Methods, apparatus, electronic devices and storage media for positioning accuracy estimation

Abstract

This application provides a method, apparatus, device, and storage medium for positioning accuracy estimation, applicable to fields such as mapping, navigation, autonomous driving, vehicle networking, intelligent transportation, and cloud computing. The positioning accuracy estimation method includes: acquiring first driving information of a vehicle collected by sensors; obtaining first position information of the vehicle based on the first driving information, and a first intermediate variable used in determining the first position information; determining a target accuracy estimation model based on the first driving information and the first intermediate variable in a pre-trained accuracy estimation model; inputting the first position information and the first intermediate variable into the target accuracy estimation model to obtain second position information of the vehicle, and the accuracy error of the first position information relative to the second position information. This application embodiment can effectively evaluate the accuracy error of high-precision positioning.

Description

Technical Field

[0001] This application relates to the field of transportation, and more specifically, to methods, apparatus, devices, and storage media for estimating positioning accuracy. Background Technology

[0002] Intelligentization is one of the main themes of current automotive development. The progress of automotive intelligence depends on the maturity of vehicle sensors, algorithms, and decision-making platforms. The combination of high-precision positioning and high-precision maps provides vehicles with accurate absolute position information, which complements the relative position information from sensors, working together to improve the safety of intelligent driving. As the level of automotive intelligence continues to increase, the importance of high-precision positioning becomes increasingly prominent.

[0003] Assessing accuracy error is a crucial aspect of high-precision positioning. The accuracy error of a single positioning operation—the magnitude of the potential error—can significantly aid subsequent vehicle control, collision avoidance, intelligent speed control, path planning, and behavioral decision-making. However, currently, there is no effective method for evaluating the accuracy error of high-precision positioning. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for estimating positioning accuracy, which can effectively evaluate the accuracy error of high-precision positioning.

[0005] In a first aspect, embodiments of this application provide a method for estimating positioning accuracy, including: Acquire the vehicle's initial driving information collected by the sensors; Based on the first driving information, the first location information of the vehicle is obtained, as well as the first intermediate variable used in determining the first location information; Based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in a pre-trained accuracy estimation model; the accuracy estimation model is obtained by training a machine learning model based on a training sample set, wherein the training samples in the training sample set include the position information of the sample vehicle and the intermediate variables used in determining the position information; The first location information and the first intermediate variable are input into the target accuracy estimation model to obtain the second location information of the vehicle and the accuracy error of the first location information relative to the second location information.

[0006] Secondly, embodiments of this application provide a method for training a model, including: Acquire the second driving information of the sample vehicle collected by the sensor in the first moment; Based on the second driving information, the third location information of the sample vehicle is obtained, as well as the second intermediate variable used in determining the third location information; Obtain the fourth location information of the sample vehicle collected by the positioning device at the first time. A training sample set is determined, the training sample set including the third position information, the fourth position information, and the second intermediate variable; The accuracy estimation model is trained based on the training sample set.

[0007] Thirdly, embodiments of this application provide a device for estimating positioning accuracy, comprising: The acquisition unit is used to acquire the vehicle's first driving information collected by the sensors; The processing unit is configured to obtain the first location information of the vehicle based on the first driving information, and a first intermediate variable used in determining the first location information. The determining unit is configured to determine a target accuracy estimation model based on the first driving information and the first intermediate variable in a pre-trained accuracy estimation model; the accuracy estimation model is obtained by training a machine learning model based on a training sample set, wherein the training samples in the training sample set include the position information of the sample vehicle and the intermediate variables used in determining the position information; The target accuracy estimation model is used to take the first position information and the first intermediate variable as input to obtain the second position information of the vehicle, and the accuracy error of the first position information relative to the second position information.

[0008] Fourthly, an apparatus for training a model is provided, characterized in that it comprises: The first acquisition unit is used to acquire the second driving information of the sample vehicle collected by the sensor in the first moment; The processing unit is configured to obtain the third location information of the sample vehicle based on the second driving information, and a second intermediate variable used in determining the third location information; The second acquisition unit is used to acquire the fourth location information of the sample vehicle collected by the positioning device at the first time. A determining unit is used to determine a training sample set, the training sample set including the third position information, the fourth position information, and the second intermediate variable; A training unit is used to train the accuracy estimation model based on the training sample set.

[0009] Fifthly, embodiments of this application provide an electronic device, including: Processor, adapted to implement computer instructions; and, A memory that stores computer instructions adapted for loading by a processor and executing the methods of the first or second aspect described above.

[0010] In a sixth aspect, embodiments of this application provide a computer-readable storage medium storing computer instructions that, when read and executed by a processor of a computer device, cause the computer device to perform the methods described in the first or second aspect.

[0011] Fifthly, embodiments of this application provide a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods described in the first or second aspect.

[0012] Based on the above technical solutions, the embodiments of this application can obtain the first position information of the vehicle according to the first driving information, and obtain the first intermediate variable used in the process of determining the first position information. Then, based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in a pre-trained accuracy estimation model, and the first position information and the first intermediate variable are input into the target accuracy estimation model to obtain the second position information of the vehicle and the accuracy error of the first position information relative to the second position information, thereby effectively evaluating the accuracy error of high-precision positioning. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of a system architecture according to an embodiment of this application; Figure 2 A schematic flowchart illustrating a method for training a model provided in an embodiment of this application; Figure 3 This is a schematic diagram of a network architecture applicable to an embodiment of this application; Figure 4 A schematic flowchart illustrating a method for obtaining training sample data provided in an embodiment of this application; Figure 5 A schematic flowchart illustrating a positioning accuracy estimation method provided in an embodiment of this application; Figure 6 This application provides a specific example of a method for estimating positioning accuracy. Figure 7 A schematic block diagram of a positioning accuracy estimation device provided in an embodiment of this application; Figure 8 A schematic block diagram of an apparatus for training a model provided in an embodiment of this application; Figure 9 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0015] It should be understood that in the embodiments of this application, "B corresponding to A" means that B is associated with A. In one implementation, B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0016] In the description of this application, unless otherwise stated, "at least one" means one or more, and "multiple" means two or more. Additionally, "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0017] It should also be understood that the descriptions of "first", "second", etc. appearing in the embodiments of this application are only for illustration and to distinguish the objects being described, and there is no order to them. They do not indicate any special limitation on the number of devices in the embodiments of this application, and cannot constitute any limitation on the embodiments of this application.

[0018] It should also be understood that specific features, structures, or characteristics relating to embodiments in the specification are included in at least one embodiment of this application. Furthermore, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0019] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products, or devices.

[0020] The embodiments of this application are applied to the field of artificial intelligence technology.

[0021] Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or computers-controlled machines to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess perception, reasoning, and decision-making capabilities.

[0022] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0023] With the research and advancement of artificial intelligence (AI) technology, AI is being studied and applied in various fields, such as smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, autonomous driving, drones, robots, smart healthcare, and smart customer service. It is believed that with the development of technology, AI will be applied in more fields and play an increasingly important role.

[0024] This application's embodiments relate to autonomous driving technology within artificial intelligence. Autonomous driving technology relies on the collaborative efforts of artificial intelligence, computer vision, radar, monitoring devices, and a global positioning system (GPS) to enable computers to automatically and safely operate motor vehicles without any active human intervention. Autonomous driving technology typically includes high-precision mapping, environmental perception, behavioral decision-making, path planning, and motion control. Autonomous driving technology has broad application prospects. Specifically, the technical solution provided in this application's embodiments relates to a technique for evaluating positioning accuracy errors, which can be used to evaluate the positioning accuracy of autonomous driving systems.

[0025] This application's embodiments may also relate to Machine Learning (ML) in artificial intelligence technology. ML is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and many other disciplines. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0026] Figure 1 This is a schematic diagram of a system architecture involved in an embodiment of this application, such as... Figure 1 As shown, the system architecture may include user equipment 101, data acquisition equipment 102, training equipment 103, execution equipment 104, database 105, and content library 106.

[0027] The data acquisition device 102 is used to read training data from the content library 106 and store the read training data in the database 105. The training data involved in this embodiment includes the location information #1 and location information #2 of the sample vehicle, and intermediate variables used in determining the location information #1. The location information #1 is determined based on the driving information of the sample vehicle collected by the sensor, and the location information #2 is the location information of the sample vehicle collected by the positioning device.

[0028] Training device 103 trains a machine learning model based on training data maintained in database 105, enabling the trained machine learning model to effectively evaluate positioning accuracy. The machine learning model obtained by training device 103 can be applied to different systems or devices.

[0029] Additionally, refer to Figure 1 The execution device 104 is equipped with an I / O interface 107 for data interaction with external devices. For example, it receives vehicle driving information collected by sensors from the user equipment 101 via the I / O interface. The calculation module 109 in the execution device 104 uses this driving information and a trained machine learning model to obtain the vehicle's positioning accuracy. The machine learning model can then send the results back to the user equipment 101 via the I / O interface.

[0030] User equipment 101 may include intelligent vehicles, in-vehicle terminals, mobile phones, tablets, laptops, handheld computers, mobile internet devices (MIDs), or other terminal devices.

[0031] The execution device 104 can be a server.

[0032] For example, the server can be a rack server, blade server, tower server, or cabinet server, etc. The server can be a standalone test server or a test server cluster composed of multiple test servers.

[0033] There may be one or more servers. When there are multiple servers, at least two servers are used to provide different services, and / or at least two servers are used to provide the same service, such as providing the same service in a load-balanced manner. This application does not limit this.

[0034] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. A server can also become a node in a blockchain.

[0035] In this embodiment, the execution device 104 is connected to the user equipment 101 via a network. The network can be an intranet, the Internet, the Global System for Mobile Communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network, Bluetooth, Wi-Fi, voice communication network, or other wireless or wired networks.

[0036] It should be noted that, Figure 1 This is merely a schematic diagram of a system architecture provided in this application embodiment, and the positional relationships between the devices, components, modules, etc. shown in the figure do not constitute any limitation. In some embodiments, the data acquisition device 102, user device 101, training device 103, and execution device 104 may be the same device. The database 105 may be distributed across one server or multiple servers, and the content library 106 may be distributed across one server or multiple servers.

[0037] The relevant terms involved in the embodiments of this application are described below.

[0038] High-precision positioning generally refers to positioning with an accuracy at the decimeter or centimeter level or even higher. It can provide vehicles with highly accurate positioning results and is one of the core technologies indispensable for safe driving such as autonomous driving and remote driving. High-precision positioning plays an important role in precise lateral / longitudinal positioning of vehicles, obstacle detection and collision avoidance, intelligent speed control, path planning, and behavioral decision-making.

[0039] High-precision positioning allows for accurate determination of a vehicle's absolute position using a high-precision absolute reference frame. This high-precision absolute reference frame can be, for example, a high-precision map. The map layer of a high-precision map contains numerous road attribute elements with centimeter-level precision, including but not limited to information on road edges, lane edges, and center lines. During operation, the vehicle can perform precise navigation based on the information in the high-precision map.

[0040] Multi-source fusion positioning is a technology that integrates multiple positioning techniques based on information fusion strategies. It can combine various positioning methods, including satellite positioning, wireless communication signal positioning, and sensor positioning, to obtain better fusion positioning results than a single positioning scheme. High-precision positioning can be achieved through multi-source fusion positioning.

[0041] Satellite positioning is a technology that uses satellites (such as GNSS) for positioning; wireless communication signal positioning is a technology that uses wireless communication signals (such as WiFi signals, Ultra Wideband (UWB) signals) for positioning; and sensor positioning is a technology that uses information collected by sensors (such as vision sensors, vehicle speed sensors, and inertial measurement unit (IMU) sensors) for positioning.

[0042] Visual positioning refers to the technology of location using information collected by visual sensors. Visual positioning uses visual sensors to identify and perceive road attribute elements in a high-precision map positioning layer, and relies on visual algorithms to calculate vehicle location information. Visual positioning can highly reuse high-precision maps and sensors such as cameras without requiring additional hardware deployment, resulting in a significant cost advantage.

[0043] Inertial positioning refers to the technology of positioning using information collected by IMU sensors. Inertial positioning measures the angular rate and acceleration of a vehicle using IMU sensors, and automatically calculates the instantaneous velocity and position of the vehicle using Newton's laws of motion. It has the characteristics of not relying on external information, not radiating energy to the outside world, being resistant to interference, and having good concealment.

[0044] Global Navigation Satellite System (GNSS): This term broadly refers to satellite navigation systems, including global, regional, and augmentation systems such as the Global Positioning System (GPS), GLONASS, Europe's Galileo, and the BeiDou Navigation Satellite System, as well as related augmentation systems like the Wide Area Augmentation System (WAAS), the European Geostationary Navigation Overlay System (EGNOS), and the Multifunctional Transport Satellite Augmentation System (MSAS). It can also encompass other satellite navigation systems under construction or planned for the future. The international GNSS system is a complex, multi-system, multi-layered, and multi-mode integrated system.

[0045] Typically, conventional satellite positioning achieves location by simultaneously receiving signals from multiple satellites. However, satellite signals experience fluctuations as they pass through the ionosphere and troposphere, introducing errors and limiting positioning accuracy to the meter level. Ground-based augmentation stations can use real-time kinematic (RTK) carrier phase differential technology to calculate satellite positioning errors and perform further position corrections, thereby improving positioning accuracy. For example, satellite positioning accuracy can be improved from the meter level to the centimeter level.

[0046] In addition, satellite positioning is prone to signal weakness or loss when there are obstructions on the ground. In such cases, vehicles can use inertial positioning, visual positioning, or other positioning functions to ensure that the navigation system continues to work.

[0047] RTK carrier phase differential technology is a real-time method for processing the carrier phase observations of two measurement stations. It involves sending the carrier phase data acquired by the base station to the user receiver, and then calculating the coordinates by difference. RTK carrier phase differential technology employs a dynamic real-time carrier phase differential method, enabling centimeter-level positioning accuracy in the field. This provides new measurement principles and methods for engineering layout, topographic mapping, and various control surveys, improving operational efficiency.

[0048] Assessing accuracy error is a crucial aspect of high-precision positioning. The accuracy error of a single positioning operation—the magnitude of the potential error—can significantly aid subsequent vehicle control, collision avoidance, intelligent speed control, path planning, and behavioral decision-making. However, currently, there is no effective method for evaluating the accuracy error of high-precision positioning.

[0049] In view of this, embodiments of this application provide a method, apparatus, device, and storage medium for estimating positioning accuracy, which can effectively evaluate the accuracy error of high-precision positioning.

[0050] Specifically, embodiments of this application can acquire first driving information of a vehicle collected by sensors, and obtain first position information of the vehicle based on the first driving information, as well as first intermediate variables used in determining the first position information. Then, based on the first driving information and the first intermediate variables, a target accuracy estimation model is determined in a pre-trained accuracy estimation model, and the first position information and the first intermediate variables are input into the target accuracy estimation model to obtain second position information of the vehicle and the accuracy error of the first position information relative to the second position information.

[0051] The accuracy estimation model is obtained by training a machine learning model using a training sample set. The training sample set includes the location information of the sample vehicle and intermediate variables used in determining the location information. For example, the training sample set may include third and fourth location information of the sample vehicle, and a second intermediate variable used in determining the third location information. The third location information is determined based on second driving information of the sample vehicle collected by sensors at a first time. The fourth location information includes the location information of the sample vehicle collected by the positioning device at that first time.

[0052] Therefore, the embodiments of this application can obtain the first location information of the vehicle based on the first driving information, and obtain the first intermediate variable used in determining the first location information. Then, based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in a pre-trained accuracy estimation model. The first location information and the first intermediate variable are input into the target accuracy estimation model to obtain the second location information of the vehicle and the accuracy error of the first location information relative to the second location information, thereby effectively evaluating the accuracy error of high-precision positioning.

[0053] In this embodiment of the application, the accuracy estimation model can predict the true location information of the vehicle collected by the positioning device and estimate the accuracy error of the positioning information relative to the true location information based on the vehicle's driving information collected by the sensor and the intermediate variables used in determining the positioning information.

[0054] The positioning accuracy estimation method of this application embodiment can be divided into two stages: an offline modeling stage and an online estimation stage. In the offline modeling stage, an accuracy estimation model is trained based on the collected sample vehicle data. In the online estimation stage, vehicle driving information is collected in real time, and the vehicle's positioning accuracy is estimated based on this driving information and the accuracy estimation model established in the offline modeling stage.

[0055] The following sections describe each stage in detail with reference to the accompanying drawings.

[0056] First, let's introduce the offline modeling stage.

[0057] Figure 2 This is a schematic flowchart illustrating a method 200 for training a model, provided in an embodiment of this application. Method 200 can be executed by any electronic device with data processing capabilities. For example, the electronic device can be implemented as a server or terminal device; or, for example, the electronic device can be implemented as... Figure 1 The training device 103 in this application is not limited thereto.

[0058] Figure 3 This is a schematic diagram of a network architecture applicable to embodiments of this application, including a vehicle information acquisition module 301, a real-time positioning module 302, an information statistics module 303, a truth value acquisition module 304, a modeling module 305, an information judgment module 306, and an accuracy estimation model 307. The vehicle information acquisition module 301, real-time positioning module 302, information statistics module 303, truth value acquisition module 304, and modeling module 305 can be used for model training in the offline modeling stage. The following will combine... Figure 3 The method 200 for training the model is described.

[0059] like Figure 2 As shown, method 200 includes steps 210 to 250.

[0060] 210, acquire the second driving information of the sample vehicle collected by the sensor at the first moment.

[0061] For example, it can be made by Figure 3 The vehicle information acquisition module 301 acquires the second driving information of the sample vehicle collected by the sensor at a first moment. The first moment can be the current time or a period before, without limitation. After acquiring the second driving information, the vehicle information acquisition module 301 can send the second driving information to the real-time positioning module 302. Optionally, the second driving information can also be sent to the information statistics module 303.

[0062] In some embodiments, the second driving information includes driving information of a sample vehicle collected by at least one of an inertial measurement unit (IMU) sensor, a vehicle speed sensor, and a vision sensor.

[0063] For example, the sensor can be a variety of sensors installed on the sample vehicle to collect vehicle driving information, such as GNSS sensors, IMU sensors, vehicle speed sensors, vision sensors, etc., or it can be other sensor devices.

[0064] The driving information of the sample vehicle collected by the GNSS sensor may include, for example, the longitude, latitude, DOP value, accuracy value, whether the current positioning is valid, and whether the current positioning is fixed.

[0065] In general, a GNSS sensor is considered effective when it receives GNSS signals normally under unobstructed or weakly obstructed conditions (such as on a normal driving road), thus ensuring valid positioning. However, in conditions with significant obstruction (such as in tunnels, overpasses, or mountainous areas), the GNSS sensor may fail to receive GNSS signals, resulting in invalid positioning. In this case, the GNSS sensor may be considered ineffective or malfunctioning.

[0066] The solution obtained by a GNSS sensor using RTK carrier phase differential technology can be called a fixed solution. When GNSS signals are blocked (e.g., on mountain roads), making RTK carrier phase differential technology unusable, other types of solutions can be obtained, such as floating-point solutions. Whether the current positioning is fixed refers to whether the current positioning used RTK carrier phase differential technology to obtain a fixed solution. Among these, the fixed solution has the highest accuracy, reaching the centimeter level.

[0067] The IMU sensor collects driving information of the sample vehicle, which may include acceleration values ​​in the x, y, and z axes of the accelerometer and angular velocities in the x, y, and z axes of the gyroscope.

[0068] The vehicle speed sensor collects driving information of the sample vehicle, which may include, for example, the vehicle's speed, direction, and magnitude.

[0069] The driving information of the sample vehicle collected by the visual sensor may include, for example, the equation of the lane line where the sample vehicle is located, the type of the lane line (such as solid line, dashed line, white line or yellow line, etc.), and the coordinates of obstacles.

[0070] 220. Based on the second driving information, the third location information of the sample vehicle is obtained, as well as the second intermediate variable used in determining the third location information.

[0071] For example, it can be made by Figure 3 The real-time positioning module 302 obtains the third location information of the sample vehicle and the second intermediate variable used in determining the third location information based on the second driving information. After obtaining the third location information and the second intermediate variable, the real-time positioning module 302 can send the third location information and the second intermediate variable to the modeling module 305.

[0072] In some embodiments, the real-time positioning module 302 can obtain second driving information of the sample vehicle collected by each sensor from the vehicle information acquisition module 301, and can combine it with other information to obtain third location information of the sample vehicle. For example, other information may include map information, such as a high-precision map.

[0073] In some embodiments, the third location information may include the longitude, latitude, and vehicle heading angle of the sample vehicle.

[0074] The vehicle steering angle can be the angle by which the front of the vehicle deviates from a preset direction. For example, the angle by which the front of the vehicle deviates clockwise from due north can be used as the vehicle steering angle.

[0075] In the process of determining the third location information of the sample vehicle based on the second driving information mentioned above, some intermediate variables can also be obtained, namely the second intermediate variables mentioned above.

[0076] In some embodiments, third location information and a second intermediate variable can be obtained based on the second driving information and map information described above. The second intermediate variable may include error information between lane information determined based on the second driving information and lane information of the sample vehicle in the map information.

[0077] For example, the second intermediate variable may include the following two types: 1) Intermediate variables derived from information collected by visual sensors and high-precision maps, such as the difference between sensor lane width and map lane width, lane line optimization distance, initial optimization error, and final optimization error; 2) Intermediate variables derived from information collected by GNSS sensors and visual sensors and high-precision maps, such as the probability of a location point in each lane, the lane with the highest probability of a location point, and whether the lane information given by the GNSS sensor and the lane information given by the visual sensor are consistent.

[0078] In some embodiments, the second intermediate variable may further include at least one of the covariance of the optimization algorithm used to determine the third location information and the statistical value of the second driving information in the second time period.

[0079] For example, the second intermediate variable may also include the following two categories: 3) Optimize the covariance of the algorithm, such as the velocity covariance, the covariance of IMU sensor bias, and the covariance of attitude estimation (such as the estimated longitude, latitude, and heading angle of the vehicle). 4) Statistical values ​​of parameters collected by the statistical sensors over a period of time, such as the statistical values ​​of information collected by the GNSS sensor within 20 seconds, the mean and variance of vehicle speed within 1 second, the mean and variance of acceleration values ​​in the x, y, and z axes of the accelerometer and the mean and variance of angular velocity values ​​in the x, y, and z axes of the gyroscope within 10 seconds, and the distance between the last positioning point obtained by the GNSS sensor and the positioning result.

[0080] In the above 1), the lane line optimization distance can refer to the optimized distance of the lane line obtained by visual positioning when aligning the lane line of the vehicle's lane obtained by visual positioning with the lane line of the vehicle in the high-precision map; the initial optimization error can refer to the error between the two before aligning the lane line of the vehicle's lane obtained by visual positioning with the lane line of the vehicle in the high-precision map; and the final optimization error can refer to the error between the two after aligning the lane line of the vehicle's lane obtained by visual positioning with the lane line of the vehicle in the high-precision map.

[0081] In 2) above, the GNSS sensor can use RTK carrier phase differential technology for high-precision positioning, obtaining the probability of the vehicle's location point in each lane, and can then select the lane with the highest location point probability as the vehicle's lane. Based on the optimization results in 1) above, the visual sensor can also further obtain the probability of the vehicle's location point in each lane, and select the lane with the highest location point probability as the vehicle's lane.

[0082] In section 3) above, different optimization algorithms and different optimization objectives can be used, and this application does not limit them. For example, the optimization algorithm can be an optimization algorithm based on the overall optimization form, such as optimizing the information collected by all sensors within a certain period of time, or it can be an optimization algorithm based on filtering, such as optimizing the information collected by sensors at different time points.

[0083] In 4) above, the statistical values ​​of the information collected by the GNSS sensor within 20 seconds include, but are not limited to, the effective ratio, the fixed ratio, the mean and variance of the accuracy value, the mean and variance of the DOP value, the accuracy of the last GNSS positioning point, the DOP value, whether it is fixed, and whether it is effective.

[0084] Among them, the effective ratio can refer to the proportion of effective positioning by the GNSS sensor to all positioning results obtained by the GNSS sensor; the fixed ratio can refer to the proportion of fixed solutions in the positioning results obtained by the GNSS sensor to all positioning results.

[0085] For example, it can be made by Figure 3The information statistics module 303 collects statistical values ​​of the parameters given by the vehicle information collection module 301 over a period of time, obtaining the statistical values ​​mentioned in 4) above. This application does not limit this. After obtaining the statistical values, the information statistics module 303 can send the statistical values ​​to the real-time positioning module 302 or to the modeling module 305.

[0086] 230, obtain the fourth location information of the sample vehicle collected by the positioning device at the first moment.

[0087] For example, the positioning device can be Figure 3 The truth acquisition module 304 can be deployed on the sample vehicle. Correspondingly, the fourth location information can also be called truth location information, that is, the fourth location information can be considered as the true location of the sample vehicle, which is not limited in this application.

[0088] Positioning devices can have high positioning accuracy, such as SPAN-ISA-100C or SPAN-CPT. Generally, the higher the accuracy of the positioning device used, the better the positioning accuracy estimation.

[0089] In some embodiments, the fourth location information may include the longitude, latitude, and vehicle heading angle of the sample vehicle.

[0090] 240. Determine the training sample set, which includes the third position information, the fourth position information, and the second intermediate variable mentioned above.

[0091] For example, it can be made by Figure 3 The modeling module 305 determines the training sample set, which may include a large number of training samples. Each training sample may include the third position information obtained in step 220, the second intermediate variable, and the fourth position information obtained in step 230.

[0092] Because different vehicles have different Operational Design Domains (ODDs), for example, an autonomous vehicle may have one control system designed for driving in an urban environment and another designed for driving on a highway. Therefore, a large amount of data needs to be collected in different scenarios, such as at least highways, urban expressways, ordinary urban roads, and highway or urban tunnel sections.

[0093] This application embodiment can collect a large amount of data generated by sample vehicles during operation in different scenarios to generate a training sample set. For example, driving information of sample vehicles during operation can be collected by sensors on a sufficient number of road sections such as highways, urban expressways, and ordinary urban roads, as well as location information of sample vehicles during operation collected by positioning devices. As an example, data can be collected for at least 5,000 kilometers, or more than 10,000 kilometers, in each scenario.

[0094] As one possible implementation, when determining the training sample set, the positioning results obtained by the real-time positioning module 302 (such as third location information) and the positioning results obtained by the ground truth acquisition module 304 (such as fourth location information) can be aligned by time, and all intermediate variables of the real-time positioning module 302 can be extracted. As an example, one training sample can be generated for each time point, that is, each training sample can include the positioning results obtained by the real-time positioning module 302, the positioning results obtained by the ground truth acquisition module 304, and the intermediate variables of the real-time positioning module 302 at the corresponding time point.

[0095] In some embodiments, for tunnel sections, the data of sample vehicles collected in normal road sections (i.e., road sections without tunnels) can be used to simulate tunnel sections to obtain training sample data corresponding to tunnel sections.

[0096] Specifically, in tunnel sections, GNSS signals are blocked, making it impossible to obtain data collected by GNSS sensors. Relying solely on other sensors, such as IMU sensors, results in accumulated errors that cannot meet accuracy requirements. Furthermore, the accuracy of location information collected by positioning devices decreases in tunnel sections.

[0097] For autonomous or remote driving technologies, the requirement for maintaining high positioning accuracy in tunnel sections is typically limited to a certain period or distance within the tunnel. Therefore, tunnel sections can be simulated by artificially invalidating portions of the data collected by GNSS sensors. The invalidated portions of these GNSS sensor data correspond to tunnel sections, thus generating training sample data. Furthermore, the fourth position information of the sample vehicle collected by the positioning device in this training sample is sufficiently accurate compared to the position information collected by the positioning device in a real tunnel.

[0098] Figure 4 A schematic flowchart of a method 400 for obtaining training sample data corresponding to a tunnel section is shown. In method 400, training sample data corresponding to a tunnel section can be generated by periodically invalidating and validating the driving information of sample vehicles collected by GNSS sensors over time.

[0099] It should be understood that Figure 4 The steps or operations for obtaining a training sample set corresponding to a tunnel section are illustrated, but these steps or operations are merely examples. Other operations or variations of the operations shown in the figures can also be performed in the embodiments of this application. Furthermore, Figure 4 The steps in the diagram may be performed in a different order than those shown in the diagram, and it is possible that not all operations shown in the diagram need to be performed.

[0100] like Figure 4 As shown, method 400 includes steps 401 to 409.

[0101] 401, retrieves all valid files acquired by the GNSS sensor.

[0102] For example, during the road test phase, the driving information of the sample vehicle collected by sensors during a section of road in a tunnel-free scenario can be acquired. Here, the driving information of the vehicle collected by the GNSS sensor can be saved in file format.

[0103] 402. Is the time since the start of the file greater than 200 seconds? Here, the file start time can be the time when the sample vehicle begins operation, that is, the time when the GNSS sensor begins collecting the vehicle's driving information. By ensuring that the file start time is greater than 200 seconds, the training sample data of the acquired simulated tunnel can include the driving information of the sample vehicle collected by the GNSS sensor in the 200 seconds before entering the tunnel.

[0104] 403, was the speed ever greater than 20 km / h? Here, by confirming that the speed has exceeded 20 km / h, it helps to ensure that the vehicle is in a normal driving state, thereby ensuring that the vehicle driving information acquired by the sensors is meaningful.

[0105] 404, invalidating files acquired by GNSS sensors.

[0106] Optionally, when the files collected by the GNSS sensor are invalidated, information can also be read from files formed by vehicle driving information collected by other sensors on the sample vehicle besides the GNSS sensor (such as IMU sensor, vision sensor and speed sensor, etc.), as well as the location information of the sample vehicle collected by the positioning device.

[0107] 405, has the invalid time exceeded 300 seconds? Here, the invalid time is 300 seconds, which is equivalent to a simulated travel time of 300 seconds in the tunnel section. When the invalid time exceeds 300 seconds, the next step is to execute step 406.

[0108] 406, Enables the file acquired by the GNSS sensor.

[0109] It should be understood that after the invalid time exceeds 300 seconds, the file collected by the GNSS sensor is set to valid, which is equivalent to simulating the scenario of a vehicle driving on a normal road after exiting a tunnel.

[0110] 407, Read file information.

[0111] Here, corresponding to the scenario where the sample vehicle is on a normal road section, information can be read from the file formed by the driving information of the sample vehicle collected by the GNSS sensor and other sensors on the vehicle, as well as the location information of the sample vehicle collected by the positioning device.

[0112] 408, has the end of file been reached? The process ends when the end of the file is reached. If the end of the file is not reached, proceed to step 409.

[0113] 409. Has the effective time exceeded 200 seconds? If the valid time exceeds 200s, the next step is to execute step 404, which means continuing to invalidate the files collected by the GNSS sensor. In this way, by periodically (i.e., alternately) invalidating the GNSS sensor files for 300s and validating them for 200s, it is possible to generate sample vehicle data for tunnel sections based on the sample vehicle data collected from normal road sections.

[0114] Typically, once an autonomous vehicle enters a tunnel, the driver can take over. Therefore, the most important information in a tunnel is the vehicle's driving information collected by sensors during a period of time (e.g., 200 seconds) or a certain distance before entering the tunnel. Based on this, data of sample vehicles in tunnel sections can be generated by periodically setting the GNSS sensor files to invalid for 300 seconds and valid for 200 seconds.

[0115] It should be understood that the times of 200s and 300s and the speed of 20km / h mentioned above are specific examples given to facilitate understanding of the solutions in the embodiments of this application. The above time or speed values ​​can also be replaced with other values, and the embodiments of this application are not limited thereto.

[0116] Therefore, by periodically invalidating and validating the files collected by the GNSS sensor over time, this embodiment of the application can simulate tunnel sections and generate sample vehicle data corresponding to the tunnel sections. That is, the road sections corresponding to the invalidated parts in the files collected by the GNSS sensor can be considered as tunnel sections, thereby obtaining a training sample set corresponding to the tunnel sections, and ensuring that the accuracy of the location information collected by the positioning device in the training sample set is high enough.

[0117] 250, The accuracy estimation model is trained based on the training sample set.

[0118] For example, it can be made by Figure 3 The modeling module 305 trains the accuracy estimation model 307 based on the training sample set. For example, the modeling module 305 can input the training samples into the accuracy estimation model 307 to update the parameters of the accuracy estimation model 307.

[0119] In some embodiments, different accuracy estimation models can be trained for different scenarios, thereby achieving the best accuracy error estimation results in each scenario. For example, the following three scenarios may be included.

[0120] Scenario 1: Non-tunnel scene + scene with map Scene 2: Tunnel scene + scene with map Scenario 3: Non-tunnel scene + map-less scene For example, the map mentioned above can be a high-precision map, and this application does not limit it.

[0121] In scenario 1 above, the GNSS sensor is effective and can combine the vehicle's driving information collected by the sensor with map information for real-time positioning.

[0122] In scenario 1, the accuracy estimation model can specifically be a first accuracy estimation model. The third position information and the second intermediate variable in the training samples corresponding to the first accuracy estimation model are determined based on the second driving information and map information (such as a high-precision map). The second driving information includes the driving information of the sample vehicle collected by the GNSS sensor, and the second intermediate variable includes the error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information.

[0123] As a specific example, in the training samples of the first accuracy estimation model, the second intermediate variable may include the four types of intermediate variables mentioned in step 220 above: 1) intermediate variables derived from information collected by visual sensors and high-precision maps, 2) intermediate variables derived from information collected by GNSS sensors, visual sensors and high-precision maps, 3) the covariance of the optimization algorithm, and 4) statistical values ​​of parameters collected by statistical sensors over a period of time.

[0124] In scenario 2 above, if the GNSS sensor fails, the vehicle's driving information collected by other sensors besides the GNSS sensor can be combined with map information for real-time positioning.

[0125] In scenario 2, the accuracy estimation model can specifically be a second accuracy estimation model. The third position information and the second intermediate variable in the training samples corresponding to the second accuracy estimation model are determined based on the valid part of the second driving information and map information (such as a high-precision map). The driving information part of the sample vehicle collected by the GNSS sensor in the second driving information is set to invalid. The second intermediate variable includes the error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information.

[0126] Optionally, the driving information of the sample vehicles collected by the GNSS sensor in the second driving information is periodically set to invalid and valid according to time.

[0127] Specifically, the method for invalidating the driving information of sample vehicles collected by GNSS sensors can be found in the description of step 240 above, and will not be repeated here.

[0128] As a specific example, the second intermediate variable in the training samples of the second accuracy estimation model can include four types of intermediate variables from step 220 above: 1) intermediate variables derived from information collected by the visual sensor and the high-precision map; 3) the covariance of the optimization algorithm; and 4) statistical values ​​of parameters collected by the statistical sensor over a period of time (excluding parameters collected by the GNSS sensor). For example, the statistical values ​​in 4) could be the mean and variance of the vehicle speed within 1 second, the mean and variance of the acceleration values ​​in the x, y, and z axes of the accelerometer within 10 seconds, and the mean and variance of the angular velocity values ​​in the x, y, and z axes of the gyroscope.

[0129] In scenario 3 above, the GNSS sensor is effective, but because there is no map information, the vehicle driving information collected by the sensor cannot be combined with the map information for real-time positioning.

[0130] In scenario 3, the accuracy estimation model can specifically be a third accuracy estimation model. The third position information and the second intermediate variable in the training samples corresponding to the third accuracy estimation model are determined based on the second driving information, which includes the driving information of the sample vehicles collected by the GNSS sensor.

[0131] As a specific example, in the training samples of the third precision estimation model, the second intermediate variable may include four types of intermediate variables, such as 3) the covariance of the optimization algorithm and 4) the statistical values ​​of the parameters collected by the statistical sensor over a period of time.

[0132] After preparing training sample sets for each of the above scenarios, we can model each scenario separately, that is, train an accuracy estimation model for each scenario based on the training sample set corresponding to that scenario. Here, the accuracy estimation model can be a machine learning model, such as a random forest model, an XGBoost model, a deep neural network model, etc., without limitation.

[0133] The following is an example of a set of algorithm configurations when the accuracy estimation model uses the xgboost model: Sample size: 300,000 'booster': 'gbtree' 'objective': 'reg:gamma' or 'reg:squarederror' 'gamma': 0.1, 'max_depth': 6, 'lambda': 3, 'subsample': 0.7, 'colsample_bytree': 0.7, 'min_child_weight': 3, 'silent': 1, 'eta': 0.1 For example, the third position information, fourth position information, and second intermediate variable from the training samples in each scenario can be input into the accuracy estimation model of the corresponding scenario. The accuracy estimation model can obtain the fifth position information of the sample vehicle and the accuracy error of the third position information based on the third position information and the second intermediate variable. Here, the accuracy error can be the error between the third position information and the fifth position information, and this application does not limit it to this.

[0134] Optionally, during model training, a first loss can be determined based on the fifth position information and the aforementioned fourth position information. Optionally, during model training, a second loss can be determined based on the precision error of the third position information relative to the fifth position information and the precision error of the third position information relative to the fourth position information. Then, the parameters of the accuracy estimation model can be updated based on at least one of the first and second losses.

[0135] In some embodiments, the aforementioned accuracy error includes at least one of lateral distance error, longitudinal distance error, and orientation angle error.

[0136] Taking the error between the third and fourth location information as an example, the component of the distance between the third and fourth location information decomposed into the direction perpendicular to the road is called the lateral distance error, and the component decomposed into the direction of the road is called the longitudinal distance error. As an example, the direction of the positioning device (such as a truth acquisition device) can be taken as the road direction. The road direction can also be called the driving direction, without limitation.

[0137] When the fourth location information includes longitude lon0, latitude lat0, and vehicle heading angle heading0, and the third location information includes longitude lon, latitude lat, and vehicle heading angle heading, the accuracy error of the third location information relative to the fourth location information can be determined to include lateral distance error disthorizontal, longitudinal distance error distvetical, and heading angle error.

[0138] The heading angle error can be the difference between the vehicle heading angle heading0 and the vehicle heading angle heading, i.e., heading0 - heading.

[0139] The calculation process for the lateral distance error (disthorizontal) and longitudinal distance error (distvetical) is as follows: First, calculate the distance *dist* between the two positions corresponding to the fourth and third position information. Specifically, this can be obtained based on the Haversine formula: (1) (2) (3) in, Convert longitude or latitude to radians The average radius of the Earth is r = 6371.393. For arcsine operation, For sine calculation, This is for cosine operations.

[0140] Then, the orientation angle (point_degree) between the two positions corresponding to the fourth and third position information can be calculated. This orientation angle (point_degree) is the angle by which the direction formed by the two positions corresponding to the fourth and third position information deviates from the direction of the positioning device, and its value ranges from 0 to 360°. The specific calculation process for the orientation angle (point_degree) is as follows: (4) (5) (6) (7) Where x represents the vertical component of the distance between the two locations corresponding to the fourth and third location information, and y represents the component of that distance in the direction of the positioning device. For inverse cosine calculation, the corresponding angle can be returned based on the signs of x and y, and % indicates the remainder operation. dlon in formulas (5) and (6) can be obtained from formula (4).

[0141] Finally, the lateral distance error (disthorizontal) and longitudinal distance error (distvetical) can be calculated based on the orientation angle (point_degree). The specific calculation process is as follows: (8) when hour, (9) (10) when hour, (11) (12) when hour, (13) (14) when hour, (13) (14) The error between the third position information and the fifth position information can be calculated using formulas (1)-(14), and will not be repeated here.

[0142] Another type of scenario exists: tunnel scenarios combined with map-free scenarios. In this scenario, even high-precision sensors, such as IMU sensors, cannot guarantee sufficiently high positioning accuracy. Therefore, in this scenario, positioning can be set to unavailable, and there is no need to estimate positioning accuracy.

[0143] Therefore, in this embodiment, the accuracy estimation model is trained by using the third position information of the sample vehicle determined by the driving information of the sample vehicle collected by sensors, the second intermediate variable used in determining the third position information, and the fourth position information of the sample vehicle collected by the positioning device, thus obtaining a trained accuracy estimation model. In this embodiment, the accuracy estimation model integrates the vehicle position information determined by the driving information collected by sensors, and the intermediate variable used in determining the position information, enabling the estimation of the accuracy error of the position information.

[0144] After training to obtain an accurate estimation model, the online estimation stage can begin. The online estimation stage is described below.

[0145] Figure 5 This is a schematic flowchart illustrating a positioning accuracy estimation method 500 provided in an embodiment of this application. Method 500 can be executed by any electronic device with data processing capabilities. For example, the electronic device can be implemented as a server or terminal device; or, for example, the electronic device can be implemented as... Figure 1 The calculation module 109 in this application is not limited in this respect.

[0146] In some embodiments, the electronic device may include (e.g., deploy) a machine learning model, which may be the accuracy estimation model described above. See also... Figure 3 The vehicle information acquisition module 301, real-time positioning module 302, information statistics module 303, information judgment module 306, accuracy estimation model 307, and accuracy estimation module 308 can be used to estimate the accuracy of vehicle positioning. Below, we will combine... Figure 3 The method for estimating positioning accuracy is described in section 500.

[0147] like Figure 5 As shown, method 500 includes steps 510 to 540.

[0148] 510, acquire the first driving information of the vehicle collected by the sensor.

[0149] For example, it can be made by Figure 3 The vehicle information acquisition module 301 acquires the first driving information of the vehicle collected by the sensor. The sensor can collect the first driving information of the vehicle's current state and a period of time prior, without limitation. After acquiring the first driving information, the vehicle information acquisition module 301 can send the first driving information to the real-time positioning module 302 and the information judgment module 306. Optionally, the first driving information can also be sent to the information statistics module 303.

[0150] In some embodiments, the first driving information includes vehicle driving information collected by at least one of an inertial measurement unit (IMU) sensor, a vehicle speed sensor, and a vision sensor.

[0151] Specifically, the first driving information collected by the sensor is... Figure 2 The second driving information collected by the sensor in step 210 is similar and can be referred to the description above, so it will not be repeated here.

[0152] 520. Based on the first driving information, the first location information of the vehicle is obtained, as well as the first intermediate variable used in determining the first location information.

[0153] For example, it can be made by Figure 3 The real-time positioning module 302 obtains the vehicle's first location information and the first intermediate variable used in determining the first location information based on the aforementioned first driving information. After obtaining the first location information and the first intermediate variable, the real-time positioning module 302 can send the first location information and the first intermediate variable to the accuracy estimation model 307.

[0154] In some embodiments, the first location information includes the vehicle's longitude, latitude, and vehicle heading angle.

[0155] In some embodiments, first location information and first intermediate variable can be obtained based on first driving information and map information, wherein the first intermediate variable includes error information between lane information determined based on the first driving information and lane information of the vehicle in the map information.

[0156] In some embodiments, the first intermediate variable may include at least one of the covariance of the optimization algorithm used to determine the first location information and the statistical value of the first driving information in a first time period.

[0157] For example, it can be made by Figure 3 The information statistics module 303 collects statistical values ​​of parameters given by the vehicle information collection module 301 over a period of time to obtain the statistical value of the first driving information in the first time period. After obtaining the statistical value, the information statistics module 303 can send the statistical value to the real-time positioning module 302 or send the statistical variable to the accuracy estimation model 307.

[0158] Specifically, the first location information and Figure 2 Similar to the third position information in step 220, the first intermediate variable is... Figure 2 The second intermediate variable in step 220 is similar and can be referred to the description above, so it will not be repeated here.

[0159] 530. Based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in the pre-trained accuracy estimation model. The accuracy estimation model is obtained by training a machine learning model based on a training sample set. The training samples in the training sample set include the position information of the sample vehicle and the intermediate variables used in determining the position information.

[0160] Specifically, the training sample set may include the third location information, the fourth location information, and the second intermediate variable used in determining the third location information. The third location information is determined based on the second driving information of the sample vehicle collected by the sensor at the first moment. The fourth location information includes the location information of the sample vehicle collected by the positioning device at the first moment.

[0161] Specifically, the training process of this accuracy estimation model can be found in the above text. Figure 2 The description in the text will not be repeated here.

[0162] In some embodiments, the pre-trained accuracy estimation model includes multiple models trained separately for different scenarios. In this case, the target accuracy estimation model can be determined from the pre-trained accuracy estimation model based on the first driving information of the vehicle collected by the sensors and the first intermediate variable in the process of determining the first position information; that is, the applicable target accuracy estimation model to be invoked is determined according to the specific application scenario.

[0163] For example, it can be made by Figure 3 The information judgment module 306 determines the applicable accuracy estimation model to be invoked based on the first driving information and the first intermediate variable, that is, by combining the first driving information and whether there is currently available map information, such as a high-precision map.

[0164] As an example, when the first driving information includes the driving information of the vehicle collected by the GNSS sensor, and there is currently available map information (such as a high-precision map), the first accuracy estimation model can also be determined as the target accuracy estimation model, that is, the first accuracy estimation model under scenario 1 above is determined to be called.

[0165] As another example, when the first driving information does not include the vehicle driving information collected by the GNSS sensor, and there is currently available map information (such as a high-precision map), the second accuracy estimation model can also be determined as the target accuracy estimation model, that is, the second accuracy estimation model under scenario 2 above is determined to be called.

[0166] As another example, when the first driving information includes the vehicle's driving information collected by GNSS sensors, and there is currently no available map information (such as a high-precision map), the third precision estimation model can also be determined as the target precision estimation model, that is, the third precision estimation model under scenario 3 above is determined to be called.

[0167] Specifically, the first-precision estimation model, the second-precision estimation model, and the third-precision estimation model can be found in [reference needed]. Figure 2 The description of step 250 will not be repeated here.

[0168] 540. Input the first position information and the first intermediate variable into the target accuracy estimation model to obtain the second position information of the vehicle and the accuracy error of the first position information relative to the second position information.

[0169] For example, see [link to example]. Figure 3 After the information judgment module 306 determines the applicable target accuracy estimation model to be called based on the first driving information of the vehicle obtained by the sensors in the vehicle information acquisition module 301 and the availability of map information, such as a high-precision map, the real-time positioning module 302 and the information statistics module 303 can send the relevant input features required by the benchmark accuracy estimation model to the target accuracy estimation model.

[0170] For example, when calling the first accuracy estimation model, the first intermediate variable may include the following intermediate variables used in determining the first location information: 1) intermediate variables derived from information collected by the visual sensor and the high-precision map; 2) intermediate variables derived from information collected by the GNSS sensor, the visual sensor and the high-precision map; 3) the covariance of the optimization algorithm; and 4) the statistical values ​​of the parameters collected by the statistical sensor over a period of time.

[0171] For example, when calling the second accuracy estimation model, the first intermediate variable may include the following intermediate variables used in determining the first location information: 1) intermediate variables derived from information collected by the visual sensor and the high-precision map, 3) the covariance of the optimization algorithm, and 4) statistical values ​​of parameters collected by the statistical sensor over a period of time (excluding parameters collected by the GNSS sensor).

[0172] For example, when calling the third precision estimation model, the first intermediate variable may include the following intermediate variables used in determining the first location information: 3) the covariance of the optimization algorithm and 4) the statistical values ​​of the parameters collected by the statistical sensor over a period of time.

[0173] After inputting relevant features, the first accuracy estimation model, the second accuracy estimation model, or the third accuracy estimation model can output the predicted second position information of the vehicle, as well as the accuracy error of the first position information relative to the second position.

[0174] The second location information can be either the vehicle's location information collected by the positioning device as predicted by the accuracy estimation model, or the vehicle's actual location information predicted by the accuracy estimation model; there is no limitation on either. For example, the second location information includes the vehicle's longitude, latitude, and heading angle.

[0175] For example, the target accuracy estimation model can predict the true location information of the vehicle collected by the positioning device and estimate the accuracy error of the positioning information relative to the true location information by using the vehicle's positioning location information determined by the vehicle's driving information collected by the sensor and the intermediate variables used in determining the positioning location information.

[0176] Therefore, the embodiments of this application can obtain the first location information of the vehicle based on the first driving information, and obtain the first intermediate variable used in determining the first location information. Then, based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in a pre-trained accuracy estimation model. The first location information and the first intermediate variable are input into the target accuracy estimation model to obtain the second location information of the vehicle and the accuracy error of the first location information relative to the second location information, thereby effectively evaluating the accuracy error of high-precision positioning.

[0177] Compared to the traditional method of using optimized covariance, the positioning accuracy estimation method of this application significantly improves both the CEP90 and mean of the accuracy error. CEP90 refers to the value at the 90th percentile when all accuracy errors are sorted from smallest to largest. For example, comparing a dataset of approximately 2-3 hours per day across different road segments over 20 days, the mean accuracy error determined by this application is improved by an average of approximately 0.5m compared to the covariance method.

[0178] Figure 6 A specific example of a positioning accuracy estimation method 600 provided in this application embodiment is illustrated. Method 600 can be executed by any electronic device with data processing capabilities. For example, the electronic device can be implemented as a server or terminal device; or, for example, the electronic device can be implemented as... Figure 1 The calculation module 109 in this application is not limited in this respect.

[0179] It should be understood that Figure 6 The steps or operations of a method for estimating positioning accuracy are shown, but these steps or operations are merely examples. Other operations or variations of the operations shown in the figures can also be performed in the embodiments of this application. Furthermore, Figure 6 The steps in the diagram may be performed in a different order than those shown in the diagram, and may not necessarily be performed in the correct order. Figure 6 All operations within.

[0180] like Figure 6 As shown, method 600 includes steps 601 to 610.

[0181] 601, Obtain the vehicle's initial driving information.

[0182] Specifically, step 601 can be found in the description of step 510, and will not be repeated here.

[0183] Is 602 a tunnel? Specifically, the vehicle's location within a tunnel can be determined based on the first driving information in section 601. For example, if the first driving information includes vehicle driving information collected by GNSS sensors, it can be determined that the vehicle is not in a tunnel; if the first driving information does not include vehicle driving information collected by GNSS sensors, it can be determined that the vehicle is in a tunnel.

[0184] If the vehicle is in a tunnel, proceed to step 603. If the vehicle is not in a tunnel, proceed to step 605.

[0185] 603, is there a high-precision map? In the tunnel scenario, the next step is to determine whether a high-definition map exists. If a high-definition map exists, proceed to step 604; otherwise, proceed to step 610.

[0186] 604, Invoke the second precision estimation model.

[0187] 605, is there a high-precision map? In non-tunnel scenarios, the next step is to determine whether a high-definition map exists. If a high-definition map exists, proceed to step 606; otherwise, proceed to step 607.

[0188] 606, invoke the first precision estimation model.

[0189] 607, invoke the third precision estimation model.

[0190] Specifically, the first-precision estimation model, the second-precision estimation model, and the third-precision estimation model can be found in [reference needed]. Figure 2 and Figure 5 The description in the text will not be repeated here.

[0191] 608, Parameter Statistics.

[0192] Specifically, the driving parameters in the first driving information in step 601 can be statistically analyzed to obtain the statistical values ​​of each driving parameter.

[0193] 609, Real-time location.

[0194] Specifically, the vehicle's first location information can be determined based on the first driving information obtained in step 601 and the statistical values ​​obtained in step 608. The real-time positioning result (i.e., this first location information), and the intermediate variables used in determining this first location information, can be input into the corresponding accuracy estimation model. For details, the parameters input to different accuracy estimation models can be found in [reference needed]. Figure 5 The description in the text will not be repeated here.

[0195] 610, Location unavailable.

[0196] In other words, in tunnel scenarios and without high-precision maps, positioning is determined to be unavailable.

[0197] 611, Determine the accuracy error.

[0198] Specifically, the accuracy error of real-time positioning results in different scenarios can be determined by calling the accuracy estimation model in different scenarios.

[0199] Therefore, the embodiments of this application can obtain the first location information of the vehicle based on the first driving information, and obtain the first intermediate variable used in determining the first location information. Then, based on the first driving information and the first intermediate variable, a target accuracy estimation model is determined in a pre-trained accuracy estimation model. The first location information and the first intermediate variable are input into the target accuracy estimation model to obtain the second location information of the vehicle and the accuracy error of the first location information relative to the second location information, thereby effectively evaluating the accuracy error of high-precision positioning.

[0200] The specific embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this application, various simple modifications can be made to the technical solutions of this application, and these simple modifications all fall within the protection scope of this application. For example, the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this application will not describe the various possible combinations separately. Furthermore, various different embodiments of this application can also be arbitrarily combined, as long as they do not violate the spirit of this application, they should also be considered as the content disclosed in this application.

[0201] It should also be understood that, in the various method embodiments of this application, the sequence numbers of the above processes do not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. It should be understood that these sequence numbers can be interchanged where appropriate so that the embodiments of this application described can be implemented in a sequence other than those illustrated or described.

[0202] The method embodiments of this application have been described in detail above. The following description, in conjunction with... Figures 7 to 9 The following describes in detail the device embodiments of this application.

[0203] Figure 7 This is a schematic block diagram of a positioning accuracy estimation device 700 provided in an embodiment of this application. Figure 7 As shown, the positioning accuracy estimation device 700 may include an acquisition unit 710, a processing unit 720, and an accuracy estimation model 730.

[0204] The acquisition unit 710 is used to acquire the first driving information of the vehicle collected by the sensor; The processing unit 720 is configured to obtain first location information of the vehicle based on the first driving information, and a first intermediate variable used in determining the first location information. The determining unit 730 is used to determine a target accuracy estimation model in a pre-trained accuracy estimation model based on the first driving information and the first intermediate variable; the accuracy estimation model is obtained by training a machine learning model based on a training sample set, wherein the training samples in the training sample set include the position information of the sample vehicle and the intermediate variables used in determining the position information; The target accuracy estimation model 740 is used to take the first position information and the first intermediate variable as input to obtain the second position information of the vehicle and the accuracy error of the first position information relative to the second position information.

[0205] In some embodiments, the processing unit 720 is specifically used for: Based on the first driving information and map information, the first location information and the first intermediate variable are obtained, wherein the first intermediate variable includes error information between the lane information determined based on the first driving information and the lane information of the vehicle in the map information.

[0206] In some embodiments, the first driving information includes driving information of the vehicle collected by a Global Navigation Satellite System (GNSS) sensor.

[0207] Specifically, the determining unit 730 is used for: Based on the first driving information and the first intermediate variable, the first accuracy estimation model is determined as the target accuracy estimation model, wherein the location information and the intermediate variable in the training sample set of the first accuracy estimation model are determined based on the driving information of the sample vehicles and the map information collected by the GNSS sensor.

[0208] In some embodiments, the determining unit 730 is specifically used for: Based on the first driving information and the first intermediate variable, the second accuracy estimation model is determined as the target accuracy estimation model. The location information and the intermediate variable in the training sample set corresponding to the second accuracy estimation model are determined based on the valid part of the driving information of the sample vehicle collected by the GNSS sensor and the map information. The driving information part of the sample vehicle collected by the GNSS sensor is set to invalid.

[0209] In some embodiments, the driving information of the sample vehicle collected by the GNSS sensor is periodically set to invalid and valid over time.

[0210] In some embodiments, the first driving information includes driving information of the vehicle collected by GNSS sensors.

[0211] Specifically, the determining unit 730 is used for: The third accuracy estimation model is determined based on the first driving information and the first intermediate variable, wherein the location information and the intermediate variable in the training sample set corresponding to the third accuracy estimation model are determined based on the driving information of the sample vehicles collected by the GNSS sensor.

[0212] In some embodiments, the first driving information includes driving information of the vehicle collected by at least one of an inertial measurement unit (IMU) sensor, a vehicle speed sensor, and a vision sensor.

[0213] In some embodiments, the first intermediate variable includes at least one of the covariance of the optimization algorithm used to determine the first location information and the statistical value of the first driving information in a first time period.

[0214] In some embodiments, the first location information includes the vehicle's first longitude, first latitude, and first vehicle heading angle, and the second location information includes the vehicle's second longitude, second latitude, and second vehicle heading angle.

[0215] In some embodiments, the accuracy error includes at least one of lateral distance error, longitudinal distance error, and orientation angle error.

[0216] It should be understood that the device embodiments and method embodiments can correspond to each other, and similar descriptions can be referred to the method embodiments. To avoid repetition, they will not be repeated here. Specifically, in this embodiment, the positioning accuracy estimation device 700 can correspond to the corresponding subject executing the method 500 or 600 of the embodiments of this application, and the foregoing and other operations and / or functions of each module in the device 700 are respectively to implement the corresponding processes in the method 500 or 600 above. For the sake of brevity, they will not be repeated here.

[0217] Figure 8 This is a schematic block diagram of the apparatus 800 for training a model provided in an embodiment of this application. Figure 8 As shown, the apparatus 800 for training the model may include a first acquisition unit 810, a processing unit 820, a second acquisition unit 830, a determination unit 840, and a training unit 850.

[0218] The first acquisition unit 810 is used to acquire the second driving information of the sample vehicle collected by the sensor in the first moment; The processing unit 820 is configured to obtain the third location information of the sample vehicle based on the second driving information, and a second intermediate variable used in determining the third location information; The second acquisition unit 830 is used to acquire the fourth location information of the sample vehicle collected by the positioning device at the first time. The determining unit 840 is used to determine a training sample set, the training sample set including the third position information, the fourth position information and the second intermediate variable; Training unit 850 is used to train the accuracy estimation model based on the training sample set.

[0219] In some embodiments, the second driving information includes driving information of the sample vehicle collected by a Global Navigation Satellite System (GNSS) sensor; Specifically, the processing unit 820 is used for: Based on the second driving information and map information, the third location information and the second intermediate variable are obtained, wherein the second intermediate variable includes error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information; Specifically, training unit 850 is used for: The first accuracy estimation model is trained based on the training sample set.

[0220] In some embodiments, the second driving information includes driving information of the sample vehicle collected by a Global Navigation Satellite System (GNSS) sensor; The processing unit 820 is further configured to: invalidate the driving information portion of the sample vehicle collected by the GNSS sensor; Based on the valid portion of the second driving information and the map information, the third location information and the second intermediate variable are obtained, wherein the second intermediate variable includes error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information; Specifically, the training unit is used to train the second accuracy estimation model based on the training sample set.

[0221] In some embodiments, the second driving information includes driving information of the sample vehicle collected by a Global Navigation Satellite System (GNSS) sensor; wherein, the training unit 850 is specifically used for: The third accuracy estimation model is trained based on the training sample set.

[0222] In some embodiments, the second driving information includes driving information of the sample vehicle collected by at least one of an inertial measurement unit (IMU) sensor, a vehicle speed sensor, and a vision sensor.

[0223] In some embodiments, the second intermediate variable includes at least one of the covariance of the optimization algorithm used to determine the third location information and the statistical value of the second driving information in the second time period.

[0224] It should be understood that the device embodiments and method embodiments can correspond to each other, and similar descriptions can be referred to the method embodiments. To avoid repetition, they will not be repeated here. Specifically, in this embodiment, the device 800 for training the model can correspond to the corresponding subject that executes the method 200 of the present application embodiments, and the foregoing and other operations and / or functions of each module in the device 800 are respectively for implementing the corresponding processes in the method 200 above. For the sake of brevity, they will not be repeated here.

[0225] The apparatus and system of this application embodiments have been described above from the perspective of functional modules in conjunction with the accompanying drawings. It should be understood that these functional modules can be implemented in hardware, in software instructions, or in a combination of hardware and software modules. Specifically, the steps of the method embodiments in this application can be completed by integrated logic circuits in the processor's hardware and / or by software instructions. The steps of the methods disclosed in this application embodiments can be directly embodied as being executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. Optionally, the software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps in the above method embodiments.

[0226] like Figure 9 This is a schematic block diagram of the electronic device 1100 provided in the embodiments of this application.

[0227] like Figure 9 As shown, the electronic device 1100 may include: The system includes a memory 1110 and a processor 1120. The memory 1110 stores computer programs and transfers the program code to the processor 1120. In other words, the processor 1120 can retrieve and run the computer program from the memory 1110 to implement the methods described in the embodiments of this application.

[0228] For example, the processor 1120 can be used to execute the steps in the method 200 described above according to the instructions in the computer program.

[0229] In some embodiments of this application, the processor 1120 may include, but is not limited to: General-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0230] In some embodiments of this application, the memory 1110 includes, but is not limited to: Volatile memory and / or non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0231] In some embodiments of this application, the computer program may be divided into one or more modules, which are stored in the memory 1110 and executed by the processor 1120 to complete the method provided in this application. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the electronic device 1100.

[0232] Optional, such as Figure 9 As shown, the electronic device 1100 may further include a transceiver 1130, which can be connected to the processor 1120 or the memory 1110. The processor 1120 can control the transceiver 1130 to communicate with other devices; specifically, it can send information or data to other devices or receive information or data sent by other devices. The transceiver 1130 may include a transmitter and a receiver. The transceiver 1130 may further include an antenna, and the number of antennas may be one or more.

[0233] It should be understood that the various components in the electronic device 1100 are connected through a bus system, which includes a data bus, a power bus, a control bus, and a status signal bus.

[0234] According to one aspect of this application, a communication device is provided, including a processor and a memory for storing a computer program, the processor for calling and running the computer program stored in the memory, causing the encoder to perform the method of the above-described method embodiment.

[0235] According to one aspect of this application, a computer storage medium is provided that stores a computer program thereon, which, when executed by a computer, enables the computer to perform the methods of the above-described method embodiments. Alternatively, embodiments of this application also provide a computer program product containing instructions that, when executed by a computer, cause the computer to perform the methods of the above-described method embodiments.

[0236] According to another aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method described in the above-described method embodiments.

[0237] In other words, when implemented using software, it can be implemented wholly or partially in the form of a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video disc (DVD)), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0238] It is understood that specific embodiments of this application may involve user information and other related data. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0239] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0240] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or modules may be electrical, mechanical, or other forms.

[0241] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. For example, the functional modules in the various embodiments of this application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0242] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for estimating positioning accuracy, characterized in that, include: Acquire the vehicle's initial driving information collected by the sensors; Based on the first driving information, the first location information of the vehicle is obtained, as well as the first intermediate variable used in determining the first location information; Based on the availability of the first driving information and map information, a target accuracy estimation model is determined from a pre-trained accuracy estimation model; the accuracy estimation model is obtained by training a machine learning model based on a training sample set, wherein the training samples in the training sample set include the location information of the sample vehicle and intermediate variables used in determining the location information; The first location information and the first intermediate variable are input into the target accuracy estimation model to obtain the second location information of the vehicle and the accuracy error of the first location information relative to the second location information.

2. The method according to claim 1, characterized in that, The step of obtaining the first location information of the vehicle based on the first driving information, and the first intermediate variable used in determining the first location information, includes: Based on the first driving information and map information, the first location information and the first intermediate variable are obtained, wherein the first intermediate variable includes error information between the lane information determined based on the first driving information and the lane information of the vehicle in the map information.

3. The method according to claim 2, characterized in that, The first driving information includes driving information of the vehicle collected by the Global Navigation Satellite System (GNSS) sensors; The step of determining the target accuracy estimation model from a pre-trained accuracy estimation model based on the availability of the first driving information and map information includes: Based on the availability of the first driving information and the map information, the first accuracy estimation model is determined as the target accuracy estimation model, wherein the location information and the intermediate variables in the training sample set of the first accuracy estimation model are determined based on the driving information of the sample vehicles and the map information collected by the GNSS sensor.

4. The method according to claim 2, characterized in that, The step of determining the target accuracy estimation model from the pre-trained accuracy estimation model based on the availability of the first driving information and map information includes: Based on the availability of the first driving information and the map information, the second accuracy estimation model is determined as the target accuracy estimation model. The location information and the intermediate variables in the training sample set corresponding to the second accuracy estimation model are determined based on the valid portion of the driving information of the sample vehicles collected by the GNSS sensor and the map information. The portion of the driving information of the sample vehicles collected by the GNSS sensor is set to invalid.

5. The method according to claim 4, characterized in that, The driving information of the sample vehicle collected by the GNSS sensor is periodically set to invalid and valid over time.

6. The method according to claim 1, characterized in that, The first driving information includes the driving information of the vehicle collected by the GNSS sensor; The step of determining the target accuracy estimation model from a pre-trained accuracy estimation model based on the availability of the first driving information and map information includes: The third accuracy estimation model is determined as the accuracy estimation model based on the fact that the first driving information and the map information are unavailable. The location information and the intermediate variables in the training sample set corresponding to the third accuracy estimation model are determined based on the driving information of the sample vehicles collected by the GNSS sensor.

7. The method according to any one of claims 1-6, characterized in that, The first driving information includes driving information of the vehicle collected by at least one of the following sensors: an inertial measurement unit (IMU) sensor, a vehicle speed sensor, and a vision sensor.

8. The method according to any one of claims 1-6, characterized in that, The first intermediate variable includes at least one of the covariance of the optimization algorithm used to determine the first location information and the statistical value of the first driving information in the first time period.

9. The method according to any one of claims 1-6, characterized in that, The first location information includes the vehicle's first longitude, first latitude, and first vehicle heading angle, and the second location information includes the vehicle's second longitude, second latitude, and second vehicle heading angle.

10. The method according to any one of claims 1-6, characterized in that, The accuracy error includes at least one of lateral distance error, longitudinal distance error, and orientation angle error.

11. A method for training a model, characterized in that, include: Acquire the second driving information of the sample vehicle collected by the sensor in the first moment; Based on the second driving information, the third location information of the sample vehicle is obtained, as well as the second intermediate variable used in determining the third location information; Obtain the fourth location information of the sample vehicle collected by the positioning device at the first time. A training sample set is determined, the training sample set including the third position information, the fourth position information, and the second intermediate variable; The accuracy estimation model is trained based on the training sample set; The second driving information includes driving information of the sample vehicle collected by the Global Navigation Satellite System (GNSS) sensors; The step of obtaining the third location information of the sample vehicle based on the second driving information, and the second intermediate variable used in determining the third location information, includes: Based on the second driving information and map information, the third location information and the second intermediate variable are obtained, wherein the second intermediate variable includes error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information; The step of training the accuracy estimation model based on the training sample set includes: The first accuracy estimation model is trained based on the training sample set.

12. The method according to claim 11, characterized in that, The second driving information includes driving information of the sample vehicle collected by the Global Navigation Satellite System (GNSS) sensors; The method further includes: invalidating a portion of the driving information of the sample vehicle collected by the GNSS sensor; The step of obtaining the third location information of the sample vehicle based on the second driving information, and the second intermediate variable used in determining the third location information, includes: Based on the valid portion of the second driving information and the map information, the third location information and the second intermediate variable are obtained, wherein the second intermediate variable includes error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information; The step of training the accuracy estimation model based on the training sample set includes: The second accuracy estimation model is trained based on the training sample set.

13. The method according to claim 11, characterized in that, The second driving information includes driving information of the sample vehicle collected by the Global Navigation Satellite System (GNSS) sensors; The step of training the accuracy estimation model based on the training sample set includes: The third accuracy estimation model is trained based on the training sample set.

14. A device for estimating positioning accuracy, characterized in that, include: The acquisition unit is used to acquire the vehicle's first driving information collected by the sensors; The processing unit is configured to obtain the first location information of the vehicle based on the first driving information, and a first intermediate variable used in determining the first location information. The determining unit is configured to determine a target accuracy estimation model from a pre-trained accuracy estimation model based on the availability of the first driving information and map information; the accuracy estimation model is obtained by training a machine learning model based on a training sample set, wherein the training samples in the training sample set include the location information of the sample vehicle and intermediate variables used in determining the location information; The target accuracy estimation model is used to take the first position information and the first intermediate variable as input to obtain the second position information of the vehicle, and the accuracy error of the first position information relative to the second position information.

15. An apparatus for training a model, characterized in that, include: The first acquisition unit is used to acquire the second driving information of the sample vehicle collected by the sensor in the first moment; The processing unit is configured to obtain the third location information of the sample vehicle based on the second driving information, and a second intermediate variable used in determining the third location information; The second acquisition unit is used to acquire the fourth location information of the sample vehicle collected by the positioning device at the first time. A determining unit is used to determine a training sample set, the training sample set including the third position information, the fourth position information, and the second intermediate variable; A training unit is used to train the accuracy estimation model based on the training sample set; The second driving information includes driving information of the sample vehicle collected by the Global Navigation Satellite System (GNSS) sensors; Specifically, the determining unit is used to: obtain the third location information and the second intermediate variable based on the second driving information and map information, wherein the second intermediate variable includes error information between the lane information determined based on the second driving information and the lane information of the sample vehicle in the map information; Specifically, the training unit is used to train the first accuracy estimation model based on the training sample set.

16. An electronic device, characterized in that, The method includes a processor and a memory, wherein the memory stores instructions, and when the processor executes the instructions, it causes the processor to perform the method according to any one of claims 1-13.

17. A computer storage medium, characterized in that, Used to store a computer program, the computer program including a method for performing any one of claims 1-13.

18. A computer program product, characterized in that, It includes computer program code, which, when executed by an electronic device, causes the electronic device to perform the method of any one of claims 1-13.

Images