Method, device, electronic device and storage medium for determining speed of autonomous driving vehicle

Abstract

The present application relates to a method, device, electronic device and storage medium for determining the speed of an autonomous driving vehicle. The method includes: obtaining a first speed and a second speed of the autonomous driving vehicle, wherein the first speed is the vehicle speed obtained when the autonomous driving vehicle is traveling straight and the quality of the RTK differential signal is good, and the second speed is the calibrated speed of the first speed; training a preset linear model according to the first speed and the second speed to obtain a speed calibration model; obtaining a third speed when the quality of the RTK differential signal of the autonomous driving vehicle is not good, and determining the calibrated speed of the third speed based on the speed calibration model and the third speed. The solution provided by the present application obtains a speed calibration model when the quality of the RTK differential signal is good, selects a preset speed interval according to the speed when the RTK differential signal is not good, and selects a speed calibration model corresponding to the preset speed interval to calibrate the speed in real time, thereby improving the accuracy of the speed calibration.

Description

Technical Field

[0001] The present application relates to the field of navigation technology, and in particular to a method, device, electronic device, and storage medium for determining the speed of an autonomous driving vehicle. Background Art

[0002] In the related art, the speed of the vehicle is an important control variable in autonomous driving. The control of the vehicle's speed closely affects the safety of the passengers, and the accuracy of measuring the vehicle's speed affects the judgment of the autonomous driving system. The speed of an autonomous driving vehicle is generally measured by the internal sensors of the autonomous driving vehicle, but there is often a deviation in the vehicle's driving speed obtained by the internal wear sensor. The vehicle speed usually measured by the internal sensor includes the vehicle chassis speed and the wheel speed of the autonomous driving vehicle. Electromagnetic sensors are often used to measure the chassis speed of the vehicle. The electromagnetic sensor collects electromagnetic induction signals through the rotation of the wheel. When the road surface is uneven or the vehicle vibrates, the measured speed has errors. The bottom speed of the vehicle can be measured by using an optical sensor, and the time it takes for the light beam to reflect from the ground to determine the chassis speed of the vehicle, but this method has high requirements on the placement and stability of the optical device, and there are errors in the measuring device. Therefore, the calibrated speed of the autonomous driving vehicle can be obtained from the positioning system of the satellite navigation positioning and the inertial measurement unit to calibrate the vehicle speed.

[0003] However, in some scenarios, such as under bridges, in culverts, tunnels, and between densely populated buildings where the positioning signal is poor, the calibrated speed of the autonomous driving vehicle obtained by the positioning system based on RTK differential signals is often not accurate enough, or even unable to provide the calibrated speed of the autonomous driving vehicle.

[0004] Therefore, it is necessary to provide a method for accurately calibrating the vehicle speed when the RTK differential signal is poor. Summary of the invention

[0005] In order to solve or partially solve the problems existing in the related technology, the present application provides a method, device, electronic device, and storage medium for determining the speed of an autonomous driving vehicle, which can solve the problem that the speed of the autonomous driving vehicle cannot be accurately calibrated when the RTK differential signal is not good.

[0006] A first aspect of the present application provides a method for determining the speed of an autonomous driving vehicle, comprising:

[0007] Obtaining a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is a vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle satisfies a first preset condition and the heading angle of the autonomous driving vehicle satisfies a second preset condition, the first preset condition refers to a judgment condition when the quality of the RTK differential signal of the autonomous driving vehicle is available, and the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state;

[0008] Obtaining a second vehicle speed of the autonomous driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed;

[0009] A preset linear model is trained according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, wherein the vehicle speed calibration model is used to receive the first vehicle speed and output a second vehicle speed corresponding to the first vehicle speed;

[0010] A third vehicle speed is obtained, where the third vehicle speed is a vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet a first preset condition, and a calibrated speed of the third vehicle speed is determined based on a vehicle speed calibration model and the third vehicle speed.

[0011] Preferably, the first vehicle speed is the vehicle chassis speed measured by the vehicle speed sensor of the autonomous driving vehicle, the second vehicle speed is the vehicle travel speed calculated based on the RTK differential signal corresponding to the first vehicle speed, and the moment of the second vehicle speed is the same as the moment of the first vehicle speed corresponding to the second vehicle speed.

[0012] Preferably, a preset linear model is trained according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, including:

[0013] Determine a model error parameter of a preset linear model according to the speed error coefficient;

[0014] Setting a preset speed interval, and initializing a model error parameter of a preset linear model within the preset speed interval;

[0015] According to a first vehicle speed within a preset speed interval and a second vehicle speed corresponding to the first vehicle speed, a training sample set of a preset linear model within the preset speed interval is obtained;

[0016] The sample training set within the preset speed range is input into the preset linear model, and the model error parameters are optimized to obtain the vehicle speed calibration model corresponding to each preset speed range.

[0017] Preferably, the sample training set within the preset speed range is input into the preset linear model, and the model error parameter is optimized to obtain the vehicle speed calibration model corresponding to each preset speed range, including:

[0018] Inputting a first vehicle speed into a preset linear model, and obtaining a fourth vehicle speed output by the preset linear model according to the first vehicle speed;

[0019] Calculating a value of a loss function of a preset linear model according to a fourth vehicle speed outputted from the first vehicle speed and a second vehicle speed corresponding to the first vehicle speed;

[0020] The model error parameters of the preset linear model are optimized according to the value of the loss function to obtain the vehicle speed calibration model corresponding to each preset speed interval.

[0021] Preferably, the model error coefficient includes a first calibration coefficient and a second calibration coefficient, and setting the model error parameter of the preset linear model according to the speed error coefficient includes:

[0022] Obtaining a speed error coefficient, and setting a first calibration coefficient and a second calibration coefficient of a preset linear model according to the error coefficient of the autonomous driving vehicle, wherein the speed error coefficient is an error coefficient between the first vehicle speed and the second vehicle speed;

[0023] The product of the first calibration coefficient and the first vehicle speed is obtained, and a preset linear model is determined according to the product of the first calibration coefficient and the first vehicle speed and the second calibration coefficient.

[0024] Preferably, in a preset speed range, initializing the error parameters of the preset linear model includes:

[0025] In each preset speed interval, the first calibration coefficient of the preset linear model is set to 1, and the second calibration coefficient of the preset linear model is set to 0.

[0026] Preferably, determining a calibrated speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed includes:

[0027] determining a preset speed interval in which the third vehicle speed is located;

[0028] The third vehicle speed is input into a vehicle speed calibration model corresponding to a preset speed interval to obtain a calibrated speed corresponding to the third vehicle speed.

[0029] A second aspect of the present application provides a speed determination device for an autonomous driving vehicle, comprising:

[0030] a first acquisition unit, configured to acquire a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is a vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle satisfies a first preset condition and the heading angle of the autonomous driving vehicle satisfies a second preset condition, wherein the first preset condition refers to a judgment condition when the quality of the RTK differential signal of the autonomous driving vehicle is available, and the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state;

[0031] A second acquisition unit is used to acquire a second vehicle speed of the automatic driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed;

[0032] A training unit is used to train a preset linear model according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, wherein the vehicle speed calibration model outputs a second vehicle speed corresponding to the first vehicle speed according to the first vehicle speed;

[0033] The calibration unit is used to obtain a third vehicle speed, where the third vehicle speed is the vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet the first preset condition, and determine the calibration speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed.

[0034] Preferably, the device further comprises a parameter calibration unit, which is used to obtain a model error parameter of the vehicle speed calibration model and adjust a parameter value of the inertial measurement unit according to the model error parameter of the vehicle speed calibration model.

[0035] A third aspect of the present application provides an electronic device, including:

[0036] Processor; and

[0037] The memory stores executable codes thereon, and when the executable codes are executed by the processor, the processor is caused to execute the method as described above.

[0038] A fourth aspect of the present application provides a computer-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor is caused to execute the method as described above.

[0039] The technical solution provided by the present application may include the following beneficial effects: on the one hand, since the original sample set used to train the speed calibration model is obtained when the RTK differential signal of the autonomous driving vehicle meets the preset conditions, when the RTK differential signal is not good, the vehicle chassis speed can be accurately calibrated based on the trained speed calibration model; on the other hand, since each preset speed interval corresponds to its original sample set, the problem of inaccurate calibration of the target speed calibration model caused by the large span of sample data can be eliminated, that is, by subdividing the original sample set, a more accurate training sample can be provided for the speed calibration model, so that the target calibration model obtained thereby can calibrate the vehicle driving speed more accurately.

[0040] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and other objects, features and advantages of the present application will become more apparent by describing in more detail exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0042] Figure 1 is a flow chart of a method for determining the speed of an autonomous driving vehicle shown in an embodiment of the present application;

[0043] Figure 2is another flow chart of a method for determining the speed of an autonomous driving vehicle shown in an embodiment of the present application;

[0044] Figure 3 is another flow chart of a method for determining the speed of an autonomous driving vehicle shown in an embodiment of the present application;

[0045] Figure 4 is another flow chart of a method for determining the speed of an autonomous driving vehicle shown in an embodiment of the present application;

[0046] Figure 5 is a schematic diagram of the structure of a speed determination device for an autonomous driving vehicle shown in an embodiment of the present application;

[0047] Figure 6 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0049] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0050] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0051] In related technologies, there are often deviations when directly obtaining the vehicle's driving speed through the internal sensors of the autonomous vehicle, such as the chassis speed and wheel speed of the autonomous vehicle obtained through the internal sensors of the autonomous vehicle. Therefore, the calibrated speed of the autonomous vehicle can be obtained from the positioning system and inertial measurement unit of the satellite navigation positioning. However, in some scenarios, such as under bridges, culverts, tunnels, and between dense buildings where the positioning signal is poor, the calibrated speed of the autonomous vehicle obtained by the positioning system based on RTK differential signals is often not accurate enough, or even unable to provide the calibrated speed of the autonomous vehicle.

[0052] In order to solve or partially solve the problems existing in the related technology, the present application provides a method, device, electronic device, and storage medium for determining the speed of an autonomous driving vehicle, which can solve the problem that the speed of the autonomous driving vehicle cannot be accurately calibrated when the RTK differential signal is not good.

[0053] See also Figure 1 , a method for determining the speed of an autonomous driving vehicle, comprising:

[0054] Step S101, obtaining a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is a vehicle speed obtained when the RTK differential signal quality of the autonomous driving vehicle satisfies a first preset condition and the heading angle of the autonomous driving vehicle satisfies a second preset condition, the first preset condition refers to a judgment condition when the RTK differential signal quality of the autonomous driving vehicle is available, and the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state.

[0055] In step S101, the first vehicle speed is the vehicle chassis speed measured by the vehicle speed sensor of the autonomous driving vehicle. The satellite navigation system for obtaining the RTK differential signal can be any one of the GPS (Global Positioning System), the Beidou satellite positioning system and the RTK (Real Time Kinematic) positioning system. In the scenario where the RTK differential signal is good, the calibration speed obtained by the autonomous driving vehicle based on the RTK differential signal is more accurate. However, in the scenario where the quality of the RTK differential signal is poor, the calibration speed obtained based on the RTK differential signal is not very accurate. Therefore, it is necessary to obtain a sample sampling set for each preset speed interval in the scenario where the satellite quality signal is good. Since the poor quality RTK differential signal will cause the data of the original sample set to be inaccurate, before collecting the original sample set, the quality of the RTK differential signal must first be judged, and the RTK differential signal that meets the preset conditions must be selected as the data source for the calibration speed. Therefore, when the first vehicle speed and the second vehicle speed are used to train the preset linear model to determine the speed calibration model, the first vehicle speed and the second vehicle speed when the quality of the RTK differential signal is good need to be used.

[0056] Specifically, the step of judging the quality of the RTK differential signal includes: receiving the observation value of the RTK differential signal, solving the RTK differential signal according to the observation value, wherein the solution of the observation value includes a single point fixed solution, a pseudo-range solution, a fixed solution, and a floating point solution. If the solution of the obtained observation value is a fixed solution, it is judged that the quality of the RTK differential signal is good and meets the preset conditions.

[0057] In one embodiment, the driving state of the autonomous driving vehicle includes a straight-line driving state, a curve driving state, etc. of the autonomous driving vehicle.

[0058] In one embodiment, step S101 determines the straight-ahead state of the heading angle of the autonomous vehicle based on the RTK differential signal, specifically including: obtaining the heading angle of the autonomous vehicle based on the RTK differential signal of the autonomous vehicle; if the change angle of the heading angle is less than a preset change angle, then determining that the driving state of the autonomous vehicle is a straight-ahead state. The heading angle of the autonomous vehicle can be calculated based on the RTK differential signal, or it can be calculated based on data provided by the inertial measurement unit.

[0059] Step S102, obtaining a second vehicle speed of the automatic driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed.

[0060] In step S102, the time of the second vehicle speed is the same as the time of the first vehicle speed corresponding to the second vehicle speed. Since the second vehicle speed is obtained when the quality of the RTK differential signal is good, this second vehicle speed is a more accurate calibration speed of the first vehicle speed.

[0061] Step S103, training a preset linear model according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, where the vehicle speed calibration model is used to receive the first vehicle speed and output a second vehicle speed corresponding to the first vehicle speed.

[0062] In one embodiment, if Figure 2 As shown, a preset linear model is trained according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, including:

[0063] S201. Determine a model error parameter of a preset linear model according to a speed error coefficient.

[0064] Step S201 eliminates the error between the first vehicle speed and the second vehicle speed by setting a linear model. The speed error coefficient is a calibrated error coefficient between the first vehicle speed and the second vehicle speed. In this application, a preset linear model is used between the first vehicle speed and the second vehicle speed. Therefore, the first calibration coefficient and the second calibration coefficient of the preset linear model can be assumed first, and the first calibration coefficient and the second calibration coefficient are determined by the first vehicle speed, the second vehicle speed and the speed error coefficient. In step S201, the model error parameters of the preset linear model are determined according to the speed error coefficient, including: obtaining the error coefficient of the autonomous driving vehicle, setting the first calibration coefficient and the second calibration coefficient of the preset linear model according to the error coefficient of the autonomous driving vehicle; obtaining the product of the first calibration coefficient and the first vehicle speed, and determining the preset linear model according to the product of the first calibration coefficient and the first vehicle speed and the second calibration coefficient.

[0065] In one embodiment, the preset linear model includes:

[0066] v = k 1 *V 0 +b 1 ;

[0067] Among them, V 0 is the first vehicle speed, and v is the calibrated speed corresponding to the first vehicle speed, i.e., the second vehicle speed. 1 is the first calibration coefficient of the speed calibration model, b 1 is the second calibration coefficient of the speed calibration model, and the model error parameter includes the first calibration coefficient and the second calibration coefficient.

[0068] S202: Setting a preset speed interval, and initializing a model error parameter of a preset linear model within the preset speed interval.

[0069] The preset speed interval is obtained according to the vehicle chassis speed. By dividing the chassis speed into multiple preset speed intervals and then obtaining the speed calibration method according to the preset speed intervals, the problem of the speed calibration method being inaccurate due to the large span of the original sample set can be reduced. Specifically, the chassis speed of the autonomous driving vehicle can be divided into a low-speed preset speed interval, a medium-speed preset speed interval, and a high-speed preset speed interval according to the driver's driving habits and the standards of the autonomous driving industry. For example, the low-speed preset speed interval is 2m / s to 10m / s, the medium-speed preset speed interval is 10m / s to 25m / s, and the high-speed preset speed interval is greater than 25m / s.

[0070] In one embodiment, a preset speed interval can also be obtained according to the driving state of the autonomous driving vehicle. For example, according to the driving state of the autonomous driving vehicle, the autonomous driving vehicle is divided into a starting preset speed interval, a normal driving preset speed interval, an acceleration preset speed interval, and a high-speed driving interval. Among them, each driving state interval can be divided into multiple gear preset speed intervals according to the vehicle driving speed. For example, for the starting preset speed interval, the preset speed interval can be divided into a first gear speed, a second gear speed, and a third gear speed, wherein the speed interval corresponding to the first gear speed is 0km / h~15km / h, the second gear preset speed interval is: 10km / h~25km / h, and the third gear preset speed interval is: 20km / h~45km / h. For the preset speed intervals of other states, the method for obtaining the preset speed interval in the starting preset speed interval can be referred to, and the preset speed interval can be obtained according to the driving state of the autonomous driving vehicle.

[0071] Optionally, in one embodiment, the environmental conditions and preset speed intervals of the autonomous driving vehicle can be combined to obtain an original sample set of preset speed intervals under relevant environmental conditions, so that the speed calibration model trained with the original sample set is more targeted. For example, the environmental conditions of the autonomous driving vehicle are determined by obtaining driving scene information of the autonomous driving vehicle, road condition information of the autonomous driving vehicle, and driving environment information, and the maximum driving speed and the minimum driving speed of the autonomous driving vehicle are determined according to the environmental conditions of the autonomous driving vehicle, and the preset speed interval of the autonomous driving vehicle is obtained according to the maximum driving speed and the minimum driving speed.

[0072] For example, during the calibration process, the driving speed of the autonomous driving vehicle can be divided into multiple preset speed intervals. In this case, the preset speed interval corresponding to the chassis speed is selected during the driving process, and the target speed calibration model corresponding to the preset speed interval is obtained, and the chassis speed is calibrated based on the target speed calibration model. By selecting the preset speed interval for the chassis speed of the autonomous driving vehicle in real time, the target calibration speed model of the preset speed interval can be obtained, thereby achieving real-time calibration of the vehicle speed according to the preset speed interval.

[0073] For example, for the first vehicle speed V 0 , select V 0 The corresponding preset speed interval is selected, and the first calibration coefficient and the second calibration coefficient corresponding to the preset speed interval are selected, and the first calibration coefficient and the second calibration coefficient are substituted into the speed calibration model, and V is input according to the speed calibration model. 0 , we get V 0 The rated speed.

[0074] In step S202, by training the preset linear model in each preset speed interval to obtain the speed calibration model, the problem of inaccurate calibration of the target speed calibration model caused by the large span of the original sample data can be eliminated. Calibrating the vehicle speed in the preset speed interval can provide a more accurate speed calibration model.

[0075] In each preset speed interval, a first vehicle speed that satisfies a preset driving state of the automatic driving vehicle and a second vehicle speed corresponding to the first vehicle speed are obtained to generate an original sample set corresponding to each preset speed interval.

[0076] In step S202, the error parameters of the preset linear model are initialized within the preset speed interval, including: setting the first calibration coefficient of the preset linear model to 1 and setting the second calibration coefficient of the preset linear model to 0 within each preset speed interval.

[0077] In step S202, in the embodiment of the present invention, only the vehicle speed calibration method when the autonomous driving vehicle is driving in a straight line is discussed.

[0078] S203: Acquire a training sample set of a preset linear model within the preset speed interval according to a first vehicle speed within the preset speed interval and a second vehicle speed corresponding to the first vehicle speed.

[0079] In one embodiment, generating the original sample set corresponding to each preset speed interval includes: obtaining a first vehicle speed and a second vehicle speed corresponding to the first vehicle speed within each preset speed interval, and marking the first vehicle speed and the second vehicle speed corresponding to the first vehicle speed with corresponding labels.

[0080] For example, for an original sample<V1,V2,θ> ,in<V1,V2> It means that it is necessary to determine whether it is a pair of calibrated speeds, V1 is the first vehicle speed, V2 is the calibrated speed corresponding to the first vehicle speed, θ is the corresponding label obtained according to V1 and V2, and θ is usually expressed by similarity.

[0081] Optionally, the original sample can also be directly obtained according to the first vehicle speed and the second vehicle speed.<V1,V2> ,<V1,V2> Indicates that it is necessary to determine whether it is a pair of calibrated speeds, V1 is the first vehicle speed, and V2 is the calibrated speed corresponding to the first vehicle speed.

[0082] S204: Input the sample training set within the preset speed range into the preset linear model to optimize the model error parameters to obtain the vehicle speed calibration model corresponding to each preset speed range.

[0083] Specifically, a sample training set within a preset speed range is input into a preset linear model, and model error parameters are optimized to obtain a vehicle speed calibration model corresponding to each preset speed range, including: inputting a first vehicle speed into the preset linear model to obtain a fourth vehicle speed output by the preset linear model based on the first vehicle speed; calculating a value of a loss function of the preset linear model based on the fourth vehicle speed output from the first vehicle speed and a second vehicle speed corresponding to the first vehicle speed; and optimizing model error parameters of the preset linear model based on the value of the loss function to obtain a vehicle speed calibration model corresponding to each preset speed range.

[0084] In one embodiment, step S204 inputs the sample training set within the preset speed range into the preset linear model, and optimizes the model error parameters to obtain the vehicle speed calibration model corresponding to each preset speed range. Training the preset linear model includes: setting a loss function, inputting the original sample set into the speed calibration model to obtain the loss function value, iteratively training the parameters of the speed calibration model through the loss function value, and stopping the training after the obtained loss function value converges to the preset threshold. The speed calibration model obtained at this time is the target speed calibration model. Specifically, Figure 3 As shown in step S301 to step S303.

[0085] In one embodiment, a preset linear model is used for the speed calibration model, and the preset linear model includes:

[0086] v = k 1 *V 0 +b 1 ;

[0087] Among them, V 0 is the first vehicle speed, v is the second vehicle speed corresponding to the first vehicle speed, k 1 is the first calibration coefficient of the speed calibration model, b 1 is the second calibration coefficient of the speed calibration model, and the model error parameter includes the first calibration coefficient and the second calibration coefficient.

[0088] Step S301: Initialize a preset linear model in each preset speed interval.

[0089] In one embodiment, in each preset speed interval, the speed calibration model is initialized, including: in each preset speed interval, setting a first calibration coefficient of the speed calibration model to 1, and setting a second calibration coefficient of the speed calibration model to 0.

[0090] Step S302, input each first vehicle speed in the original sample set into the speed calibration model to obtain a loss function value of a second vehicle speed corresponding to each first vehicle speed;

[0091] Specifically, in step S302, each original sample in the original sample set is obtained, and the true similarity probability between the first vehicle speed and the second vehicle speed corresponding to the first vehicle speed in each original sample is obtained. Each original sample in the original sample set is input into the speed calibration model to obtain the similarity probability of each original sample; the value of the loss function of the original sample is calculated according to the similarity probability corresponding to each original sample and the true similarity.

[0092] For example, in each interval, the original sample<V1,V2> Input the speed calibration model to obtain the original sample<V1,V2> The true similarity label θ in . You can also directly obtain the original sample<V1,V2,θ> During the training process, each original sample in the original sample set<V1,V2> The target model is input to obtain the probability (i.e., similarity probability) of each original sample being predicted as label θ, and then the value of the loss function is calculated based on the probability of each original sample being predicted as label θ and the true similarity θ.

[0093] Step S303: optimizing the model error parameters of the speed calibration model according to the loss function value, and obtaining the speed calibration model corresponding to each preset speed interval according to the optimized model error parameters and the calibration model.

[0094] In step S303, the back propagation gradient of the speed calibration model is obtained based on the value of the loss function obtained in step S302, and the model error parameters of the speed calibration model are updated according to the gradient, so as to finally obtain a trained speed calibration model, that is, a target speed calibration model.

[0095] In one embodiment, a maximum number of iterations is set, the original sample set is input into the calibration model, and a loss function corresponding to the chassis speed is obtained according to the chassis speed, and a loss function value is obtained. The model error parameter is updated by back-propagating the gradient according to the loss function value to obtain a trained speed calibration model. The parameters of the trained speed calibration model are obtained, and a target speed calibration model is determined according to the parameters.

[0096] Specifically, step S301 to step S303 use the speed calibration model and the original sample set to obtain the target calibration model including:

[0097] a) Select a preset speed range, give the initialization model and the original sample, and then calculate the output value of the model. For example, the first input parameter is the original sample<V1,V2,θ> , V1 is the first vehicle speed, V2 is the second vehicle speed, θ is the similarity label between V1 and V2, and the output value is the similarity probability of obtaining V2 based on V1 and the speed calibration model;

[0098] b) Calculate the value of the loss function based on the output value of the model and the true similarity label;

[0099] c) Back propagate the gradient according to the value of the loss function and update the model error parameters according to the gradient;

[0100] d) Given the number of iterations and setting the similarity probability, the output value is the model error parameter of the preset similarity probability, which is the model error parameter of the target model.

[0101] For example, in step (b), the loss function value is obtained, and in step (c), the loss function value is updated to the initial value, and then the process returns to step b), and according to a preset number of iterations, the data in the initial data set is gradually input, and the model is gradually trained until the preset number of iterations is reached to complete the parameter update.

[0102] In one embodiment, step S204, inputting the sample training set within the preset speed range into the preset linear model, optimizing the model error parameters to obtain the vehicle speed calibration model corresponding to each preset speed range, also includes: calculating the first calibration coefficient and the second calibration coefficient of the speed calibration model within each preset speed range based on the original sample set and the least squares method.

[0103] For example, original samples are collected within a preset time period, and the original samples are classified according to preset speed intervals to obtain original sample sets corresponding to vehicle speeds in each interval, wherein the original sample sets include a first vehicle speed and a second vehicle speed corresponding to the first vehicle speed. Within the preset speed interval, the original sample set data set and the speed calibration model are fitted according to the least squares method to obtain the first calibration coefficient and the second calibration coefficient value of the target calibration model.

[0104] In one embodiment, step S102 obtains a speed calibration model, trains the speed calibration model according to the original sample set corresponding to each preset speed interval, and obtains a target speed calibration model corresponding to each preset speed interval, and also includes: obtaining a first calibration coefficient and a second calibration coefficient of the speed calibration model according to the optimization algorithm and the original quantity set.

[0105] Specifically, the original sample set is collected within a preset time period, the original sample set is classified according to the preset speed interval, the original sample set corresponding to each preset speed interval is obtained, and the sample data is sorted in the original sample set according to the size of the calibration speed. The objective function of the original sample set and the speed calibration model is set. In the optimization algorithm, it is necessary to find the minimum value of the objective function. In each iteration, the objective function value is required to decrease, that is, starting from the initial point, first assume a reliable maximum displacement s, and then in the trust area with the current point as the center and s as the radius, by finding the optimal point of an approximate function (quadratic) of the objective function, the real displacement is solved. After the displacement is obtained, the objective function value is calculated again. If it makes the decrease of the objective function value meet certain conditions, then it means that the displacement is reliable, and then continue to iterate and calculate according to this rule; if it cannot make the decrease of the objective function value meet certain conditions, the range of the trust region should be reduced and then solved again.

[0106] For example, the objective function of the speed calibration model is obtained, the original sample set is updated, and the objective function is iterated step by step until the minimum value of the objective function is obtained within the trust region. At this time, the parameters of the target model can be calculated based on the minimum value of the objective function and the speed calibration model.

[0107] Step S104, obtaining a third vehicle speed, where the third vehicle speed is the vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet the first preset condition, and determining a calibrated speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed.

[0108] like Figure 4 As shown, based on the vehicle speed calibration model and the third vehicle speed, determining the calibration speed of the third vehicle speed includes:

[0109] Step S401: Determine a preset speed interval in which the third vehicle speed is located.

[0110] During the calibration process, the driving speed of the autonomous driving vehicle can be divided into multiple preset speed intervals. In this case, the preset speed interval corresponding to the third vehicle speed is selected during the driving process, and the target speed calibration model corresponding to the preset speed interval is obtained, and the third vehicle speed is calibrated based on the target speed calibration model. By selecting the preset speed interval for the third vehicle speed of the autonomous driving vehicle in real time, the target calibration speed model of the preset speed interval can be obtained, thereby achieving real-time calibration of the vehicle speed according to the preset speed interval.

[0111] Step S402: inputting the third vehicle speed into a speed calibration model corresponding to a preset speed interval to obtain a calibrated speed corresponding to the third vehicle speed.

[0112] For example, for the third vehicle speed V 0 , select V 0The corresponding preset speed interval is selected, and the first calibration coefficient and the second calibration coefficient corresponding to the preset speed interval are selected, and the first calibration coefficient and the second calibration coefficient are substituted into the target calibration model, and V is input according to the target calibration model. 0 , we get V 0 The rated speed.

[0113] Corresponding to the aforementioned application function implementation method embodiment, the present application also provides a speed determination device, electronic device and corresponding embodiments of an autonomous driving vehicle.

[0114] Figure 5 It is a schematic diagram of the structure of a speed determination device for an autonomous driving vehicle shown in an embodiment of the present application.

[0115] See also Figure 5 , the speed determination device of the automatic driving vehicle comprises:

[0116] The first acquisition unit 501 is used to acquire a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is the vehicle speed obtained when the RTK differential signal quality of the autonomous driving vehicle meets a first preset condition and the heading angle of the autonomous driving vehicle meets a second preset condition, the first preset condition refers to the judgment condition when the RTK differential signal quality of the autonomous driving vehicle is available, and the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state.

[0117] A second acquisition unit 502 is used to acquire a second vehicle speed of the automatic driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed;

[0118] A training unit 503 is used to train a preset linear model according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, wherein the vehicle speed calibration model outputs a second vehicle speed corresponding to the first vehicle speed according to the first vehicle speed;

[0119] The calibration unit 504 is used to obtain a third vehicle speed, where the third vehicle speed is the vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet the first preset condition, and determine the calibrated speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed.

[0120] In one embodiment, the device further includes an RTK differential signal acquisition unit, which is used to acquire an RTK differential signal that meets preset conditions and judge the driving state of the autonomous driving vehicle according to the RTK differential signal. When the RTK differential signal acquisition unit judges that the autonomous driving vehicle meets the preset autonomous driving vehicle driving state, the first acquisition unit is further used to acquire the chassis speed that meets the preset autonomous driving vehicle driving state and the calibration speed corresponding to the chassis speed that meets the preset autonomous driving vehicle driving state within each preset speed interval, and generate an original sample set corresponding to each preset speed interval.

[0121] In one embodiment, the device further includes a parameter calibration unit, which is used to obtain a model error parameter of the vehicle speed calibration model and adjust a parameter value of the inertial measurement unit according to the model error parameter of the vehicle speed calibration model.

[0122] The technical solution provided by the present application may include the following beneficial effects: on the one hand, since the original sample set used to train the speed calibration model is obtained when the RTK differential signal of the autonomous driving vehicle meets the preset conditions, when the RTK differential signal is not good, the vehicle chassis speed can be accurately calibrated based on the trained speed calibration model; on the other hand, since each preset speed interval corresponds to its original sample set, the problem of inaccurate calibration of the target speed calibration model caused by the large span of sample data can be eliminated, that is, by subdividing the original sample set, a more accurate training sample can be provided for the speed calibration model, so that the target calibration model obtained thereby can calibrate the vehicle driving speed more accurately.

[0123] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0124] Figure 6 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application.

[0125] See also Figure 6 , the electronic device 600 includes a memory 610 and a processor 620 .

[0126] The processor 620 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0127] The memory 610 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 620 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory 610 may include any combination of computer-readable storage media, including various types of semiconductor storage chips (such as DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 610 may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0128] The memory 610 stores executable codes, and when the executable codes are processed by the processor 620 , the processor 620 can execute part or all of the above-mentioned methods.

[0129] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0130] Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-transitory machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by a processor of an electronic device (or server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0131] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for determining the speed of an autonomous driving vehicle, characterized in that: include: Obtaining a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is a vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle satisfies a first preset condition and the heading angle of the autonomous driving vehicle satisfies a second preset condition, the first preset condition refers to a judgment condition when the quality of the RTK differential signal of the autonomous driving vehicle is available, the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state, and the first vehicle speed is a vehicle chassis speed measured by a speed sensor of the autonomous driving vehicle; Acquire a second vehicle speed of the automatic driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed; A preset linear model is trained according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, which includes: determining a model error parameter of the preset linear model according to a speed error coefficient, setting a preset speed interval, initializing the model error parameter of the preset linear model within the preset speed interval, obtaining a training sample set of the preset linear model within the preset speed interval according to the first vehicle speed within the preset speed interval and the second vehicle speed corresponding to the first vehicle speed, inputting the sample training set within the preset speed interval into the preset linear model, and optimizing the model error parameter to obtain a vehicle speed calibration model corresponding to each preset speed interval; the vehicle speed calibration model is used to receive the first vehicle speed and output a second vehicle speed corresponding to the first vehicle speed; Obtain a third vehicle speed, where the third vehicle speed is a vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet the first preset condition, and determine a calibrated speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed.

2. The method according to claim 1, characterized in that The second vehicle speed is a vehicle speed calculated according to the RTK differential signal corresponding to the first vehicle speed, and the time of the second vehicle speed is the same as the time of the first vehicle speed corresponding to the second vehicle speed.

3. The method according to claim 1, characterized in that The step of inputting the sample training set within the preset speed interval into the preset linear model and optimizing the model error parameters to obtain the vehicle speed calibration model corresponding to each preset speed interval includes: Inputting the first vehicle speed into the preset linear model, and obtaining a fourth vehicle speed output by the preset linear model according to the first vehicle speed; Calculating a value of a loss function of the preset linear model according to a fourth vehicle speed outputted from the first vehicle speed and a second vehicle speed corresponding to the first vehicle speed; The model error parameters of the preset linear model are optimized according to the value of the loss function to obtain the vehicle speed calibration model corresponding to each preset speed interval.

4. The method according to claim 1, characterized in that: The model error coefficient includes a first calibration coefficient and a second calibration coefficient, and setting the model error parameter of the preset linear model according to the speed error coefficient includes: Acquire the speed error coefficient, and set a first calibration coefficient and a second calibration coefficient of the preset linear model according to the automatic driving vehicle error coefficient, wherein the speed error coefficient is an error coefficient between the first vehicle speed and the second vehicle speed; The product of the first calibration coefficient and the first vehicle speed is obtained, and the preset linear model is determined according to the product of the first calibration coefficient and the first vehicle speed and the second calibration coefficient.

5. The method according to claim 3, characterized in that: Initializing the model error parameters of the preset linear model within the preset speed range includes: In each of the preset speed intervals, the first calibration coefficient of the preset linear model is set to 1, and the second calibration coefficient of the preset linear model is set to 0.

6. The method according to claim 1, characterized in that The speed mark based on The method of determining a calibration speed of the third vehicle speed by using a given model and the third vehicle speed includes: Determining the preset speed interval in which the third vehicle speed is located; The third vehicle speed is input into a speed calibration model corresponding to the preset speed interval to obtain a calibrated speed corresponding to the third vehicle speed.

7. A speed determination device for an autonomous driving vehicle, characterized in that: include: a first acquisition unit, configured to acquire a first vehicle speed of the autonomous driving vehicle, wherein the first vehicle speed is a vehicle travel speed obtained when the quality of an RTK differential signal of the autonomous driving vehicle satisfies a first preset condition and the heading angle of the autonomous driving vehicle satisfies a second preset condition, the first preset condition refers to a judgment condition when the quality of the RTK differential signal of the autonomous driving vehicle is available, the second preset condition refers to the heading angle of the autonomous driving vehicle when the autonomous driving vehicle is in a straight-ahead state, and the first vehicle speed is a vehicle chassis speed measured by a vehicle speed sensor of the autonomous driving vehicle; A second acquisition unit is used to acquire a second vehicle speed of the automatic driving vehicle, where the second vehicle speed is a calibrated speed corresponding to the first vehicle speed; A training unit is used to train a preset linear model according to the first vehicle speed and the second vehicle speed to obtain a vehicle speed calibration model, which includes: determining a model error parameter of the preset linear model according to a speed error coefficient, setting a preset speed interval, initializing the model error parameter of the preset linear model within the preset speed interval, obtaining a training sample set of the preset linear model within the preset speed interval according to the first vehicle speed within the preset speed interval and the second vehicle speed corresponding to the first vehicle speed, inputting the sample training set within the preset speed interval into the preset linear model, optimizing the model error parameter, so as to obtain a vehicle speed calibration model corresponding to each preset speed interval; the vehicle speed calibration model outputs a second vehicle speed corresponding to the first vehicle speed according to the first vehicle speed; A calibration unit is used to obtain a third vehicle speed, where the third vehicle speed is the vehicle speed obtained when the quality of the RTK differential signal of the autonomous driving vehicle does not meet the first preset condition, and determine a calibrated speed of the third vehicle speed based on the vehicle speed calibration model and the third vehicle speed.

8. An electronic device, characterized in that: include: processor; as well as A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: Executable code is stored on it. When the executable code is executed by a processor of an electronic device, the processor is caused to execute the method as claimed in any one of claims 1 to 6.

Images