determined according to the first acceleration and the second acceleration.

4. The method according to claim 1, characterized in that The determining of the second acceleration according to the longitudinal motion information of the main target vehicle and the vehicle speed of the current vehicle comprises: The second acceleration is determined based on the following formula: Where AX2 is the second acceleration, v t is the longitudinal velocity of the main target vehicle, v e is the vehicle speed of the current vehicle, and s is the longitudinal distance between the main target vehicle and the current vehicle.

5. The method according to claim 1, characterized in that The determining the target acceleration according to the first acceleration and the second acceleration includes: determining an acceleration difference according to an absolute value of the first acceleration and an absolute value of the second acceleration; If the acceleration difference is greater than a preset acceleration threshold, determining a target acceleration according to the first acceleration, the second acceleration and a preset correction value; If the acceleration difference is less than or equal to the preset acceleration threshold, the second acceleration is used as the target acceleration.

6. The method according to claim 5, characterized in that The determining the target acceleration according to the first acceleration, the second acceleration and the preset correction value includes: If the first acceleration is greater than 0, the sum of the second acceleration and the preset correction value is used as the target acceleration; If the first acceleration is less than or equal to 0, the difference between the second acceleration and the preset correction value is used as the target acceleration.

7. The method according to claim 1, characterized in that The acceleration model is trained based on the following method: Obtaining the vehicle speed and artificial acceleration of each reference vehicle, and obtaining longitudinal motion information of the sample vehicles corresponding to each reference vehicle; The vehicle speed of each reference vehicle and the longitudinal motion information of the sample vehicle corresponding to each reference vehicle are respectively input into the untrained acceleration model to obtain each output acceleration; Based on a preset loss function, the artificial acceleration and output acceleration of each benchmark vehicle are calculated to determine the target loss of the untrained acceleration model; If the target loss is less than the preset loss, the untrained acceleration model is used as the pre-trained acceleration model; If the target loss is greater than or equal to the preset loss, the model parameters of the untrained acceleration model are adjusted, the untrained acceleration model is updated, and the step of inputting the vehicle speed of each benchmark vehicle and the longitudinal motion information of the sample vehicle corresponding to each benchmark vehicle into the untrained acceleration model to obtain each output acceleration is returned to execute until the target loss is less than the preset loss.

8. The method according to claim 1, characterized in that The obtaining of longitudinal motion information of each reference vehicle comprises: In response to receiving the smart cruise signal, starting the smart camera of the current vehicle; Based on the smart camera, obtaining the longitudinal distance between each reference vehicle and the current vehicle, the longitudinal speed of each reference vehicle, and the longitudinal acceleration of each reference vehicle; For each of the reference vehicles, the longitudinal distance between the reference vehicle and the current vehicle, the longitudinal speed of the reference vehicle, and the longitudinal acceleration of the reference vehicle are used as the longitudinal motion information of the reference vehicle.

9. An electronic device, characterized in that: The electronic device comprises: Processor and memory; The processor is used to execute the steps of the vehicle control method as described in any one of claims 1 to 8 by calling the program or instruction stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program or an instruction that causes a computer to execute the steps of the vehicle control method according to any one of claims 1 to 8.

Images