An Unmanned Environment Perception and Precise Positioning System and Method

Abstract

The present invention discloses an unmanned driving environment perception precise positioning system and method, which relates to the field of positioning technology. It includes dividing point clouds into a static layer point cloud set and a dynamic layer point cloud set; performing ground projection on the static layer point cloud set, extracting geometric features of road markings and generating a topological coding vector; synchronously receiving GNSS positioning data, IMU inertial data and the topological coding vector, and generating a confidence matrix according to the signal-to-noise ratio of each data source; performing projection matching between the topological coding vector and the marking coding matrix in the high-precision map, and activating the visual weight enhancement mode of the confidence matrix when the matching similarity is greater than the matching threshold, and outputting the fused positioning coordinates. The present invention matches and verifies the topological coding vector with the marking information in the high-precision map, further improves the positioning accuracy, solves the problem of poor adaptability of traditional maps, and improves the positioning accuracy and robustness of unmanned vehicles in complex environments.

Description

Technical Field

[0001] The present invention relates to the field of positioning technology, and particularly to an accurate positioning system and method for unmanned driving environment perception. Background Art

[0002] With the continuous development of unmanned driving technology, the environmental perception and accurate positioning of autonomous driving systems have become one of the key technologies for the safe and stable driving of unmanned vehicles. In the past few decades, with the progress of sensor technology, technologies such as lidar (LiDAR), vision sensors, and GNSS (Global Navigation Satellite System) have been gradually applied to autonomous vehicles, providing rich data sources for environmental perception. Lidar can effectively perceive the three-dimensional structure of the surrounding environment through its high-precision point cloud data, especially showing good detection capabilities in complex urban environments.

[0003] However, there are still some deficiencies in the existing technologies. For example, although multi-sensor fusion technology can improve the positioning accuracy, most of the existing methods rely on simple weighted fusion strategies and fail to effectively consider the signal-to-noise ratio and reliability of different sensor data in different environments, resulting in the accuracy of sensor data fusion being affected in some complex scenarios (such as high-dynamic scenarios or environmental occlusions). In addition, existing lidar point cloud processing methods often only focus on static environment modeling and lack effective discrimination in the detection and processing of dynamic objects, making it difficult to accurately identify dynamic obstacles and static environmental features. Summary of the Invention

[0004] In view of the problems that the existing unmanned driving environment perception systems usually have unstable positioning accuracy and poor environmental adaptability, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to improve the positioning accuracy and robustness of unmanned vehicles in complex environments.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an unmanned driving environment perception and precise positioning system, which includes a dynamic point cloud layering module that receives the original lidar point cloud data, divides the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputs the static layer point cloud set; a marking feature encoding module that projects the static layer point cloud set onto the ground, extracts the geometric features of the road markings, and generates a topological encoding vector, where the topological encoding vector includes a triple encoding sequence of lane line type, radius of curvature, and solid-to-dashed ratio; a confidence matrix generation module that synchronously receives GNSS positioning data, IMU inertial data, and the topological encoding vector, and generates a confidence matrix based on the signal-to-noise ratio of each data source; and a cross-modal verification module that projects and matches the topological encoding vector with the marking encoding matrix in the high-precision map, and activates the visual weight enhancement mode of the confidence matrix when the matching similarity is greater than the matching threshold, and outputs the fused positioning coordinates.

[0008] As a preferred solution of the unmanned driving environment perception and precise positioning system of the present invention, the dynamic point cloud layering module includes: receiving the original lidar point cloud data, and performing time series segmentation on the original lidar point cloud data; calculating the dynamic changes of the point cloud based on the difference analysis of the point cloud trajectories, and determining whether it belongs to a static scene or a dynamic object; if it belongs to a static scene, performing clustering analysis to identify the fixed structures in the environment, and outputting the distribution and boundary information of each object in the static point cloud; and outputting the static layer point cloud set.

[0009] As a preferred solution of the unmanned driving environment perception and precise positioning system of the present invention, the marking feature encoding module includes: using a ground detection algorithm to identify the ground points in the static point cloud; projecting all the points in the static layer point cloud set onto the ground plane according to their spatial coordinates to obtain the corresponding ground coordinates, and obtaining the ground projection point cloud set; extracting the road markings from the ground projection point cloud set, and performing geometric feature analysis on the extracted road markings; and generating a topological encoding vector based on the extracted geometric features of the road markings, where the topological encoding vector is composed of a triple encoding sequence of three features: lane line type, radius of curvature, and solid-to-dashed ratio.

[0010] As a preferred solution of the unmanned driving environment perception and precise positioning system of the present invention, the geometric feature analysis of the extracted road markings includes: analyzing the shape of the markings to determine the lane line type; calculating the radius of curvature according to the bending degree of the markings; and calculating the solid-to-dashed ratio according to the ratio of the solid and dashed parts of the markings, that is, the length ratio of the continuous solid line to the dashed line in the markings.

[0011] As a preferred embodiment of the unmanned driving environment perception and precise positioning system of the present invention, the confidence matrix generation module includes: a signal-to-noise ratio calculation unit for calculating the signal-to-noise ratio of each data source according to GNSS positioning data, IMU inertial data, and visual data; an initial confidence matrix generation unit for generating an initial confidence matrix according to the calculated signal-to-noise ratio; and an environment correction unit for generating an environment correction matrix based on the traffic flow density factor and the environmental visibility factor, correcting the initial confidence matrix, and obtaining the confidence matrix.

[0012] As a preferred embodiment of the unmanned driving environment perception and precise positioning system of the present invention, the initial confidence matrix is calculated by weighted calculation based on the signal-to-noise ratio and the importance of different data sources; according to the weight values, an initial confidence matrix is generated, and the initial confidence matrix includes the confidence values of each data source.

[0013] As a preferred embodiment of the unmanned driving environment perception and precise positioning system of the present invention, the environment correction unit includes: the traffic flow density factor is the number of vehicles detected by a camera or radar per unit time; the environmental visibility factor is obtained through a light intensity sensor or meteorological data; an environment correction matrix is generated according to a pre-stored environmental impact coefficient table; the initial confidence matrix and the environment correction matrix are subjected to a Hadamard product operation, wherein the non-diagonal elements are updated to the coupling weights between sensors to obtain the corrected confidence matrix; the dynamic layer point cloud density is used as a real-time attenuation factor for visual confidence, and the confidence of visual positioning is adjusted twice to output the final confidence matrix.

[0014] As a preferred embodiment of the unmanned driving environment perception and precise positioning system of the present invention, the cross-modal verification module includes: according to the current position, direction of the vehicle, and the geometric characteristics of the road markings, performing spatial projection alignment on the topological coding vector and the marking coding matrix in the high-precision map, and calculating the similarity; comparing the calculated similarity with a matching threshold, and when the matching similarity is greater than the matching threshold, activating the visual weight strengthening mode of the confidence matrix to strengthen the visual data weight in the confidence matrix; using the strengthened visual weight to perform weighted fusion on the positioning results of each data source in the confidence matrix to obtain the final fused positioning coordinates as the current position of the vehicle.

[0015] Second aspect, the present invention provides an accurate positioning method for unmanned driving environment perception, which includes: receiving the original point cloud data of the lidar, segmenting the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputting the static layer point cloud set; performing ground projection on the static layer point cloud set, extracting the geometric features of the road markings and generating a topological coding vector, where the topological coding vector includes a triple coding sequence of lane line type, radius of curvature, and virtual-real ratio; synchronously receiving GNSS positioning data, IMU inertial data, and the topological coding vector, generating a confidence matrix according to the signal-to-noise ratio of each data source; performing projection matching between the topological coding vector and the marking coding matrix in the high-precision map, and activating the visual weight enhancement mode of the confidence matrix when the matching similarity is greater than the matching threshold, and outputting the fused positioning coordinates.

[0016] Third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program instructions are executed by the processor, the steps of the accurate positioning system for unmanned driving environment perception as described in the first aspect of the present invention are implemented.

[0017] Fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program instructions are executed by the processor, the steps of the accurate positioning system for unmanned driving environment perception as described in the first aspect of the present invention are implemented.

[0018] The beneficial effects of the present invention are as follows: through the separation processing of the static layer and the dynamic layer of the dynamic point cloud, the present invention not only improves the system's modeling ability for the static environment, but also makes the recognition of dynamic obstacles more accurate. In addition, by introducing a method for generating a confidence matrix corrected by the signal-to-noise ratio, the system can adaptively adjust the weights according to the reliability of different sensor data, thereby avoiding overfitting in high-dynamic or complex environments. And through the cross-modal verification module, the topological coding vector is matched and verified with the marking information in the high-precision map, further improving the positioning accuracy, solving the problem of poor adaptability of traditional maps, and enhancing the positioning accuracy and robustness of unmanned vehicles in complex environments. Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings.

[0020] Figure 1 It is a structural diagram of an accurate positioning system for unmanned driving environment perception.

[0021] Figure 2It is a structural diagram of the confidence matrix generation module of the precise positioning system for driverless environment perception.

[0022] Figure 3 It is a schematic diagram of the precise positioning method for driverless environment perception. Specific embodiments

[0023] To make the above objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings of the specification.

[0024] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0025] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0026] As mentioned in the above background art, although multi-sensor fusion technology can improve the positioning accuracy, most of the existing methods rely on simple weighted fusion strategies and fail to effectively consider the signal-to-noise ratio and reliability of different sensor data in different environments, resulting in the accuracy of sensor data fusion being affected in some complex scenarios (such as high-dynamic scenarios or environmental occlusion). In addition, existing lidar point cloud processing methods often only focus on static environment modeling and lack effective discrimination in the detection and processing of dynamic objects, making it difficult to accurately identify dynamic obstacles and static environmental features.

[0027] To address the above problems, the present invention improves the positioning accuracy and robustness of driverless vehicles in complex environments through dynamic point cloud layering, marking feature encoding, and cross-modal verification, as follows.

[0028] Figure 1 It is a system block diagram of the precise positioning system for driverless environment perception according to an embodiment of the present application. As Figure 1As shown in the figure, in the precise positioning system for unmanned driving environment perception, it includes: a dynamic point cloud layering module 100, which receives the original point cloud data of the lidar, divides the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputs the static layer point cloud set; a marking feature encoding module 200, which projects the static layer point cloud set onto the ground, extracts the geometric features of the road markings and generates a topological encoding vector, and the topological encoding vector includes a triple encoding sequence of lane line type, radius of curvature, and virtual-real ratio; a confidence matrix generation module 300, which synchronously receives GNSS positioning data, IMU inertial data, and the topological encoding vector, and generates a confidence matrix according to the signal-to-noise ratio of each data source; a cross-modal verification module 400, which projects and matches the topological encoding vector with the marking encoding matrix in the high-precision map, and activates the visual weight enhancement mode of the confidence matrix when the matching similarity is greater than the matching threshold, and outputs the fused positioning coordinates.

[0029] In the embodiment of the present application, the dynamic point cloud layering module 100 is used to receive the original point cloud data of the lidar, divide the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and output the static layer point cloud set. It should be understood that the differential analysis method based on the point cloud trajectory is one of the key algorithms for environment perception in the unmanned driving system. This method can not only distinguish static and dynamic objects, but also provide stable dynamic object detection in a complex real-time driving environment; the effectiveness and real-time performance of this method directly affect the safety and reliability of the unmanned driving system. Therefore, in the system implementation, the dynamic point cloud layering module often needs to be fused with other sensor data (such as cameras, radars, etc.) to jointly improve the overall perception ability.

[0030] Among them, the original point cloud data of the lidar contains the information of all three-dimensional space points obtained by the lidar scanning in the unmanned driving environment, and each point contains position coordinates and reflection intensity. According to the sensor timestamp information, the original point cloud data is sliced into multiple time segments, and each time segment represents all the point cloud data collected by the lidar scanning within a time window.

[0031] Optionally, in order to improve the accuracy and reliability of the point cloud data, other sensors (such as cameras, radars, etc.) are usually combined to fuse the point cloud data. This can make up for the blind spots or errors of a single sensor (such as lidar).

[0032] Using motion detection algorithms (such as differential analysis based on point cloud trajectories), calculate the dynamic changes of the point cloud. For each point cloud in a time segment, by comparing the motion differences between consecutive time segments, determine whether the point belongs to a static scene or a dynamic object. Among them, the position of dynamic objects changes significantly, reflecting the movement of objects such as vehicles and pedestrians; while the static scene has no obvious position change, reflecting fixed structures such as buildings, trees, and roads in the environment. Based on this judgment, the point cloud data is divided into a static layer point cloud set and a dynamic layer point cloud set.

[0033] Exemplarily, using motion detection algorithms (such as differential analysis based on point cloud trajectories), calculating the dynamic changes of the point cloud includes the following steps:

[0034] Preprocess the original point cloud data of the lidar, including operations such as denoising and filtering. Among them, for the denoising step, methods such as voxel grid filtering (Voxel Grid Filter) can be used to reduce redundant point cloud data to an accuracy suitable for subsequent processing; according to the data acquisition frequency of the lidar and the time stamps of the sensors, the original point cloud data is sliced into multiple time segments (time windows), and each time segment contains all the point cloud data scanned at a certain moment; register the point cloud data in each time segment to ensure that the point clouds of each time segment are aligned in the same coordinate system.

[0035] Preferably, the ICP algorithm or other point cloud registration algorithms can be used for this processing to obtain globally consistent point cloud data.

[0036] Furthermore, after completing the point cloud registration, use a differential analysis algorithm to compare the point cloud data between consecutive time segments. The Euclidean distance between point clouds can be calculated or the position change can be measured using the point-to-point distance between point clouds. For the point cloud in each time segment, by comparing its position change with the point cloud of the previous time segment, determine whether the point cloud has moved significantly. If the position difference between a certain point cloud in consecutive time segments determines whether the point belongs to a dynamic object.

[0037] Based on the differential analysis, judge dynamic objects and static objects according to the set threshold. The point cloud positions of dynamic objects change greatly, reflecting the movement trajectories of objects, such as vehicles or pedestrians in motion; while the point cloud positions of static objects change little, reflecting fixed structures in the environment, such as buildings and roads. Using the dynamic detection results, the point cloud data is divided into a dynamic layer point cloud set and a static layer point cloud set.

[0038] The above method has good adaptability to complex scenarios (such as dynamic traffic, crowded pedestrian areas, etc.).

[0039] Further, perform clustering analysis on the static layer point cloud set to identify fixed structures in the environment.

[0040] Exemplarily, the purpose of clustering analysis is to group the static point cloud by spatial position, identify each static object such as buildings, roads, obstacles, etc., and output the distribution and boundary information of each object in the static point cloud to help the unmanned driving system better understand the static objects in the surrounding environment.

[0041] In the embodiment of the present application, the marking feature encoding module 200 is used to perform ground projection on the static layer point cloud set, extract road marking geometric features and generate a topological coding vector, and the topological coding vector includes a triple coding sequence of lane line type, radius of curvature, and solid-to-dashed ratio, which specifically includes the following content:

[0042] Receive the static layer point cloud set, which contains the static point cloud data output from the dynamic point cloud layering module, representing the immovable parts (such as fixed objects like roads and buildings) in the road environment; use a ground detection algorithm to identify the ground points in the static point cloud; project all points in the static layer point cloud set onto the ground plane according to their spatial coordinates, that is, perform vertical projection, to obtain the points on the ground to eliminate the influence caused by height differences, and obtain the ground projection point cloud set;

[0043] Extract road markings from the ground projection point cloud set, and use a line fitting algorithm (such as the least squares method, Hough transform, etc.) to identify the markings on the road. Analyze the geometric features of the extracted markings, mainly including: lane line type: determine its type by analyzing the shape of the marking (solid line or dashed line); radius of curvature: calculate the radius of curvature R according to the degree of bending of the marking, and the radius of curvature can be obtained by performing quadratic curve fitting on the fitted marking; solid-to-dashed ratio: calculate the solid-to-dashed ratio according to the ratio of the solid and dashed parts of the marking, that is, the length ratio of the continuous solid line to the dashed line in the marking.

[0044] Further, based on the extracted marking geometric features, generate a topological coding vector, which is composed of a triple coding sequence of three features: lane line type, radius of curvature, and solid-to-dashed ratio, and each feature coding can be performed in the following manner:

[0045] The lane line type is represented by a binary value (such as 1 represents a solid line, 0 represents a dashed line) or a classification number; discretize the radius of curvature into values within a finite range and generate a code according to the specific value. For example, the curvature can be divided into different levels (such as large curvature, medium curvature, small curvature, etc.) through a certain threshold; discretize the solid-to-dashed ratio into several levels (such as high ratio, moderate ratio, low ratio), and generate the corresponding code.

[0046] Exemplarily, lane line types are represented using binary values or classification numbers: a solid line can be represented using the binary value 1; a dashed line can be represented using the binary value 0. Classification numbers can also be used to represent different lane line types. For example, a solid line is encoded as 1 and a dashed line is encoded as 2; the radius of curvature can be discretized by a certain threshold and divided into different levels. For example, a large curvature, that is, a relatively large radius of curvature, represents a relatively straight road and can be represented by the code 1; a medium curvature, that is, a medium radius of curvature, represents a road with a certain bend and can be represented by the code 2; a small curvature, that is, a relatively small radius of curvature, represents a road with a sharp bend and can be represented by the code 3. Among them, the discretization interval and encoding of the curvature can be adjusted according to the actual curvature range and requirements. The solid-dashed ratio reflects the ratio of the dashed line to the solid line in the lane line and can be discretized by dividing the ratio into several levels. The divided levels can be: a high ratio, indicating that most of the lane lines are dashed lines and the ratio is high, which can be represented by the code 1; a moderate ratio, indicating that the ratio of the dashed line to the solid line in the lane line is moderate, which can be represented by the code 2; a low ratio, indicating that most of the lane lines are solid lines and the proportion of the dashed line is low, which can be represented by the code 3.

[0047] Suppose there are the following geometric features of the lane line: lane line type: solid line → encoded as 1; radius of curvature: medium curvature → encoded as 2; solid-dashed ratio: moderate ratio → encoded as 2; then the generated triple encoding sequence is: 。

[0048] The above triple encoding sequence provides a structured representation for the geometric features of each lane line, and the specific discretization and encoding methods can be adjusted according to the actual scenario and requirements.

[0049] In this embodiment, identifying ground points in the static point cloud using the ground detection algorithm includes: randomly selecting point pairs from the point cloud and performing plane fitting, and using the RANSAC algorithm to identify and extract the maximum number of ground points. The specific steps are as follows: randomly select a certain number of points from the static point cloud set, assuming that these points belong to the ground plane; calculate the plane equation and count the points with the smallest variance above this plane equation (i.e., the points with the smallest vertical distance from the plane); through multiple iterations, update the plane model and determine which points belong to the ground and which do not according to the threshold.

[0050] Once the ground points are identified, all points in the static point cloud set need to be projected onto the ground plane next. This operation maps the spatial coordinates of each point onto the ground plane through vertical projection to eliminate the influence brought by the height difference. In this way, point cloud data parallel to the ground plane can be obtained, which is convenient for further analysis and processing.

[0051] In this embodiment, projecting all the points in the static layer point cloud set onto the ground plane according to their spatial coordinates to obtain the ground projection point cloud set includes the following operations:

[0052] For each point in the static point cloud Project it onto the plane through its distance from the ground plane (i.e., the perpendicular distance from the point to the plane); assume the equation of the ground plane is , where is the normal vector of the plane, and D is the offset; for each point The new point projected onto the ground plane Can be calculated by the following formula:

[0053]

[0054] Where is the vertical coordinate of the projected point on the ground plane, and Can be projected by keeping and Coordinates unchanged.

[0055] This process eliminates the influence caused by different heights in the static point cloud (such as height differences of buildings or other objects), ensuring that the point cloud data on the ground is more accurate and consistent. After vertical projection, all the obtained point cloud data is the ground projection point cloud set. These point cloud data represent all the points parallel to the ground plane in the static environment, eliminating the influence of height differences and can be used for further analysis and processing.

[0056] It should be noted that in the embodiment of this application, the marking feature encoding module projects the static layer point cloud set onto the ground, extracts the geometric features of the road markings and generates a topological coding vector, effectively improving the accuracy and robustness of the unmanned driving environment perception system. The static layer point cloud set contains the static point cloud data output from the dynamic point cloud layering module, and these data represent the immovable parts in the road environment (such as fixed objects like roads and buildings). The marking feature encoding module first uses the ground detection algorithm to identify the ground points in the static point cloud. Fits the ground plane through the RANSAC algorithm and identifies the maximum number of ground points, thereby eliminating the influence caused by height differences. Subsequently, through vertical projection, all the static point cloud data is projected onto the ground plane to obtain the ground projection point cloud set, and this step provides accurate point cloud data for further analyzing the geometric features of the road markings.

[0057] The method of generating the topological coding vector in this process has significant advantages: First, by discretizing the geometric features of the road markings (such as lane line type, radius of curvature, virtual-to-real ratio, etc.), the complexity in data processing is effectively reduced, and it can be flexibly adjusted for different scenarios. The discretization of lane line type, radius of curvature, and virtual-to-real ratio not only makes the data format adapt to the processing capacity of the system, but also can encode the road markings within a specific threshold range according to actual needs, thus providing a reliable and scalable representation method. Through this structured topological coding vector, the system can efficiently perform road marking recognition and matching.

[0058] In the embodiment of the present application, the confidence matrix generation module 300 is configured to synchronously receive GNSS positioning data, IMU inertial data, and the topological coding vector, and generate a confidence matrix according to the signal-to-noise ratio of each data source, as Figure 2 shown, including a signal-to-noise ratio calculation unit 301, configured to calculate the signal-to-noise ratio of each data source according to GNSS positioning data, IMU inertial data, and visual data; an initial confidence matrix generation unit 302, configured to generate an initial confidence matrix according to the calculated signal-to-noise ratio; and an environment correction unit 303, configured to generate an environment correction matrix based on the traffic flow density factor and the environmental visibility factor, and correct the initial confidence matrix to obtain the confidence matrix.

[0059] In this embodiment, the signal-to-noise ratio calculation unit 301 specifically includes:

[0060] The GNSS positioning data provides the position of the current vehicle in the global coordinate system, including longitude, latitude, and altitude information; the IMU inertial data provides the acceleration and angular velocity data of the vehicle for estimating the motion state of the vehicle; the topological coding vector, including lane line type, radius of curvature, and virtual-to-real ratio, provides the geometric feature information of the road markings.

[0061] Specifically, the calculation of the GNSS signal-to-noise ratio is based on the GNSS positioning accuracy, which is usually expressed as the positioning error. The higher the positioning accuracy, the higher the signal-to-noise ratio. Generally, the calculation of the GNSS signal-to-noise ratio can be performed according to the following steps:

[0062] First, it is necessary to obtain the accuracy of GNSS positioning, which is usually expressed as , the smaller it is, the smaller the positioning error and the higher the accuracy:

[0063]

[0064] where is the error range of GNSS positioning.

[0065] The accuracy of the IMU is generally determined by estimating the error ranges of the accelerometer and gyroscope. For example, the inertial measurement error may be expressed as , which represents the standard deviation or deviation of the IMU output data. The signal-to-noise ratio formula can be expressed as:

[0066]

[0067] Furthermore, the visual signal-to-noise ratio is estimated by analyzing the stability and availability of the coded vectors of the lane line features in the image. The visual signal-to-noise ratio is usually adjusted according to the lane line information (such as lane line features) in the image. Specifically, the visual signal-to-noise ratio can be calculated based on the stability of the lane line information in the image. For example, whether the lane lines are clearly visible and whether they are continuously stable in the image. In practical applications, the signal-to-noise ratio can be adjusted according to the quality of the lane line features. For example, a simple quality factor (such as the clarity index of the lane lines) can be used to adjust the visual signal-to-noise ratio, which can be expressed as:

[0068]

[0069] where is the error range of lane line recognition in the visual image, and is the image quality factor, which is used to represent the stability and visibility of the lane line features in the image.

[0070] The above three signal-to-noise ratios can be dynamically calculated by an algorithm, and the perception and positioning capabilities of the system can be adjusted according to different input data sources and accuracies.

[0071] In this embodiment, the initial confidence matrix generation unit 302 specifically includes:

[0072] The initial confidence matrix is calculated based on the signal-to-noise ratio and the importance weighting of different data sources, as follows:

[0073] For each data source, the weight is determined according to the reciprocal of its signal-to-noise ratio; according to the determined weight values, an initial confidence matrix is generated, which contains the confidence values of each data source, and its calculation method is:

[0074]

[0075] where , , are the weights corresponding to the data sources, that is, the reciprocals of the corresponding signal-to-noise ratios; , , are the values corresponding to the data sources respectively, and the initial values of the non-diagonal elements are 0, indicating that there is no preset correlation between the sensors.

[0076] In this embodiment, by combining the signal-to-noise ratio calculation of multiple sensor data and the confidence matrix generation technology, the environmental perception accuracy and positioning accuracy of the driverless system can be significantly improved. Specifically, the signal-to-noise ratio calculation unit dynamically adjusts the weights of each data source according to the signal quality of GNSS positioning data, IMU inertial data, and topological coding vectors. The calculation of the GNSS signal-to-noise ratio is based on the GNSS positioning error. The higher the positioning accuracy, the higher the signal-to-noise ratio, which can ensure the reliability of GNSS data. The IMU data determines its signal-to-noise ratio by estimating the range of inertial errors, enabling the inertial navigation system to more accurately estimate the vehicle motion state during positioning. The signal-to-noise ratio of the visual signal is evaluated by analyzing the stability and availability of the lane line features in the image, improving the effectiveness of the image data in harsh environments. Therefore, the signal-to-noise ratio calculation unit can dynamically adjust its weight in positioning according to the stability of different sensor data sources.

[0077] In this embodiment, the environmental correction unit 303 specifically includes:

[0078] Performing matrix correction according to the traffic flow density factor and the environmental visibility factor, including:

[0079] The traffic flow density factor adjusts the lane-related part of the confidence matrix according to the change of the real-time traffic flow. When the traffic flow density is high, the lane line confidence is low; the environmental visibility factor adjusts the overall confidence according to the current environmental visibility (such as weather conditions like haze, rain, and snow). The lower the visibility, the lower the confidence should be.

[0080] Setting the traffic flow density factor and the environmental visibility factor and generating an environmental correction matrix , whose value is adjusted according to the above two factors: ; Performing the Hadamard product operation (i.e., element-wise multiplication) on the initial confidence matrix and the environmental correction matrix to obtain the corrected confidence matrix , where the non-diagonal elements are updated to the coupling weights between sensors. For example, the collaborative credibility between GNSS and lidar or the error compensation relationship between vision and inertial navigation, etc.

[0081] Preferably, when the dynamic layer point cloud density exceeds the threshold (e.g., >30%), the dynamic layer point cloud density is superimposed as the visual confidence attenuation factor to further adjust the confidence of visual positioning. For example:

[0082]

[0083] Among them, is the attenuation factor, and the final confidence matrix is output.

[0084] It should be noted that the confidence matrix generation unit determines the importance of each data source in positioning through weighted calculation based on the signal-to-noise ratio of each sensor. Specifically, the generation of the initial confidence matrix assigns weights according to the reciprocal of the signal-to-noise ratio of each sensor, which enables high-precision data sources (such as high-quality GNSS positioning data) to play a greater role in positioning, while low-precision data sources (such as visual data with high noise) are appropriately suppressed.

[0085] The environment correction unit further corrects the confidence matrix and adjusts the confidence of the lane-related part through the traffic flow density factor and the environmental visibility factor. When the traffic flow density is high, the confidence of the road markings will decrease, and in the case of poor visibility, the overall positioning confidence will decrease. This dynamic adjustment mechanism can effectively cope with environmental impacts such as traffic complexity and bad weather, so as to maintain the stability and accuracy of the system. The Hadamard product operation multiplies the initial confidence matrix and the environment correction matrix element by element to further optimize the calculation of the confidence.

[0086] In the embodiment of the present application, the cross-modal verification module 400 is used to perform projection matching between the topological coding vector and the road marking coding matrix in the high-precision map. When the matching similarity is greater than the matching threshold, the visual weight enhancement mode of the confidence matrix is activated, and the fused positioning coordinates are output, including the following content:

[0087] Receive the topological coding vector and the high-precision map road marking coding matrix. Among them, the topological coding vector contains information such as lane line type, curvature radius, and virtual-real ratio; the high-precision map road marking coding matrix contains the geometric features and topological information of the road markings in the high-precision map, such as lane line type, position, curvature, etc. Perform projection matching between the topological coding vector and the road marking coding matrix in the high-precision map, and perform spatial projection alignment on the two according to the current position, direction of the vehicle, and the geometric features of the road markings. During the projection matching process, considering the similarity of topological information, matching measurement methods such as cosine similarity and Euclidean distance are used to calculate the similarity between the two.

[0088] Compare the calculated similarity with the matching threshold. If the similarity is greater than the matching threshold, it is considered that the topological coding vector matches the high-precision map road marking information successfully, and the visual weight enhancement mode of the visual confidence matrix is activated. When the matching is successful, activate the visual weight enhancement mode in the confidence matrix, and strengthen the visual data weight in the confidence matrix according to the preset strengthening coefficient:

[0089]

[0090] Among them, is the original visual confidence; is the enhanced visual confidence, is the visual enhancement coefficient. The enhanced visual confidence improves the importance of visual information in fusion positioning and enhances the visual positioning accuracy.

[0091] Using the enhanced visual weights, the positioning results of each data source are weighted and fused in the confidence matrix. Through methods such as weighted average and Kalman filtering, the final fused positioning coordinates are obtained as the current position of the vehicle.

[0092] In the embodiment of the present application, the cross-modal verification module significantly improves the positioning accuracy of the unmanned driving system in complex environments by performing projection matching between the topological coding vector and the marking coding matrix in the high-precision map. The topological coding vector contains key information such as lane line type, radius of curvature, and virtual-real ratio, while the marking coding matrix in the high-precision map contains geometric features and topological information of road markings, such as the position, shape, and curvature of lane lines. By precisely aligning the spatial projection of the topological coding vector and the high-precision map marking coding matrix, the system can consider the similarity of topological information such as lane line type and position, and use methods such as cosine similarity and Euclidean distance to quantify the matching degree between the two. The calculation of the matching similarity can compare the road marking features in the high-precision map according to the current position and direction of the vehicle, so as to ensure a high degree of consistency between the topological coding vector and the marking information in the high-precision map. When the calculated similarity is greater than the preset matching threshold, the cross-modal verification module will automatically activate the visual weight enhancement mode in the confidence matrix. This process enhances the importance of visual information in the overall positioning by strengthening the weight of visual data. The enhanced visual confidence further enhances the role of visual data in multi-sensor fusion. Especially when the high-precision map matching is successful, the data of the visual sensor can better support the fusion positioning calculation. Through this mechanism, the system can improve the signal-to-noise ratio of visual signals according to the matching result of the high-precision map and the topological coding vector, thereby effectively improving the visual positioning accuracy.

[0093] Figure 3 is a flowchart of the unmanned driving environment perception and precise positioning method according to the embodiment of the present application. As Figure 3 shown, in the unmanned driving environment perception and precise positioning method, it includes:

[0094] Receiving the original point cloud data of the lidar, segmenting the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputting the static layer point cloud set;

[0095] Performing ground projection on the static layer point cloud set, extracting the geometric features of road markings and generating a topological coding vector, where the topological coding vector contains a triple coding sequence of lane line type, radius of curvature, and virtual-real ratio;

[0096] Synchronously receive GNSS positioning data, IMU inertial data, and the topological coding vector, and generate a confidence matrix according to the signal-to-noise ratio of each data source;

[0097] Perform projection matching between the topological coding vector and the marking coding matrix in the high-precision map. When the matching similarity is greater than the matching threshold, activate the visual weight enhancement mode of the confidence matrix and output the fused positioning coordinates.

[0098] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned precise positioning method for unmanned driving environment perception have been described in detail in the description of the precise positioning system for unmanned driving environment perception above with reference to Figures 1 to 2 and therefore, the repeated description thereof will be omitted.

[0099] This embodiment also provides a computer device applicable to the case of the precise positioning system for unmanned driving environment perception, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the precise positioning system for unmanned driving environment perception proposed in the above embodiment.

[0100] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0101] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the precise positioning system for unmanned driving environment perception proposed in the above embodiment.

[0102] In summary, through the separation process of the static layer and the dynamic layer of the dynamic point cloud, the present invention not only improves the system's modeling ability for the static environment, but also enables more accurate recognition of dynamic obstacles. In addition, by introducing a method for generating a confidence matrix corrected by signal-to-noise ratio, the system can adaptively adjust weights according to the reliability of different sensor data, thereby avoiding overfitting in high-dynamic or complex environments. And through the cross-modal verification module, the topological coding vector is matched and verified with the marking information in the high-precision map, further improving the positioning accuracy, solving the problem of poor adaptability of traditional maps, and enhancing the positioning accuracy and robustness of driverless vehicles in complex environments.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An unmanned driving environment perception and precise positioning system, characterized by: include: The dynamic point cloud stratification module receives the original point cloud data of the laser radar, divides the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputs the static layer point cloud set; A road marking feature coding module is used to perform ground projection on the static layer point cloud set, extract road marking geometric features and generate a topological coding vector, wherein the topological coding vector includes a triple coding sequence of lane line type, curvature radius and virtual-real ratio; A confidence matrix generation module synchronously receives GNSS positioning data, IMU inertial data and the topological coding vector, and generates a confidence matrix according to the signal-to-noise ratio of each data source; The cross-modal verification module performs projection matching on the topological coding vector and the marking coding matrix in the high-precision map, activates the visual weight enhancement mode of the confidence matrix when the matching similarity is greater than the matching threshold, and outputs the fused positioning coordinates; The cross-modal verification module includes: according to the current position, direction and geometric features of the vehicle's road markings, spatially aligning the topological coding vector with the road marking coding matrix in the high-precision map, and calculating the similarity; comparing the calculated similarity with the matching threshold, and when the matching similarity is greater than the matching threshold, activating the visual weight enhancement mode of the confidence matrix to enhance the visual data weight in the confidence matrix; using the enhanced visual weight, weightedly fusing the positioning results of each data source in the confidence matrix to obtain the final fused positioning coordinates as the current position of the vehicle.

2. The unmanned driving environment perception and precise positioning system as claimed in claim 1, characterized in that: The dynamic point cloud layering module includes: Receiving raw laser radar point cloud data, and performing time series segmentation on the raw laser radar point cloud data; Based on the difference analysis of point cloud trajectories, the dynamic changes of point clouds are calculated to determine whether it belongs to a static scene or a dynamic object. If it belongs to a static scene, cluster analysis is performed to identify the fixed structure in the environment and output the distribution and boundary information of each object in the static point cloud. Output static layer point cloud set.

3. The unmanned driving environment perception and precise positioning system as claimed in claim 2, characterized in that: The marking feature encoding module includes: Use ground detection algorithms to identify ground points in static point clouds; Projecting all points in the static layer point cloud set onto the ground plane according to their spatial coordinates to obtain corresponding ground coordinates, thereby obtaining a ground projection point cloud set; Extracting road markings from the ground projection point cloud set, and performing geometric feature analysis on the extracted road markings; A topological coding vector is generated based on the extracted geometric features of the road markings. The topological coding vector is composed of a triplet coding sequence of three features: lane line type, curvature radius, and virtual-real ratio.

4. The unmanned driving environment perception and precise positioning system as claimed in claim 3, characterized in that: The geometric feature analysis of the extracted road markings comprises: Analyze the shape of the markings to determine the lane line type; Calculate the radius of curvature according to the curvature of the marking line; According to the ratio of the solid and dotted parts of the marking line, the real-dotted ratio is calculated, that is, the ratio of the length of the continuous solid line to the dotted line in the marking line.

5. The unmanned driving environment perception and precise positioning system as claimed in claim 4, characterized in that: The confidence matrix generation module includes: A signal-to-noise ratio calculation unit, used to calculate the signal-to-noise ratio of each data source based on GNSS positioning data, IMU inertial data and visual data; An initial confidence matrix generating unit, used for generating an initial confidence matrix according to the calculated signal-to-noise ratio; The environment correction unit is used to generate an environment correction matrix based on the vehicle flow density factor and the environment visibility factor, and correct the initial confidence matrix to obtain the confidence matrix.

6. The unmanned driving environment perception and precise positioning system as claimed in claim 5, characterized in that: The initial confidence matrix is ​​weightedly calculated based on the signal-to-noise ratio and the importance of different data sources; An initial confidence matrix is ​​generated according to the weight values, wherein the initial confidence matrix includes the confidence value of each data source.

7. The unmanned driving environment perception and precise positioning system as claimed in claim 6, characterized in that: The environment correction unit comprises: The vehicle flow density factor is the number of vehicles detected by a camera or radar per unit time; The environmental visibility factor is obtained by a light intensity sensor or meteorological data; Generate an environmental correction matrix according to a pre-stored environmental impact coefficient table; Performing a Hadamard product operation on the initial confidence matrix and the environmental correction matrix, wherein the non-diagonal elements are updated to the inter-sensor coupling weights, to obtain a corrected confidence matrix; The density of the dynamic layer point cloud is used as a real-time attenuation factor of the visual confidence, the confidence of the visual positioning is adjusted secondary, and the final confidence matrix is ​​output.

8. An unmanned driving environment perception and precise positioning method, based on the unmanned driving environment perception and precise positioning system according to any one of claims 1 to 7, characterized in that: Also includes: Receiving original point cloud data from a laser radar, dividing the point cloud into a static layer point cloud set and a dynamic layer point cloud set, and outputting the static layer point cloud set; Performing ground projection on the static layer point cloud set, extracting geometric features of road markings and generating a topological coding vector, wherein the topological coding vector includes a triple coding sequence of lane line type, curvature radius and virtual-real ratio; synchronously receiving GNSS positioning data, IMU inertial data and the topological coding vector, and generating a confidence matrix according to the signal-to-noise ratio of each data source; Projection matching is performed on the topological coding vector and the marking coding matrix in the high-precision map, and when the matching similarity is greater than the matching threshold, the visual weight enhancement mode of the confidence matrix is ​​activated, and the fused positioning coordinates are output; The cross-modal verification module includes: performing spatial projection alignment on the topological coding vector and the road marking coding matrix in the high-precision map according to the current position and direction of the vehicle and the geometric features of the road markings, and calculating the similarity; The calculated similarity is compared with the matching threshold. When the matching similarity is greater than the matching threshold, the visual weight enhancement mode of the confidence matrix is ​​activated to enhance the visual data weight in the confidence matrix. The enhanced visual weight is used to perform weighted fusion on the positioning results of each data source in the confidence matrix to obtain the final fused positioning coordinates as the current position of the vehicle.

Images