Ai based vehicle dynamics model learning and virtual environment based driving data verification system

Abstract

According to one embodiment of the present invention, a device for verifying vehicle driving data may include a construction unit configured to construct a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle, a simulation unit configured to acquire virtual driving data by simulating the driving of the vehicle in a virtual environment based on the dynamic model, and a verification unit configured to verify the dynamic model by comparing the virtual driving data with the actual driving data.

Description

Technology Field

[0001] The present invention relates to a vehicle dynamics model, and more specifically, to a system for verifying driving data acquired in a virtual environment using an AI (Artificial Intelligence)-based vehicle dynamics model. Background Technology

[0002] Various simulations are required to verify software functions and detect unexpected behavior prior to actual vehicle testing, and accurate vehicle dynamics must be estimated to precisely control the vehicle during simulation. However, since conventional vehicle dynamics estimation is mostly performed using physics-based models, it requires sophisticated test setup and tuning. Consequently, physics-based vehicle dynamics models are difficult to implement, require significant time and cost, and often fail to respond flexibly to various conditions such as road surface, tire conditions, and vehicle load. Therefore, there is a need for a method that can reduce the costs associated with implementing vehicle dynamics models while enabling the creation of models with higher accuracy. Prior art literature

[0003] Korean Patent Publication No. 10-2614816 (Registration Date: December 13, 2023) The problem to be solved

[0004] The technical problem of the present invention is to provide an apparatus and method capable of constructing a vehicle dynamics model based on a neural network.

[0005] Another technical objective of the present invention is to provide an apparatus and method capable of verifying the reliability and accuracy of a vehicle dynamics model.

[0006] The problems that the present invention aims to solve are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0007] According to one embodiment of the present invention, a device for verifying vehicle driving data may include a construction unit configured to construct a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle, a simulation unit configured to acquire virtual driving data by simulating the driving of the vehicle in a virtual environment based on the dynamic model, and a verification unit configured to verify the dynamic model by comparing the virtual driving data with the actual driving data.

[0008] Here, the dynamic model can predict vehicle parameters at the current time point by combining features extracted from state information within the actual driving data and features extracted from control information within the actual driving data.

[0009] Additionally, the state information includes information on the longitudinal speed, lateral speed, angular velocity, wheel speed, longitudinal acceleration, and lateral acceleration of the vehicle, and the control information includes information on the steering angle, wheel torque, accelerator pedal, and brake pressure of the vehicle, and the vehicle parameters may include tire model parameters, drivetrain parameters, and moment of inertia.

[0010] In addition, the simulation unit can construct a virtual environment corresponding to the environment in which the actual driving data was acquired, and acquire driving data of a virtual vehicle implemented based on the dynamic model using a virtual sensor within the virtual environment.

[0011] In addition, the simulation unit can generate a three-dimensional terrain mesh using a triangulation method based on raster data for the region where the actual driving data was acquired, and apply a texture generated based on satellite imagery to the three-dimensional terrain mesh to construct a virtual environment corresponding to the region where the actual driving data was acquired.

[0012] In addition, the verification unit can determine the similarity between the actual driving data and the virtual driving data based on the rates of change of roll, pitch, and acceleration of the actual driving data and the virtual driving data.

[0013] According to another embodiment of the present invention, a method for a verification device to verify vehicle driving data may include the steps of: constructing a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle; simulating the driving of the vehicle in a virtual environment based on the dynamic model to acquire virtual driving data; and verifying the dynamic model by comparing the virtual driving data with the actual driving data.

[0014] According to another embodiment of the present invention, a computer program stored in a computer-readable recording medium may include instructions for the processor to perform a vehicle driving data verification method, which, when executed by a processor, comprises the steps of: constructing a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle; simulating the driving of the vehicle in a virtual environment based on the dynamic model to acquire virtual driving data; and verifying the dynamic model by comparing the virtual driving data with the actual driving data. Effects of the invention

[0015] According to one embodiment of the present invention, since a vehicle dynamics model can be constructed based on a neural network, a vehicle dynamics model having higher accuracy than existing physics-based models can be provided.

[0016] In addition, according to one embodiment of the invention, since the reliability and accuracy of the vehicle dynamics model can be verified in a virtual environment by replacing high-cost actual driving tests, the cost and time required for driving tests can be reduced, and the dynamic characteristics of the actual vehicle can be precisely estimated by enhancing the model's adaptability to various driving environments (uneven roads, wetlands, endurance roads, etc.). Brief explanation of the drawing

[0017] FIG. 1 is a drawing showing a verification device according to one embodiment of the present invention. FIG. 2 is an exemplary diagram showing an artificial neural network-based vehicle dynamics model according to one embodiment of the present invention. FIG. 3 is a diagram illustrating a method for verifying vehicle driving data according to an embodiment of the present invention. FIG. 4 is a diagram showing the process of constructing a virtual environment according to an embodiment of the present invention. FIG. 5 is a diagram illustrating the process of acquiring virtual driving data according to an embodiment of the present invention. Specific details for implementing the invention

[0018] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. Accordingly, in some embodiments, well-known processing steps, well-known device structures, and well-known techniques are not specifically described to avoid the present invention being interpreted ambiguously.

[0019] The terms used in this specification have been selected from among currently widely used general terms while considering their functions in the present invention; however, these may vary depending on the intent of those skilled in the art, case law, the emergence of new technologies, etc. Additionally, in specific cases, terms have been arbitrarily selected by the applicant, and in such cases, their meanings will be described in detail in the relevant description of the invention. Therefore, the terms used in this specification should be interpreted not merely by their names, but based on their meanings and the overall content of the present invention.

[0020] Throughout this specification, when a part is described as 'comprising' a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0021] Additionally, in this specification, the components of the verification device may refer to software or hardware components such as a Field Programmable Gate Array (FPGA) or an Application Specific Integrated Circuit (ASIC), and perform at least one function or operation. However, the components are not limited to software or hardware. The components may be configured to reside in an addressable storage medium or configured to run one or more processors. Accordingly, by example, the components include components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided by the components of the present invention may be combined into a smaller number of components or further separated into additional components.

[0022] Additionally, while terms such as first, second, third, etc. may be used in this specification to describe various components, these components are not limited by said terms. These terms are used for the purpose of distinguishing one component from other components.

[0023] Hereinafter, embodiments of the present invention are described in detail with reference to the attached drawings so that those skilled in the art can easily implement the invention. In addition, parts unrelated to the description are omitted in the drawings to clearly explain the invention. Throughout the specification, the same reference numerals refer to the same components.

[0024] FIG. 1 is a drawing showing a verification device according to one embodiment of the present invention.

[0025] Referring to FIG. 1, a verification device (100) according to one embodiment of the present invention may include a construction unit (110), a simulation unit (120), and a verification unit (130).

[0026] The construction unit (110) may be configured to construct a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle. To this end, the construction unit (110) may utilize sensor data of the vehicle that has been driven in various regions. Here, the sensor data may include data obtained from inertial measurement sensors, GPS (Global Positioning System) sensors, torque sensors, accelerator pedal position sensors, etc. The regions where actual driving data is obtained are not limited to specific regions and may include roads with uneven surfaces, Belgian roads, washboard roads, long wave roads, body twist roads, gravel roads, stone chipping, impact bulge roads, roadway impact roads, road noise roads, wetlands (mud), and marine areas. The collected actual driving data may include vehicle state information and control information and may be used for training a dynamic model based on an artificial neural network.

[0027] A dynamic model according to one embodiment of the present invention can receive state information and control information of a vehicle and estimate the dynamic characteristics of the vehicle based on an artificial neural network. For example, a dynamic model according to one embodiment of the present invention can predict vehicle parameters at a current time point by extracting state features from state information of actual driving data and extracting control features from control information, and then combining them. Here, the state information may include information regarding the longitudinal speed, lateral speed, angular velocity, wheel speed, longitudinal acceleration, and lateral acceleration of the vehicle, and the control information may include information regarding the steering angle, wheel torque, accelerator pedal, and brake pressure of the vehicle. Furthermore, the vehicle parameters may include tire model parameters, drivetrain parameters, and moment of inertia.

[0028] The simulation unit (120) may be configured to simulate the driving of a vehicle in a virtual environment based on a dynamic model constructed by the construction unit (110) to acquire virtual driving data. For example, the simulation unit (120) may construct a virtual environment corresponding to the environment in which actual driving data was acquired, and acquire driving data of a virtual vehicle implemented based on the dynamic model using a virtual sensor within the virtual environment. To this end, the simulation unit (120) may, for example, generate a three-dimensional terrain mesh using a triangulation method based on raster data for the region in which actual driving data was acquired, and apply a texture generated based on satellite imagery to the three-dimensional terrain mesh to construct a virtual environment corresponding to the region in which actual driving data was acquired.

[0029] The verification unit (130) may be configured to verify the dynamic model by comparing virtual driving data obtained from the simulation unit (120) with actual driving data of the vehicle. To this end, the verification unit (130) may determine the similarity between the actual driving data and the virtual driving data based on the rates of change of roll, pitch, and acceleration of the actual driving data and the virtual driving data.

[0030] FIG. 2 is an exemplary diagram showing an artificial neural network-based vehicle dynamics model according to one embodiment of the present invention.

[0031] A dynamic model according to one embodiment of the present invention can extract features from vehicle state information and control information, respectively, through an independent backbone model, and predict vehicle parameters by combining them. To this end, a verification device according to one embodiment of the present invention can normalize data acquired in a real driving environment (real driving data). For example, in the case of vehicle longitudinal speed data, the verification device can divide all vehicle longitudinal speed values ​​based on the maximum vehicle longitudinal speed value among the time series data to convert them into values ​​between 0 and 1, and then use them as input data.

[0032] A dynamic model according to one embodiment of the present invention may include a first backbone model (210), a second backbone model (220), a concatenated layer (230), and a fully connected layer (240), as shown in FIG. 2.

[0033] The first backbone model (210) can extract a feature map of vehicle state data (vehicle longitudinal speed, vehicle lateral speed, angular velocity, wheel speed, vehicle longitudinal acceleration, vehicle lateral acceleration, etc.) from a past time point to a present time point, and the second backbone model (220) can extract a feature map of vehicle control data (steering angle, wheel torque, accelerator pedal, brake pressure, etc.) from a past time point to a present time point. To this end, the first backbone model (210) and the second backbone model (220) can be implemented as models capable of processing / analyzing time series data, such as LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit).

[0034] The feature map extracted by the first backbone model (210) (feature map for vehicle state data) and the feature map extracted by the second backbone model (220) (feature map for vehicle control data) can be combined by a combination layer (230), and the dynamic model can predict vehicle parameters at the current time (tire model parameters, drivetrain parameters, moment of inertia, etc.) based on the fully connected layer (230). The fully connected layer may be similar to a deep neural network structure and may include a hidden layer and an output layer.

[0035] FIG. 3 is a diagram showing a method for verifying vehicle driving data according to an embodiment of the present invention, FIG. 4 is a diagram showing a process for constructing a virtual environment according to an embodiment of the present invention, and FIG. 5 is a diagram showing a process for acquiring virtual driving data according to an embodiment of the present invention.

[0036] First, referring to FIG. 3, a verification device according to one embodiment of the present invention can construct a dynamic model that estimates the dynamic characteristics of a vehicle based on an artificial neural network using actual driving data of the vehicle (S300).

[0037] For example, when sensor data from a vehicle driven in actual regions is input, the verification device can preprocess it and use it for training an artificial neural network-based dynamic model. When the artificial neural network-based dynamic model receives the preprocessed data (actual driving data), it can extract state features and control features from state information and control information, respectively, using an independent backbone model. By repeatedly performing the process of combining these to predict vehicle parameters at the current time point, it can learn the actual driving data of the corresponding vehicle for various environments.

[0038] Once the construction of the dynamic model is completed through the process described above, the verification device can acquire virtual driving data by simulating the driving of the vehicle in a virtual environment based on the dynamic model (S310).

[0039] For example, as shown in FIG. 4, the verification device can generate a three-dimensional terrain mesh using a triangulation method based on raster data for a region where actual driving data was acquired (S400), generate a texture for the region based on satellite imagery (S410), and then render the texture of the satellite imagery onto the three-dimensional terrain mesh (S420). Once a virtual environment corresponding to the region where actual driving data was acquired is established through the above-described process, the verification device can verify the accuracy of the dynamic model by reproducing the actual driving data within the virtual environment.

[0040] For example, the verification device can set initial vehicle dynamics in a simulation model based on vehicle specification data, as shown in FIG. 5 (S500). At this time, the verification device can review the connection stability between the components of the simulation model and the input data, input tire models and drivetrain parameters, and then reflect the vehicle size (overall length, overall width, wheelbase) and curb weight. Subsequently, the verification device can check whether the simulation model meets the required dynamic level, inspect the vehicle's steering responsiveness, acceleration and deceleration performance, brake response, etc., and, if necessary, further tune the dynamic model.

[0041] Finally, the verification device can simulate the driving of a virtual vehicle in a virtual environment using data predicted based on an optimized dynamic model (S510), and then acquire virtual driving data using a virtual sensor (S520). Here, the type of virtual driving data corresponds to the type of actual driving data, and each data may include information regarding speed, acceleration, steering angle, etc.

[0042] When virtual driving data is acquired through the process described above, the verification device can verify the dynamic model by comparing the virtual driving data with the actual driving data of the vehicle (S320). For example, the verification device can determine the similarity between the actual driving data and the virtual driving data based on the rates of change of roll, pitch, and acceleration of the actual driving data and the virtual driving data. The verification device can determine the similarity between the two data by performing time series analysis or frequency analysis on the actual driving data and the virtual driving data, and can verify the reliability and practicality of the dynamic model by comparing the parameters of the actual driving data and the virtual driving data.

[0043] Through this process, the verification device according to one embodiment of the present invention can perform dynamic analysis reflecting realistic vehicle behavior and can test the vehicle's movement and control response.

[0044] Meanwhile, each step included in the method for verifying vehicle driving data by the vehicle driving data verification device according to the above-described embodiment may be implemented in a computer program including instructions for a processor to perform such steps and / or a computer-readable recording medium on which said computer program is recorded.

[0045] Combinations of each step of the flowchart attached to the present invention may be performed by computer program instructions. Since these computer program instructions may be loaded into the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing equipment, the instructions performed through the processor of the computer or other programmable data processing equipment create means for performing the functions described in each step of the flowchart. Since these computer program instructions may also be stored in a computer-available or computer-readable recording medium that can be oriented toward the computer or other programmable data processing equipment to implement the function in a specific way, the instructions stored in the computer-available or computer-readable recording medium may also produce a manufactured item containing instruction means for performing the function described in each step of the flowchart. Since computer program instructions can be loaded onto a computer or other programmable data processing equipment, instructions that execute a computer or other programmable data processing equipment by performing a series of operation steps on the computer or other programmable data processing equipment to create a process executed by the computer can also provide steps for executing the functions described in each step of the flowchart.

[0046] Additionally, each step may represent a module, segment, or part of code containing one or more executable instructions for executing a specified logical function(s). Furthermore, in some alternative embodiments, the functions mentioned in the steps may occur out of order. For example, two steps described in succession may be performed substantially simultaneously, or the steps may be performed in reverse order depending on the corresponding function.

[0047] The above description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential quality of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by such embodiments. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within the equivalent scope shall be interpreted as being included within the scope of rights of the present invention. Explanation of the symbols

[0048] 100: Verification device 110: Construction Department 120: Mock Department 130: Verification Department 210: 1st Backbone Model 220: 2nd Backbone Model 230: Combination layer 240: Fully Connected Layer

Claims

Claim 1 A device for verifying vehicle driving data, comprising: a construction unit configured to construct a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle; a simulation unit configured to acquire virtual driving data by simulating the driving of the vehicle in a virtual environment based on the dynamic model; and a verification unit configured to verify the dynamic model by comparing the virtual driving data with the actual driving data. Claim 2 The device according to claim 1, wherein the dynamic model predicts vehicle parameters at the current time point by combining features extracted from state information within the actual driving data and features extracted from control information within the actual driving data. Claim 3 A device according to claim 2, wherein the state information includes information regarding the longitudinal speed, lateral speed, angular speed, wheel speed, longitudinal acceleration, and lateral acceleration of the vehicle, the control information includes information regarding the steering angle, wheel torque, accelerator pedal, and brake pressure of the vehicle, and the vehicle parameters include tire model parameters, drivetrain parameters, and moment of inertia. Claim 4 The device according to claim 1, wherein the simulation unit constructs a virtual environment corresponding to the environment in which the actual driving data is acquired, and acquires driving data of a virtual vehicle implemented based on the dynamic model using a virtual sensor within the virtual environment. Claim 5 In claim 4, the simulation unit is characterized by generating a three-dimensional terrain mesh using a triangulation method based on raster data for the region where the actual driving data was acquired, and constructing a virtual environment corresponding to the region where the actual driving data was acquired by applying a texture generated based on satellite imagery to the three-dimensional terrain mesh. Claim 6 A device according to claim 1, wherein the verification unit determines the similarity between the actual driving data and the virtual driving data based on the rate of change of roll, pitch, and acceleration of the actual driving data and the virtual driving data. Claim 7 A method for a verification device to verify vehicle driving data comprises: a step of constructing a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle; a step of acquiring virtual driving data by simulating the driving of the vehicle in a virtual environment based on the dynamic model; and a step of verifying the dynamic model by comparing the virtual driving data with the actual driving data. Claim 8 In claim 7, the method is characterized in that the dynamic model predicts vehicle parameters at the current time point by combining features extracted from state information within the actual driving data and features extracted from control information within the actual driving data. Claim 9 A method according to claim 8, wherein the state information includes information regarding the longitudinal speed, lateral speed, angular velocity, wheel speed, longitudinal acceleration, and lateral acceleration of the vehicle, the control information includes information regarding the steering angle, wheel torque, accelerator pedal, and brake pressure of the vehicle, and the vehicle parameters include tire model parameters, drivetrain parameters, and moment of inertia. Claim 10 A method according to claim 7, wherein the acquiring step comprises: a step of constructing a virtual environment corresponding to the environment in which the actual driving data is acquired; and a step of acquiring driving data of a virtual vehicle implemented based on the dynamic model using a virtual sensor within the virtual environment. Claim 11 A method according to claim 10, wherein the step of constructing comprises: a step of generating a three-dimensional terrain mesh using a triangulation method based on raster data for the region where the actual driving data was acquired; and a step of constructing a virtual environment corresponding to the region where the actual driving data was acquired by applying a texture generated based on satellite imagery to the three-dimensional terrain mesh. Claim 12 A method according to claim 7, wherein the verification step comprises determining the similarity between the actual driving data and the virtual driving data based on the rate of change of roll, pitch, and acceleration between the actual driving data and the virtual driving data. Claim 13 A computer program stored on a computer-readable recording medium, wherein the computer program, when executed by a processor, includes instructions for the processor to perform a vehicle driving data verification method comprising the steps of: constructing a dynamic model that estimates the dynamic characteristics of the vehicle based on an artificial neural network using actual driving data of the vehicle; simulating the driving of the vehicle in a virtual environment based on the dynamic model to acquire virtual driving data; and verifying the dynamic model by comparing the virtual driving data with the actual driving data.

Images