Map construction method, device, storage medium and processor

Abstract

The present invention discloses a map construction method, device, storage medium and processor. Among them, the method includes: obtaining a first map of a preset road section, where the first map is drawn based on an observation coordinate system; querying a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determining the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree of the first map and the second map; converting the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; constructing a map according to the third map and the confidence level. The present invention solves the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

Description

Technical Field

[0001] The present invention relates to the field of maps, and in particular, to a map construction method, device, storage medium, and processor. Background Art

[0002] A high-precision map, namely a high-definition electronic map (HD Map), is different from a traditional navigation map. In addition to providing road-level navigation information, a high-precision map can also provide lane-level navigation information. In terms of both the richness and accuracy of information, it is far higher than a traditional navigation map.

[0003] There are mainly two ways to establish a high-precision map: professional collection establishment and crowdsourcing collection establishment. Professional collection establishment refers to collecting road information through professional collection equipment and vehicles, and establishing a high-precision map through subsequent data fusion, data processing, publishing, delivery, and other links. Crowdsourcing collection establishment refers to collecting road information through low-cost equipment installed on ordinary user vehicles, and establishing a high-precision map through subsequent data fusion, data processing, publishing, delivery, and other links. The main difference between the two methods lies in the equipment used. Professional collection uses expensive equipment and professional carriers, and the advantage is that high-precision road information can be obtained, with mature technology and strong scene adaptability. The disadvantage is that due to the high cost of collection equipment, large-scale map establishment is limited. The crowdsourcing collection equipment is inexpensive and can be deployed on general commercial vehicles, passenger vehicles, and other carriers. When these carriers are driving on the road surface, a large amount of road data can be collected and accumulated. The advantage is low cost, which is conducive to realizing large-scale low-cost map establishment. The disadvantage is that the data accuracy is low, and the corresponding algorithm processing technology threshold is high, and it is impossible to fuse low-precision multi-trip road information data into a high-precision map.

[0004] In view of the above problem of being unable to establish a high-precision map based on road observation data, no effective solution has been proposed yet. Summary of the Invention

[0005] Embodiments of the present invention provide a map construction method, device, storage medium, and processor to at least solve the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

[0006] According to one aspect of an embodiment of the present invention, a method for constructing a map is provided, including: obtaining a first map of a preset road section, where the first map is drawn based on an observation coordinate system; querying a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determining a registration relationship and a confidence level between the first map and the second map, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; converting the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; constructing a map according to the third map and the confidence level.

[0007] Optionally, obtaining a first map of a preset road section includes: obtaining a plurality of observation points collected by a vehicle on the preset road section; determining a vehicle observation pose of each observation point, where the vehicle observation pose is determined based on the observation coordinate system; collecting vehicle observation data of each observation point, where the vehicle observation data is determined based on the vehicle observation pose; determining the first map of the preset road section based on the vehicle observation data of the plurality of observation points.

[0008] Optionally, querying a second map corresponding to the preset road section includes: determining position information of the preset road section; querying, in a preset map library, a preset area map corresponding to the position information and an adjacent area map, where the preset area map includes the preset road section, and the adjacent area map includes adjacent road sections of the preset road section; in a case where the preset area map exists in the preset map library, determining the preset area map as the second map; in a case where the preset area map does not exist in the preset map library, determining the adjacent area map as the second map.

[0009] Optionally, in a case where the preset area map is determined as the second map, determining a registration relationship and a confidence level between the first map and the second map includes: determining a matching relationship between an observation point in the first map and a preset collection point in the second map; according to the matching relationship, determining a mapping relationship between the vehicle observation pose and a vehicle relative pose, where the vehicle relative pose is determined based on the world coordinate system; according to the mapping relationship, determining road feature data of the vehicle observation data of each observation point relative to the world coordinate system, where the road feature data is used to construct the third map; determining a matched number and an unmatched number of the observation point according to the matching relationship; determining the confidence level according to the matched number and the unmatched number.

[0010] Optionally, constructing a map based on the third map and the confidence includes: obtaining multiple pieces of road feature data of the same observation point in the multiple third maps, and the confidence corresponding to each piece of road feature data; using the confidence as a weight, superimposing the multiple pieces of road feature data of the observation point to obtain the road fusion feature of the observation point; and splicing the road fusion features of multiple observation points.

[0011] Optionally, when determining that the adjacent area map is the second map, determining the registration relationship and confidence between the first map and the second map includes: determining a first anchor point of the first map among multiple observation points in the preset road section; identifying a second anchor point in the second map that matches the first anchor point, where the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; determining the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, where the vehicle relative pose is determined based on the world coordinate system; determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, where the road feature data is used to construct the third map; and determining that the confidence is a preset value.

[0012] Optionally, constructing a map based on the third map and the confidence includes: splicing the road feature data of multiple observation points in the third map according to the confidence.

[0013] Optionally, determining the first anchor point of the first map includes: obtaining a preset feature dictionary corresponding to a target observation point in the preset road section, where the preset feature dictionary records the features of multiple observation points in the preset road section through feature descriptors; determining the mean value of the feature vectors of the target observation point and each observation point in the preset feature dictionary based on the feature descriptors; determining whether the mean value of the feature vectors is greater than a preset threshold, where the preset threshold is determined based on the feature descriptors of multiple observation points recorded in the preset feature dictionary; and determining the target observation point as the first anchor point when the mean value of the feature vectors is greater than the preset threshold.

[0014] Optionally, identifying a second anchor point in the second map that matches the first anchor point includes: obtaining the feature descriptor of a second set anchor point in the second map; determining the feature vector value between the feature descriptor of the first anchor point and the feature descriptor of the second anchor point; and determining that the first anchor point matches the second anchor point when the feature vector value is not greater than the threshold, where when the first anchor point matches the second anchor point, determining the mapping relationship between the vehicle relative pose of the second anchor point and the vehicle observation pose of the first anchor point.

[0015] According to another aspect of the embodiments of the present invention, there is also provided a map construction device, including: an acquisition unit, configured to acquire a first map of a preset road section, where the first map is drawn based on an observation coordinate system; a query unit, configured to query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; a determination unit, configured to determine a registration relationship and a confidence level between the first map and the second map, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; a conversion unit, configured to convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a construction unit, configured to construct a map according to the third map and the confidence level.

[0016] According to another aspect of the embodiments of the present invention, there is also provided a map construction system, including: at least one vehicle, configured to draw a first map of a preset road section based on an observation coordinate system; query a second map of the preset road section drawn based on a world coordinate system in a cloud server; determine a registration relationship and a confidence level between the first map and the second map, and convert the first map into a third map drawn based on the world coordinate system according to the registration relationship; upload the third map and the confidence level to the cloud server; the cloud server is further configured to construct a map according to the registration relationship and the confidence level of at least one of the third maps.

[0017] According to another aspect of the embodiments of the present invention, there is also provided a computer-readable storage medium, where the computer-readable storage medium includes a stored program, and when the program runs, it controls a device where the computer-readable storage medium is located to execute the above-mentioned map construction method.

[0018] According to another aspect of the embodiments of the present invention, there is also provided a processor, where the processor is used to run a program, and when the program runs, it executes the above-mentioned map construction method.

[0019] In the embodiments of the present invention, a first map of a preset road section is acquired, where the first map is drawn based on an observation coordinate system; a second map corresponding to the preset road section is queried, where the second map is drawn based on a world coordinate system; a registration relationship and a confidence level between the first map and the second map are determined, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; the first map is converted into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a map is constructed according to the third map and the confidence level, thereby achieving the technical effect of generating a high-precision map based on low-precision road observation data, and further solving the technical problem of being unable to establish a high-precision map based on low-precision road observation data. Description of the Drawings

[0020] The drawings described herein are provided to further understand the present invention and form a part of this application. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0021] Figure 1 is a flowchart of a method for constructing a map according to an embodiment of the present invention;

[0022] Figure 2 is a schematic diagram of a method for establishing a high-precision map crowdsourcing based on positioning according to an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of a method for establishing a crowdsourcing map according to an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of a method for establishing an incremental map according to an embodiment of the present invention;

[0025] Figure 5 is a schematic diagram of a method for establishing a backbone map according to an embodiment of the present invention;

[0026] Figure 6 is a schematic diagram of a map construction device according to an embodiment of the present invention;

[0027] Figure 7 is a schematic diagram of a map construction system according to an embodiment of the present invention. Detailed Embodiments

[0028] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0029] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0030] According to an embodiment of the present invention, an embodiment of a map construction method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0031] Figure 1 is a flowchart of a map construction method according to an embodiment of the present invention, as Figure 1 shown, the method includes the following steps:

[0032] Step S102, obtain a first map of a preset road section, where the first map is drawn based on an observation coordinate system;

[0033] Step S104, query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system;

[0034] Step S106, determine the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree of the first map and the second map;

[0035] Step S108, convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system;

[0036] Step S110, construct a map according to the third map and the confidence level.

[0037] In an embodiment of the present invention, a first map of a preset road section is obtained, where the first map is drawn based on an observation coordinate system; a second map corresponding to the preset road section is queried, where the second map is drawn based on a world coordinate system; a registration relationship and a confidence level between the first map and the second map are determined, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; the first map is converted into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a map is constructed according to the third map and the confidence level, thereby achieving the technical effect of generating a high-precision map based on low-precision road observation data, and further solving the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

[0038] In the above step S102, the first map may be a map drawn based on the observation coordinate system of the vehicle at the vehicle end, and the first map is drawn based on the observation coordinate system of the vehicle end.

[0039] As an optional embodiment, obtaining the first map of the preset road section includes: obtaining a plurality of observation points collected by the vehicle on the preset road section; determining the vehicle observation pose of each observation point, where the vehicle observation pose is determined based on the observation coordinate system; collecting the vehicle observation data of each observation point, where the vehicle observation data is determined based on the vehicle observation pose; and determining the first map of the preset road section based on the vehicle observation data of the plurality of observation points.

[0040] In the above embodiment of the present invention, the preset road section includes a plurality of observation points. According to the vehicle observation pose of the observation points, the vehicle observation data of the observation points is collected according to the observation coordinate system of the vehicle, and the first map based on the observation coordinate system is drawn according to the vehicle observation data of the plurality of observation points, thereby realizing the drawing of the first map.

[0041] Optionally, the vehicle observation data represents the characteristics of the observation points collected by the vehicle according to the vehicle observation pose.

[0042] Optionally, the vehicle observation data can be collected based on the sensors of the vehicle.

[0043] It should be noted that the road measurement data obtained by the sensors is all based on the sensor coordinate system. Usually, the sensors are fixed on the carrier and move as a rigid body with the carrier. Therefore, there is a fixed conversion relationship TBS, that is, the extrinsic parameter, between the sensor coordinate system and the carrier coordinate system (i.e., the observation coordinate system).

[0044] Optionally, the carrier coordinate system (i.e., the observation coordinate system) can also be referred to as the vehicle body coordinate system for the vehicle, and it is fixed at a certain fixed position of the carrier, such as the center of the rear axle of the vehicle.

[0045] Optionally, the vehicle pose, i.e., the 6Dof (Degree of Freedom) pose WPB = TWB of the vehicle body coordinate system (i.e., the observation coordinate system) in the world coordinate system, where the 6 degrees of freedom include x, y, z, roll (rotation around the Z axis), pitch (rotation around the x axis), and yaw (rotation around the y axis).

[0046] In the above step S104, the second map is drawn based on the world coordinate system, and the world coordinate system maintains a fixed relationship with the actual geographical location. For example, the Earth-Centered, Earth-Fixed (ECEF) coordinate system can be used.

[0047] In the above step S104, the preset map library can be set in the cloud server.

[0048] As an optional embodiment, querying the second map corresponding to the preset road section includes: determining the location information of the preset road section; querying in the preset map library for the preset area map corresponding to the location information and the adjacent area map, where the preset area map includes the preset road section and the adjacent area map includes the adjacent road sections of the preset road section; when the preset area map exists in the preset map library, determining the preset area map as the second map; when the preset area map does not exist in the preset map library, determining the adjacent area map as the second map.

[0049] In the above embodiments of the present invention, the preset map library stores multiple preset maps drawn based on the world coordinate system. Among them, the preset map can include the preset area map of the preset road section or the adjacent area map of the adjacent road sections of the preset road section; thus, it is possible to query in the preset map library based on the location information of the preset road section. If the preset area map of the preset road section exists in the preset map library, then use this preset area map as the second map, and thus the accuracy of the second map can be improved based on the first map; if the preset area map of the preset road section does not exist in the preset map library, it means that the map of the preset road section is missing in the preset map library. Therefore, it is necessary to determine the adjacent road sections of the preset road section, obtain the adjacent area map of the adjacent road sections, and splice the adjacent area map with the first map to complete the map of the missing preset road section in the preset map library.

[0050] Figure 2 is a schematic diagram of a method for establishing a high-precision map crowdsourcing based on positioning according to an embodiment of the present invention, as Figure 2 shown, the cloud stores the established crowdsourcing map (i.e., the second map), and according to the request of the vehicle, sends down the crowdsourcing map of a specific location (i.e., the preset road section) and a specific size for the vehicle to use; and the cloud also continuously receives multiple trips of data (such as the third map) uploaded by the vehicle and performs fusion to establish the crowdsourcing map in an incremental manner.

[0051] It should be noted that when there is no preset area map at the location of the preset road section, the first map of the preset road section is the backbone map; when there is a preset area map at the location of the preset road section, the first map of the preset road section is the incremental map.

[0052] Figure 3 It is a schematic diagram of establishing a crowdsourced map according to an embodiment of the present invention. As Figure 3 shown, the complete form of the crowdsourced map is a vector map containing various structural information of the road. As Figure 3 shown in part (a), during the establishment of the crowdsourced map, it is necessary to first obtain the vehicle trajectory through INS and establish a backbone map based on the vehicle's observation points. As Figure 3 shown in part (b), the backbone map is a map established from the initial single-trip acquisition data. This map provides the backbone structure of the road network, but due to being based on single-trip acquisition data, its accuracy and integrity are not high. As Figure 3 shown in part (c), an incremental map is generated through multi-trip acquisition data, and a complete crowdsourced map can be obtained through the fusion of the incremental map.

[0053] Optionally, the vehicle is equipped with low-cost sensors, mainly including GNSS (Global Navigation Satellite System), IMU (Inertial Measurement Unit), wheel speedometer or vehicle speedometer, camera or LiDAR (Laser Detection and Ranging), or other sensors and their combinations.

[0054] Optionally, the vehicle obtains low-precision absolute position information through GNSS to determine the low-precision position of the vehicle, and based on this, determines whether there is a crowdsourced map at the current position and loads the map tiles of the corresponding crowdsourced map.

[0055] Optionally, the vehicle can use GNSS, IMU, and wheel speed / vehicle speed information as inputs to obtain the vehicle trajectory through the INS algorithm; obtain road observation data through sensors such as cameras or LiDAR, establish a backbone map, and upload it to the cloud.

[0056] Optionally, for the observation points with a backbone map, the vehicle obtains the pose of the vehicle relative to the map through relative positioning, obtains road observation data through sensors such as cameras or LiDAR, establishes a single-trip map, and uploads it to the cloud.

[0057] Optionally, when it is determined that the preset regional map is the second map, it indicates that the preset regional map of the area where the preset road section is located is stored in the preset map library. Then, the first map of the preset road section can be used as the incremental map of the preset regional map to supplement the accuracy of the preset regional map. Furthermore, when using the first map to supplement the accuracy of the preset regional map, it is necessary to register the first map and the preset regional map.

[0058] It should be noted that registration refers to the matching of the geographical coordinates of different image graphics obtained by different imaging means within the same area; in this application, it refers to the process of aligning the features in the first map with the features in the second map, and the method can be performed by an optimization method, such as ICP, NDT, etc.

[0059] As an optional embodiment, when it is determined that the preset regional map is the second map, determining the registration relationship and confidence level between the first map and the second map includes: determining the matching relationship between the observation points in the first map and the preset acquisition points in the second map; according to the matching relationship, determining the mapping relationship between the vehicle observation pose and the vehicle relative pose, where the vehicle relative pose is determined based on the world coordinate system; according to the mapping relationship, determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system, where the road feature data is used to construct the third map; determining the number of matched and unmatched observation points according to the matching relationship; and determining the confidence level according to the number of matched and unmatched observation points.

[0060] Optionally, multiple observation points are used as map features in the first map, and multiple preset acquisition points are used as map features in the second map. Since both the first map and the second map are maps of the preset road section, the observation points in the first map and the preset acquisition points in the second map can represent the same feature of the preset road section. Therefore, the observation points in the first map are matched with the preset acquisition points in the second map to determine the observation points and acquisition points representing the same feature.

[0061] Optionally, the vehicle observation data of the observation points in the first map is determined according to the vehicle observation pose in the observation coordinate system, and the road feature data of the preset acquisition points in the second map is determined according to the vehicle relative pose in the world coordinate system. When the observation points and the preset acquisition points are used to represent the same feature of the preset road, the mapping relationship between the vehicle observation pose and the vehicle relative pose can be determined, and then the vehicle observation data is converted into road feature data based on this mapping relationship.

[0062] For example, the vehicle observation pose and vehicle observation data of the observation points in the first map are known, and the relative vehicle pose and road feature data of the preset collection points in the first map are known. Then, when the observation points match the preset collection points, it is equivalent to determining the vehicle observation pose and vehicle observation data of the same observation point, as well as the relative vehicle pose and road feature data, so that the mapping relationship between the vehicle observation pose and the relative vehicle pose can be determined, and the road feature data of the vehicle observation data of each observation point relative to the world coordinate system can also be determined according to the mapping relationship.

[0063] Optionally, the first map can be drawn based on the vehicle observation data of multiple observation points within a preset section. After converting the vehicle observation data of the first map into road feature data, a third map can be drawn based on the converted road feature data, so as to realize the conversion of the first map drawn based on the observation coordinate system into the third map drawn based on the world coordinate system.

[0064] Optionally, during the process of matching the observation points of the first map and the preset collection points of the second map, some observation points cannot be matched with the preset collection points of the second map. Then, the number of matched observation points that have been matched and the number of unmatched observation points that have not been matched are respectively counted, and then the confidence level is determined according to the number of matched and unmatched points.

[0065] For example, the first map includes n observation points, where m observation points can be registered with the preset collection points of the second map, and p observation points cannot be registered with the preset collection points of the second map. Then, there is a relationship n = m + p, and the confidence level C = m / n, where C is a value between 0 and 1.

[0066] As an optional embodiment, constructing a map according to the third map and the confidence level includes: obtaining multiple road feature data of the same observation point in multiple third maps, and the confidence level corresponding to each road feature data; using the confidence level as a weight, superimposing the multiple road feature data of the observation point to obtain the road fusion feature of the observation point; splicing the road fusion features of multiple observation points.

[0067] Optionally, the multiple road feature data can be the road feature data collected when the same vehicle passes through a preset section multiple times, or the road feature data collected by different vehicles in the preset section respectively.

[0068] Optionally, if the relative vehicle pose is TWBj corresponding to the vehicle observation data PBj, then when projected into the world coordinate system, it is expressed as the road feature data PWj = TWBj * PBj. At this time, the road feature data PWj is already in the world coordinate system.

[0069] Optionally, the confidence level is set as Cj, where j represents multiple observation points at the same position, and j = 1, 2, 3... n.

[0070] Optionally, the observation points at all the same positions are accumulated with the corresponding confidence levels, that is, the road fusion feature Pmean - weight = (PW1 * C1 + PW2 * C2 + PW3 * C3 + …… + PWn * Cn) / (C1 + C2 + C3 + …… + Cn), and its corresponding confidence level is Cmean = (C1 + C2 + C3 + …… + Cn) / n.

[0071] Figure 4 It is a schematic diagram of establishing an incremental map according to an embodiment of the present invention. As Figure 4 shown, registration positioning is performed based on the crowdsourcing map (i.e., the second map) and the first map to determine the confidence level and the relative pose of the vehicle; then, according to the confidence level and the relative pose of the vehicle, combined with the vehicle observation data, a single - trip map (i.e., the third map) of the incremental map is obtained. The specific steps are as follows:

[0072] (1) The 3Dof position of the vehicle body in the world coordinate system is obtained through the vehicle's sensor GNSS (i.e., the position information is obtained by determining the observation points). Query the crowdsourcing map (i.e., the second map) to see if there is a preset area map (such as the backbone map) at this position.

[0073] (2) If there is a preset area map (such as the backbone map) at the current position, then the first map within a certain range around the current position is obtained (i.e., multiple first maps or the third map converted from the first map are obtained).

[0074] Optionally, after obtaining the first map, the first map and the second map can be registered to obtain the relative pose of the vehicle at each observation point in the first map.

[0075] Optionally, the registration process can adopt ICP, NDT, etc.

[0076] (3) After obtaining the relative pose of the vehicle through registration in the previous step, calculate the confidence level of this registration.

[0077] Optionally, the confidence level is calculated based on the registration degree between the observation points in the first map and the preset collection points in the second map.

[0078] For example, the first map includes n observation points, among which m observation points can be registered with the preset collection points in the second map, and p observation points cannot be registered with the preset collection points in the second map. Then, there is a relationship n = m + p, and the confidence level C = m / n, where C is a value between 0 and 1.

[0079] (4) On a road with a preset area map (such as a backbone map), the relative vehicle pose, confidence level, and corresponding vehicle observation data at each position of the vehicle can be obtained. The vehicle observation data is projected into the world coordinate system through the relative vehicle pose to obtain road feature data, and the road feature data of multiple observation points is accumulated with the corresponding confidence level as the weight, that is, a single-trip vehicle trajectory (i.e., the third map) is obtained.

[0080] Optionally, the specific steps for weight accumulation are as follows: If the relative vehicle pose TWBj corresponds to the vehicle observation data PBj, then the projection into the world coordinate system is expressed as the road feature data PWj = TWBj * PBj. At this time, the road feature data PWj is already in the world coordinate system.

[0081] Optionally, the confidence level is set as Cj, where j represents multiple observation points at the same position, and j = 1, 2, 3... n.

[0082] Optionally, all the observation points at the same position are accumulated with the corresponding confidence level, that is, the road fusion feature Pmean - weight = (PW1 * C1 + PW2 * C2 + PW3 * C3 +... + PWn * Cn) / (C1 + C2 + C3 +... + Cn), and its corresponding confidence level is Cmean = (C1 + C2 + C3 +... + Cn) / n.

[0083] (5) The single-trip vehicle trajectory (i.e., the third map) is uploaded to the cloud for fusion processing.

[0084] Optionally, during the above process, incremental mapping is performed based on positioning; while the incremental mapping is completed, vehicle positioning information can be provided simultaneously.

[0085] Optionally, in the case where the adjacent area map is determined to be the second map, if it is shown that there is no preset area map of the area where the preset road section is located in the preset map library, then the first map of the preset road section can be used as the preset area map and stitched with the adjacent area map. Furthermore, when stitching the preset area map and the adjacent area map, it is necessary to register the first map and the adjacent area map.

[0086] As an alternative embodiment, when determining that the adjacent area map is the second map, determining the registration relationship and confidence level between the first map and the second map includes: determining a first anchor point of the first map among multiple observation points on a preset road section; identifying a second anchor point in the second map that matches the first anchor point, where the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; determining the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, where the vehicle relative pose is determined based on the world coordinate system; determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, where the road feature data is used to construct a third map; determining that the confidence level is a preset value.

[0087] Optionally, the first anchor point and the second anchor point respectively describe the same object in the first map and the second map, and thus the first image and the second image can be registered with the first anchor point and the second anchor point.

[0088] Optionally, since the first anchor point belongs to one of the multiple observation points, the vehicle observation pose and vehicle observation data of the first anchor point are known; since the second anchor point is a known feature point in the second image based on the world coordinate system, the vehicle relative pose and road feature data of the second anchor point are known. Thus, when the first anchor point and the second anchor point match, it is equivalent to determining the vehicle observation pose and vehicle observation data of the same observation point, as well as the vehicle relative pose and road feature data, so that the mapping relationship between the vehicle observation pose and the vehicle relative pose can be determined, and the road feature data of the vehicle observation data of each observation point relative to the world coordinate system can also be determined according to the mapping relationship.

[0089] Optionally, the first anchor point can be divided into a starting anchor point and an ending anchor point, and the preset road section is between the starting anchor point and the ending anchor point.

[0090] Optionally, the anchor point is used to describe road features, and the road features include but are not limited to: road surface marking information (solid lines, dashed lines, arrows, text, etc.), lamp posts, road signs, traffic lights, building planes, etc.

[0091] Optionally, when the current observation point does not belong to the observation anchor point, the vehicle continues to drive, and continues to determine whether the next observation point belongs to the first anchor point until the first anchor point is encountered.

[0092] Optionally, since there is no preset area map of the preset road section in the preset map library, the first map (or the third map converted from the first map) is defaulted to the currently most credible map, and the confidence level of the first map (or the third map converted from the first map) can be set to the maximum value, that is, the confidence level is set to 1.

[0093] As an alternative embodiment, constructing a map based on a third map and confidence includes: splicing road feature data of multiple observation points in the third map according to the confidence.

[0094] Optionally, since the confidence of the third map is the maximum confidence, that is, the confidence of the third map is 1, it means that the road feature data of each observation point in the third map is the most credible. Therefore, directly splicing the road feature data of multiple key points can complete the map construction.

[0095] Optionally, if the relative pose TWB_i of the vehicle corresponds to the vehicle observation data PB_i, then when projected into the world coordinate system, it is expressed as the road feature data PW_i = TWB_i * PB_i. At this time, the road feature data PW_i is already in the world coordinate system. Here, i represents each observation point on the preset road section, and i = 1, 2, 3... n. Summing up all the observation points in the world coordinate system, that is, Psum = PW_1 + PW_2 + PW_3 +... + PW_n, constitutes the road backbone map.

[0096] As an alternative embodiment, determining the first anchor point of the first map includes: obtaining a preset feature dictionary corresponding to a target observation point in a preset road section, where the preset feature dictionary records the features of multiple observation points in the preset road section through feature descriptors; determining the mean feature vector of the target observation point and each observation point in the preset feature dictionary based on the feature descriptors; determining whether the mean feature vector is greater than a preset threshold, where the preset threshold is determined based on the feature descriptors of multiple observation points recorded in the preset feature dictionary; and determining the target observation point as the first anchor point when the mean feature vector is greater than the preset threshold.

[0097] In the above embodiments of the present invention, the preset feature dictionary is used to record the features of each observation point in the preset road section. Furthermore, based on the preset feature dictionary, a target observation point with prominent features in the preset road section can be found as the first anchor point.

[0098] Optionally, when the mean feature vector is greater than the preset threshold, it can be determined that the target observation point is not similar to other observation points, that is, the features of the target observation point are prominent compared to the features of other observation points. Furthermore, the target observation point can be used as the first anchor point to be matched with the second anchor point in the second map.

[0099] Optionally, assuming that there are a total of m feature descriptors of observation points in the preset feature dictionary, calculate the mean vector distance (d1 - mean) between the feature descriptor of the first observation point and the feature descriptors of other observation points.

[0100] For example, take the feature descriptor of the first observation point, calculate the vector distances between the feature descriptor of the first observation point and the feature descriptors of the second to the m-th observation points. That is, a total of m - 1 vector distances are calculated, and the mean value is obtained, which is the mean distance (d1-mean) between the feature descriptor of the first observation point and the feature descriptors of other observation points. By analogy, calculate the mean distances (d2-mean) …… (dm-mean) between the feature descriptors of the second to the m-th observation points in the preset feature dictionary and the feature descriptors of other observation points.

[0101] Furthermore, calculate the mean of the mean vector distances between the feature descriptors of the first to the m-th observation points and the feature descriptors of other observation points, that is, the mean vector distance (dmm) = (d1-mean + d2-mean + …… + dm-mean) / m.

[0102] Optionally, the preset threshold (d-threshold) = 2 * dmm.

[0103] As an optional embodiment, identifying a second anchor point that matches a first anchor point in a second map includes: obtaining the feature descriptor of the second anchor point in the second map of the road feature data; determining the feature vector value between the feature descriptor of the first anchor point of the road feature data and the feature descriptor of the second anchor point of the road feature data; when the feature vector value of the road feature data is not greater than the preset threshold, determining that the first anchor point of the road feature data matches the second anchor point of the road feature data. Wherein, when the first anchor point of the road feature data matches the second anchor point of the road feature data, determining the mapping relationship between the vehicle relative pose of the second anchor point of the road feature data and the vehicle observation pose of the first anchor point of the road feature data.

[0104] In the above embodiments of the present invention, when the feature vector value of the road feature data is not greater than the threshold, it indicates that the similarity between the first anchor point and the second anchor point exceeds the preset threshold. Thus, it can be determined that the first anchor point and the second anchor point are used to describe the same object, and then it is determined that the first anchor point and the second anchor point match.

[0105] Optionally, when the feature vector value of the road feature data is less than the threshold, it means that the first anchor point and the second anchor point are not similar. Then the first anchor point belongs to a brand-new anchor point, and further, the pose of the current anchor point is used as the vehicle relative pose to construct a new map.

[0106] Figure 5 is a schematic diagram of establishing a backbone map according to an embodiment of the present invention, as Figure 5 shown, through vehicle sensors such as cameras or LiDAR, etc., obtain the vehicle observation data of the current observation point, and calculate the vehicle observation data of the current observation point, and determine whether the current observation point belongs to the first anchor point through this vehicle observation data.

[0107] Optionally, the method for determining the first anchor point is as follows:

[0108] Step S11: Establish a preset feature dictionary in advance.

[0109] Optionally, collect vehicle observation data at observation points for a certain mileage (such as 1000 KM or more) in advance. The more observation points, the richer the included scenarios, which is more beneficial for anchor point determination. Discretely sample at a specific sampling rate in the collection path (this path can be of low precision and can be obtained through INS or simply through GPS). Specifically, the sampling rate can be set to values such as 5m or 10m. Generally speaking, the higher the sampling rate, the more information about the entire road is retained, but at the same time, the dictionary will be larger, occupying more space and making subsequent dictionary queries slower. Therefore, the sampling rate generally takes a suitable empirical value.

[0110] Optionally, at each observation point, obtain the vehicle observation data at that observation point (i.e., collect through vehicle sensors such as cameras or lidar), and calculate the feature descriptor corresponding to that observation point through the vehicle observation data.

[0111] Optionally, typical methods for calculating the feature descriptor include but are not limited to ORB descriptors, bag-of-words descriptors, etc.

[0112] Optionally, the feature descriptor is an encoded data in a specific format after processing the road feature data collected at the observation point through a specific algorithm (related to the type of feature descriptor used), generally represented as an n-dimensional vector, denoted as D here.

[0113] Optionally, the set of descriptors calculated at all observation points constitutes the preset feature dictionary.

[0114] Step S12: Obtain the vehicle observation data at the current observation point and calculate the feature descriptor of the current observation point.

[0115] Optionally, in step S11, vehicle observation data at a large number of observation points are collected, and feature descriptors are taken at discrete observation points (n), and these n feature descriptors constitute the preset feature dictionary.

[0116] In this step S12, the feature descriptor is calculated at the current observation point, and this feature descriptor depicts the road features of the current observation point.

[0117] Step S13: Determine whether the current observation point belongs to the first anchor point. Calculate the vector distance between the current observation point and each observation point in the preset feature dictionary, and calculate the average distance (i.e., the average feature vector). When the average distance (i.e., the average feature vector) is greater than the preset threshold d-threshold (such as 1 - 10 times dmm), the current observation point position is considered the first anchor point; otherwise, it is not the first anchor point.

[0118] Optionally, if the current observation point is not the first anchor point, no processing is performed, and the vehicle continues to travel until it encounters the first anchor point.

[0119] Optionally, if the current position observation point is the first anchor point, the vehicle sensor GNSS, IMU and wheel speed / vehicle speed information sequence is obtained starting from the current observation point. Then the vehicle posture information sequence, i.e., the vehicle INS trajectory, is obtained through INS. At the same time, the vehicle observation posture and vehicle observation data of the vehicle at each observation point are stored until the next anchor point appears.

[0120] Optionally, the vehicle observation data is based on the road data collected by the vehicle sensor (such as a camera or lidar, etc.), and the road features are obtained by detection, segmentation, and recognition methods. The content of the road features is defined by the needs of the user of the first map, and usually includes but is not limited to road marking information (solid line, dotted line, arrow, text, etc.), lamp poles, road signs, traffic lights, building planes, etc.

[0121] Optionally, the previous step obtains the vehicle INS trajectory (TWB1, TWB2, TWB3...TWBn) with two anchor points (starting point Astart, end point Aend), and the vehicle observation pose and vehicle observation data (PB1, PB2, PB3...PBn) at each observation point. The two anchor points carry the vehicle observation pose and vehicle observation data. The following takes the starting anchor point as an example to introduce the alignment method. The alignment of the end anchor point is the same.

[0122] Step S21, with the observed vehicle posture Tstart of the starting anchor point (i.e., the first anchor point) Astart, searches in a crowdsourced map (i.e., multiple adjacent area maps of a predetermined database) with a certain radius (here determined by the absolute positioning accuracy of INS, usually the absolute positioning accuracy of INS is 10m, then the search radius can be set to 20m. That is, 2 times the absolute positioning accuracy of INS), obtains the second anchor point Amap and the relative posture Tmap of the vehicle in the second map (i.e., the adjacent area map), and calculates the vector distance (i.e., the feature vector value) between the first anchor point and the second anchor point.

[0123] Optionally, the search radius is determined by the absolute positioning accuracy of the INS, and usually twice the absolute positioning accuracy of the INS can be taken as the search radius.

[0124] Step S22: If the vector distance (i.e., the feature vector value) is not greater than the preset threshold d-threshold (e.g., 1 - 10 times dmm), then use the pose corresponding to the second anchor point as the predicted pose, and register the current road observation features corresponding to the first anchor point with the crowdsourced map (registration is the process of aligning the road features in the first map with those in the second map, and the method can be an optimization method such as ICP, NDT, etc.) to obtain the relative pose of the vehicle at the first anchor point. Optionally, the registration method can adopt ICP, NDT, etc.

[0125] For example, with the simplest assumption, such as there is only one arrow on the first map based on the world coordinate system. And assume this is a one-dimensional map, and the arrow is at 10 meters on the map. The current vehicle observation data is based on the observation coordinate system, for example, it is observed that there is an arrow at 1 meter in front of the vehicle. Registration is the process of aligning the currently observed arrow with the arrow on the map. Once the arrows are aligned, it can be known that 1 meter behind the arrow on the map is the position of the vehicle, so it can be known that the vehicle is at 9 meters on the map.

[0126] It should be noted that the predicted position helps with registration. If it is initially known that the vehicle is approximately at 10 mi, then registration only needs to be done in the vicinity; otherwise, registration needs to be done on the entire map.

[0127] Optionally, if the vector distance (i.e., the feature vector value) is greater than the preset distance threshold, then consider the first anchor point as a completely new anchor point, and use the corresponding INS pose as the relative pose of the vehicle at the first anchor point.

[0128] Optionally, using the vehicle observation pose and INS trajectory as inputs, perform graph optimization on the entire INS trajectory. Here, the anchor points are two anchor points including the start and end of a section of the INS trajectory.

[0129] Optionally, use the above INS trajectory to provide the relative pose of the vehicle, project the vehicle observation data corresponding to each relative pose of the vehicle into the world coordinate system, and directly accumulate them to establish a road backbone map. The specific steps of the direct accumulation mentioned here are as follows: Project the corresponding vehicle observation data into the world coordinate system through the relative pose of the vehicle on the INS trajectory. For example, if the relative pose of the vehicle on the INS trajectory is TWB_i corresponding to the vehicle observation data PB_i, then it is represented as the road feature data PW_i = TWB_i * PB_i in the world coordinate system. At this time, PW_i is already in the world coordinate system. Here, i represents each observation point on the INS trajectory, i = 1, 2, 3... n. Accumulate all the observation points in the world coordinate system, that is, P_sum = PW_1 + PW_2 + PW_3 +... + PW_n, to form a road backbone map.

[0130] Optionally, the cloud server performs cloud data fusion, specifically including:

[0131] (1) For the backbone map (i.e., the third map) received from the vehicle end. Since the backbone map is only established where there is no preset regional map of the preset road; and the endpoints (i.e., anchor points) of the backbone map have been aligned with the existing second map (i.e., the adjacent regional map), it can be directly added to the second map.

[0132] (2) For the incremental map (i.e., the third map) received from the vehicle end. Since the incremental map (i.e., the third map) is established by registering the first map and the second map at the vehicle end, the incremental map (i.e., the third map) has been aligned with the second map, and thus can be directly superimposed on the second map, where the superimposition is performed with the confidence level of the new map as the weight.

[0133] (3) Through the above steps, multiple third maps obtained by multi-vehicle and multi-trip collection can be used to establish a high-precision and complete high-precision map.

[0134] Optionally, based on the weight accumulation of the confidence level, the steps are as follows: The vehicle relative pose TWj and road feature data PWj of the observation points of 3Dof, which are already in the world coordinate system, and their corresponding confidence levels are set as Cj, where j represents multiple observations of the same observation point, j = 1, 2, 3... n; Accumulate all the road feature data of the same observation point with the corresponding confidence levels, that is, the road fusion feature Pmean - weight = (PW1 * C1 + PW2 * C2 + PW3 * C3 +... + PWn * Cn) / (C1 + C2 + C3 +... + Cn), and its corresponding confidence level is Cmean = (C1 + C2 + C3 +... + Cn) / n.

[0135] The technical solution provided by the present invention uses INS (Inertial Navigation System) and anchor points to establish the overall road backbone structure, and superimposes multiple trips of road information on the road backbone structure through high-precision poses and confidence level information of relative positioning to obtain a high-precision and complete map. Compared with the typical SLAM method, in addition to the advantage of high map accuracy, this method also fuses and obtains the confidence level information of map elements based on the confidence level information of relative positioning. In addition, while collecting road information at the vehicle end and establishing the map, relative positioning services can be provided.

[0136] The technical solution provided by the present invention is that in crowdsourcing map building, due to the use of low-cost sensors (such as cameras, low-line Lidar, etc.) and factors such as occlusion in the actual data collection process, high-quality and complete road information cannot be obtained by single-trip collection. Therefore, it is necessary to fuse and process the road observation data obtained by multi-trip collection to obtain a high-quality and complete high-precision map.

[0137] The technical solution provided by the present invention performs fusion processing on multi-trip road observation data with low precision. Usually, relevant methods such as SLAM (simultaneous localization and mapping) can be used for processing. A typical method is to first generate a single-trip trajectory and the corresponding single-trip map for each trip of road data through an odometer, and then generate a complete high-precision map through operations such as aggregation optimization on multiple trips of maps. The main problem is that due to the cumulative error of the odometer, there are non-overlapping parts among multiple trips of map data. The purpose of aggregation optimization itself is to fuse non-overlapping multiple trips of map data, but due to the existence of a large number of similar structures on the actual road surface, aggregation optimization is prone to misregistration, resulting in map ghosting, and the final result is that the map accuracy is not high.

[0138] The technical solution provided by the present invention obtains the confidence level while obtaining the relative positioning pose through registration. The fusion of multiple trips of data is realized spatially through relative positioning and numerically through the weight based on the confidence level. Finally, a map with confidence information is obtained, which is beneficial to the refined processing of subsequent applications.

[0139] The technical solution provided by the present invention can provide positioning information while performing incremental mapping at the vehicle end.

[0140] The technical solution provided by the present invention realizes the alignment of positioning and single-trip data and the map at the vehicle end, and directly fuses multiple trips of data in the cloud, avoiding complex optimization work. It fully exploits and utilizes edge computing resources, greatly reducing the processing pressure of massive data in the cloud.

[0141] According to an embodiment of the present invention, an embodiment of a map construction device is also provided. It should be noted that this map construction device can be used to execute the map construction method in the embodiment of the present invention, and the map construction method in the embodiment of the present invention can be executed in this map construction device.

[0142] Figure 6 is a schematic diagram of a map construction device according to an embodiment of the present invention, as Figure 6 shown, the device may include: an acquisition unit 60, configured to acquire a first map of a preset road section, where the first map is drawn based on an observation coordinate system; a query unit 62, configured to query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; a determination unit 64, configured to determine the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; a conversion unit 66, configured to convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a construction unit 68, configured to construct a map according to the third map and the confidence level.

[0143] It should be noted that the obtaining unit 60 in this embodiment can be used to execute step S102 in the embodiment of the present application, the query unit 62 in this embodiment can be used to execute step S104 in the embodiment of the present application, the determining unit 64 in this embodiment can be used to execute step S106 in the embodiment of the present application, the conversion unit 66 in this embodiment can be used to execute step S108 in the embodiment of the present application, and the constructing unit 68 in this embodiment can be used to execute step S110 in the embodiment of the present application. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the above embodiments.

[0144] In the embodiment of the present invention, a first map of a preset road section is obtained, where the first map is drawn based on an observation coordinate system; a second map corresponding to the preset road section is queried, where the second map is drawn based on a world coordinate system; the registration relationship and confidence level between the first map and the second map are determined, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; the first map is converted into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a map is constructed according to the third map and the confidence level, thereby achieving the technical effect of generating a high-precision map based on low-precision road observation data, and further solving the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

[0145] As an alternative embodiment, the obtaining unit includes: a first obtaining module, configured to obtain a plurality of observation points collected by a vehicle on a preset road section; a first determining module, configured to determine the vehicle observation pose of each observation point, where the vehicle observation pose is determined based on an observation coordinate system; a collecting module, configured to collect the vehicle observation data of each observation point, where the vehicle observation data is determined based on the vehicle observation pose; a second determining module, configured to determine the first map of the preset road section based on the vehicle observation data of the plurality of observation points.

[0146] As an alternative embodiment, the query unit includes: a third determining module, configured to determine the location information of the preset road section; a query module, configured to query, in a preset map library, a preset area map corresponding to the location information and an adjacent area map, where the preset area map includes the preset road section, and the adjacent area map includes adjacent road sections of the preset road section; a fifth determining module, configured to determine the preset area map as the second map when the preset area map exists in the preset map library; a sixth determining module, configured to determine the adjacent area map as the second map when the preset area map does not exist in the preset map library.

[0147] As an alternative embodiment, the determination unit includes: a seventh determination module, configured to determine a matching relationship between an observation point in the first map and a preset acquisition point in the second map when determining that the preset area map is the second map; an eighth determination module, configured to determine a mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship, where the vehicle relative pose is determined based on the world coordinate system; a ninth determination module, configured to determine, according to the mapping relationship, road feature data of the vehicle observation data of each observation point relative to the world coordinate system, where the road feature data is used to construct a third map; a tenth determination module, configured to determine the number of matched observation points and the number of unmatched observation points according to the matching relationship; an eleventh determination module, configured to determine a confidence level according to the number of matched observation points and the number of unmatched observation points.

[0148] As an alternative embodiment, the construction unit includes: a second acquisition module, configured to acquire multiple road feature data of the same observation point in multiple third maps and the confidence level corresponding to each road feature data; a superimposing module, configured to superimpose the multiple road feature data of the observation point with the confidence level as a weight to obtain a road fusion feature of the observation point; a first splicing module, configured to splice the road fusion features of multiple observation points.

[0149] As an alternative embodiment, the determination unit includes: a twelfth determination module, configured to determine a first anchor point of the first map among multiple observation points of a preset road section when determining that the adjacent area map is the second map; an identification module, configured to identify a second anchor point in the second map that matches the first anchor point, where the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; a thirteenth determination module, configured to determine a mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, where the vehicle relative pose is determined based on the world coordinate system; a fourteenth determination module, configured to determine, according to the mapping relationship, road feature data of the vehicle observation data of each observation point relative to the world coordinate system, where the road feature data is used to construct a third map; a fifteenth determination module, configured to determine that the confidence level is a preset value.

[0150] As an alternative embodiment, the construction unit includes: a second splicing module, configured to splice the road feature data of multiple observation points in the third map according to the confidence level.

[0151] As an alternative embodiment, determining the first anchor point of the first map includes: a third acquisition module configured to acquire a preset feature dictionary corresponding to a target observation point in a preset road section, where the preset feature dictionary records the features of multiple observation points in the preset road section through feature descriptors; a sixteenth determination module configured to determine the mean value of the feature vectors between the target observation point and each observation point in the preset feature dictionary based on the feature descriptors; a judgment module configured to judge whether the mean value of the feature vectors is greater than a preset threshold, where the preset threshold is determined based on the feature descriptors of the multiple observation points recorded in the preset feature dictionary; a seventeenth determination module configured to, when the mean value of the feature vectors is greater than the preset threshold, determine the target observation point as the first anchor point.

[0152] As an alternative embodiment, the recognition module includes: a fourth acquisition module configured to acquire the feature descriptor of the second anchor point in the second map; an eighteenth determination module configured to determine the feature vector value between the feature descriptor of the first anchor point and the feature descriptor of the second anchor point; a nineteenth determination module configured to, when the feature vector value is not greater than a threshold, determine that the first anchor point matches the second anchor point, where, when the first anchor point matches the second anchor point, determine the mapping relationship between the vehicle relative pose of the second anchor point and the vehicle observation pose of the first anchor point.

[0153] Figure 7 is a schematic diagram of a map construction system according to an embodiment of the present invention, as Figure 7 shown, the device may include: at least one vehicle 70 configured to draw a first map of a preset road section based on an observation coordinate system; query a second map of the preset road section drawn based on a world coordinate system in a cloud server; determine the registration relationship and confidence level between the first map and the second map, and convert the first map into a third map drawn based on the world coordinate system according to the registration relationship; upload the third map and the confidence level to the cloud server; the cloud server 72 is further configured to construct a map according to the registration relationship and confidence level of at least one third map.

[0154] In an embodiment of the present invention, a first map of a preset road section is acquired, where the first map is drawn based on an observation coordinate system; a second map corresponding to the preset road section is queried, where the second map is drawn based on a world coordinate system; the registration relationship and confidence level between the first map and the second map are determined, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; the first map is converted into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a map is constructed according to the third map and the confidence level, thereby achieving the technical effect of generating a high-precision map based on low-precision road observation data, and further solving the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

[0155] Embodiments of the present invention can provide a computer terminal, which can be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the above computer terminal can also be replaced with a terminal device such as a mobile terminal.

[0156] Optionally, in this embodiment, the above computer terminal can be located in at least one of multiple network devices in a computer network.

[0157] In this embodiment, the above computer terminal, through the processor, can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain a first map of a preset road section, where the first map is drawn based on an observation coordinate system; query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determine the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree of the first map and the second map; convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; construct a map according to the third map and the confidence level.

[0158] In the embodiments of the present invention, by obtaining a first map of a preset road section, where the first map is drawn based on an observation coordinate system; querying a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determining the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree of the first map and the second map; converting the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; and constructing a map according to the third map and the confidence level, the technical effect of generating a high-precision map based on low-precision road observation data is achieved, and thus the technical problem of being unable to establish a high-precision map based on low-precision road observation data is solved.

[0159] Optionally, the above processor can also execute the program code of the following steps: obtain a first map of a preset road section, where the first map is drawn based on an observation coordinate system; query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determine the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree of the first map and the second map; convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; construct a map according to the third map and the confidence level.

[0160] Optionally, the above-mentioned processor may also execute the program code of the following steps: obtaining a plurality of observation points collected by the vehicle on a preset road section; determining the vehicle observation pose of each observation point, wherein the vehicle observation pose is determined based on an observation coordinate system; collecting the vehicle observation data of each observation point, wherein the vehicle observation data is determined based on the vehicle observation pose; determining a first map of the preset road section based on the vehicle observation data of the plurality of observation points.

[0161] Optionally, the above-mentioned processor may also execute the program code of the following steps: determining the position information of the preset road section; querying in a preset map library for a preset area map corresponding to the position information and an adjacent area map, wherein the preset area map includes the preset road section and the adjacent area map includes the adjacent road sections of the preset road section; in the case where the preset area map exists in the preset map library, determining the preset area map as the second map; in the case where the preset area map does not exist in the preset map library, determining the adjacent area map as the second map.

[0162] Optionally, the above-mentioned processor may also execute the program code of the following steps: in the case where the preset area map is determined as the second map, determining the matching relationship between the observation points in the first map and the preset collection points in the second map; according to the matching relationship, determining the mapping relationship between the vehicle observation pose and the vehicle relative pose, wherein the vehicle relative pose is determined based on a world coordinate system; according to the mapping relationship, determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system, wherein the road feature data is used to construct a third map; determining the number of matched and unmatched observation points according to the matching relationship; determining the confidence level according to the number of matched and unmatched points.

[0163] Optionally, the above-mentioned processor may also execute the program code of the following steps: obtaining a plurality of road feature data of the same observation point in a plurality of third maps and the confidence level corresponding to each road feature data; superimposing the plurality of road feature data of the observation point with the confidence level as the weight to obtain the road fusion feature of the observation point; splicing the road fusion features of the plurality of observation points.

[0164] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: Optionally, the above-mentioned processor may also execute the program code for the following steps: When determining that the adjacent area map is the second map, among multiple observation points on a preset road section, determine the first anchor point of the first map; identify a second anchor point in the second map that matches the first anchor point, where the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; determine the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, where the vehicle relative pose is determined based on the world coordinate system; determine the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, where the road feature data is used to construct the third map; determine that the confidence level is a preset value.

[0165] Optionally, the above-mentioned processor may also execute the program code for the following steps: Stitch the road feature data of multiple observation points in the third map according to the confidence level.

[0166] Optionally, the above-mentioned processor may also execute the program code for the following steps: Optionally, the above-mentioned processor may also execute the program code for the following steps: Record the features of multiple observation points in the preset road section through the feature descriptor; determine the mean value of the feature vectors of the target observation point and each observation point in the preset feature dictionary based on the feature descriptor; determine whether the mean value of the feature vectors is greater than a preset threshold, where the preset threshold is determined based on the feature descriptors of multiple observation points recorded in the preset feature dictionary; when the mean value of the feature vectors is greater than the preset threshold, determine that the target observation point is the first anchor point.

[0167] Optionally, the above-mentioned processor may also execute the program code for the following steps: Obtain the feature descriptor of the second set anchor point in the second map; determine the feature vector value between the feature descriptor of the first anchor point and the feature descriptor of the second anchor point; when the feature vector value is not greater than the threshold, determine that the first anchor point matches the second anchor point, where when the first anchor point matches the second anchor point, determine the mapping relationship between the vehicle relative pose of the second anchor point and the vehicle observation pose of the first anchor point.

[0168] An embodiment of the present invention also provides a computer-readable storage medium. Optionally, in this embodiment, the above-mentioned computer-readable storage medium may be used to save the program code executed by the above data transmission method.

[0169] Optionally, in this embodiment, the above-mentioned computer-readable storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.

[0170] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining a first map of a preset road section, where the first map is drawn based on an observation coordinate system; querying a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determining a registration relationship and a confidence level between the first map and the second map, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; converting the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; constructing a map according to the third map and the confidence level.

[0171] In an embodiment of the present invention, a first map of a preset road section is obtained, where the first map is drawn based on an observation coordinate system; a second map corresponding to the preset road section is queried, where the second map is drawn based on a world coordinate system; a registration relationship and a confidence level between the first map and the second map are determined, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; the first map is converted into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; a map is constructed according to the third map and the confidence level, thereby achieving the technical effect of generating a high-precision map based on low-precision road observation data, and further solving the technical problem of being unable to establish a high-precision map based on low-precision road observation data.

[0172] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining a first map of a preset road section, where the first map is drawn based on an observation coordinate system; querying a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system; determining a registration relationship and a confidence level between the first map and the second map, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, and the confidence level is determined according to the registration degree between the first map and the second map; converting the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; constructing a map according to the third map and the confidence level.

[0173] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining a plurality of observation points collected by a vehicle on a preset road section; determining the vehicle observation pose of each observation point, where the vehicle observation pose is determined based on an observation coordinate system; collecting vehicle observation data of each observation point, where the vehicle observation data is determined based on the vehicle observation pose; determining a first map of the preset road section based on the vehicle observation data of the plurality of observation points.

[0174] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: determining the location information of a preset road section; querying in a preset map library for a preset area map corresponding to the location information and an adjacent area map, where the preset area map includes the preset road section and the adjacent area map includes adjacent road sections of the preset road section; determining the preset area map as the second map when the preset area map exists in the preset map library; and determining the adjacent area map as the second map when the preset area map does not exist in the preset map library.

[0175] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: determining the matching relationship between an observation point in the first map and a preset collection point in the second map when the preset area map is determined as the second map; determining the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship, where the vehicle relative pose is determined based on the world coordinate system; determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, where the road feature data is used to construct a third map; determining the number of matched and unmatched observation points according to the matching relationship; and determining the confidence level according to the number of matched and unmatched observation points.

[0176] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining multiple road feature data of the same observation point in multiple third maps and the confidence level corresponding to each road feature data; superimposing the multiple road feature data of the observation point with the confidence level as the weight to obtain the road fusion feature of the observation point; and stitching the road fusion features of multiple observation points.

[0177] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: determining a first anchor point of the first map among multiple observation points of the preset road section when the adjacent area map is determined as the second map; identifying a second anchor point in the second map that matches the first anchor point, where the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; determining the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, where the vehicle relative pose is determined based on the world coordinate system; determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, where the road feature data is used to construct a third map; and determining the confidence level as a preset value.

[0178] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: stitching the road feature data of multiple observation points in the third map according to the confidence level.

[0179] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining a preset feature dictionary corresponding to a target observation point in a preset road section, where the preset feature dictionary records the features of multiple observation points in the preset road section through feature descriptors; determining the mean value of the feature vectors of the target observation point and each observation point in the preset feature dictionary based on the feature descriptors; determining whether the mean value of the feature vectors is greater than a preset threshold, where the preset threshold is determined based on the feature descriptors of the multiple observation points recorded in the preset feature dictionary; and determining the target observation point as a first anchor point when the mean value of the feature vectors is greater than the preset threshold.

[0180] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: obtaining the feature descriptor of a second anchor point in the second map; determining the feature vector value between the feature descriptor of the first anchor point and the feature descriptor of the second anchor point; and determining that the first anchor point matches the second anchor point when the feature vector value is not greater than a threshold, where when the first anchor point matches the second anchor point, determining the mapping relationship between the relative vehicle pose of the second anchor point and the observed vehicle pose of the first anchor point.

[0181] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0182] In the above embodiments of the present invention, the descriptions of the various embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0183] In several embodiments provided in the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units can be a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the units or modules can be in an electrical or other form.

[0184] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0185] In addition, in each embodiment of the present invention, the functional units may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0186] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The foregoing storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0187] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for map construction, characterized in that, it includes: Obtain a first map of a preset road section, where the first map is drawn based on an observation coordinate system; Query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system, and the second map is a preset area map including the preset road section, or an adjacent area map including adjacent road sections of the preset road section. In the case where the preset area map does not exist, the second map is the adjacent area map; Determine the registration relationship and confidence level between the first map and the second map, where the registration relationship includes: the mapping relationship between the observation coordinate system and the world coordinate system, determined by matching features in the first map and features in the second map, and the confidence level is determined according to the registration degree between the first map and the second map. In the case where the second map is the preset area map, the confidence level is determined according to the number of matched and unmatched observation points in the first map and the second map. In the case where the second map is the adjacent area map, the confidence level is a preset value; Convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; Construct a map according to the third map and the confidence level.

2. The method according to claim 1, characterized in that, Obtaining a first map of a preset road section includes: Obtain a plurality of observation points collected by a vehicle on the preset road section; Determine the vehicle observation pose of each observation point, where the vehicle observation pose is determined based on the observation coordinate system; Collect vehicle observation data of each observation point, where the vehicle observation data is determined based on the vehicle observation pose; Determine the first map of the preset road section based on the vehicle observation data of multiple observation points.

3. The method according to claim 2, characterized in that, Querying a second map corresponding to the preset road section includes: Determine the location information of the preset road section; Query the preset area map and the adjacent area map corresponding to the location information in a preset map library; In the case where the preset area map exists in the preset map library, determine the preset area map as the second map; In the case where the preset area map does not exist in the preset map library, determine the adjacent area map as the second map.

4. The method according to claim 3, characterized in that, In the case where the preset area map is determined as the second map, determining the registration relationship and confidence level between the first map and the second map includes: Determine the matching relationship between the observation points in the first map and the preset collection points in the second map; According to the matching relationship, determine the mapping relationship between the vehicle observation pose and the vehicle relative pose, where the vehicle relative pose is determined based on the world coordinate system; According to the mapping relationship, determine the road feature data of the vehicle observation data of each observation point relative to the world coordinate system, where the road feature data is used to construct the third map.

5. The method according to claim 4, wherein, constructing a map according to the third map and the confidence includes: obtaining multiple road feature data of the same observation point in multiple third maps, and the confidence corresponding to each road feature data; superimposing the multiple road feature data of the observation point with the confidence as the weight to obtain the road fusion feature of the observation point; stitching the road fusion features of multiple observation points.

6. The method according to claim 3, wherein, when determining that the adjacent area map is the second map, determining the registration relationship and confidence between the first map and the second map includes: determining a first anchor point of the first map among multiple observation points on the preset road section; identifying a second anchor point matching the first anchor point in the second map, wherein the object described by the first anchor point in the first map is the same as the object described by the second anchor point in the second map; determining the mapping relationship between the vehicle observation pose and the vehicle relative pose according to the matching relationship between the first anchor point and the second anchor point, wherein the vehicle relative pose is determined based on the world coordinate system; determining the road feature data of the vehicle observation data of each observation point relative to the world coordinate system according to the mapping relationship, wherein the road feature data is used to construct a third map.

7. The method according to claim 6, wherein, constructing a map according to the third map and the confidence includes: stitching the road feature data of multiple observation points in the third map according to the confidence.

8. The method according to claim 6, wherein, determining the first anchor point of the first map includes: obtaining a preset feature dictionary corresponding to a target observation point in the preset road section, wherein the preset feature dictionary records the features of multiple observation points in the preset road section through feature descriptors; determining the mean value of the feature vectors of the target observation point and each observation point in the preset feature dictionary based on the feature descriptors; judging whether the mean value of the feature vectors is greater than a preset threshold, wherein the preset threshold is determined based on the feature descriptors of multiple observation points recorded in the preset feature dictionary; when the mean value of the feature vectors is greater than the preset threshold, determining the target observation point as the first anchor point.

9. The method according to claim 8, wherein, identifying a second anchor point matching the first anchor point in the second map includes: obtaining the feature descriptor of a second set anchor point in the second map; determining the feature vector value between the feature descriptor of the first anchor point and the feature descriptor of the second anchor point; when the feature vector value is not greater than the threshold, determining that the first anchor point matches the second anchor point, and when the first anchor point matches the second anchor point, determining the mapping relationship between the vehicle relative pose of the second anchor point and the vehicle observation pose of the first anchor point.

10. A map construction device, wherein, comprising: An acquisition unit, configured to acquire a first map of a preset road section, where the first map is drawn based on an observation coordinate system; A query unit, configured to query a second map corresponding to the preset road section, where the second map is drawn based on a world coordinate system, the second map is a preset area map including the preset road section, or an adjacent area map including adjacent road sections of the preset road section, and in the case where the preset area map does not exist, the second map is the adjacent area map; A determination unit, configured to determine a registration relationship and a confidence level between the first map and the second map, where the registration relationship includes: a mapping relationship between the observation coordinate system and the world coordinate system, determined by matching features in the first map and features in the second map, and the confidence level is determined according to the registration degree between the first map and the second map. In the case where the second map is the preset area map, the confidence level is determined according to the number of matched and unmatched observation points in the first map and the second map. In the case where the second map is the adjacent area map, the confidence level is a preset value; A conversion unit, configured to convert the first map into a third map based on the registration relationship, where the third map is drawn based on the world coordinate system; A construction unit, configured to construct a map according to the third map and the confidence level.

11. A map construction system, Characterized in that, It includes: At least one vehicle, configured to draw a first map of a preset road section based on an observation coordinate system; Query in a cloud server for a second map of the preset road section drawn based on a world coordinate system; Determine the registration relationship and the confidence level between the first map and the second map, and convert the first map into a third map drawn based on the world coordinate system according to the registration relationship; Upload the third map and the confidence level to the cloud server, where the second map is a preset area map including the preset road section, or an adjacent area map including adjacent road sections of the preset road section. In the case where the preset area map does not exist, the second map is the adjacent area map. The registration relationship is determined by matching features in the first map and features in the second map. In the case where the second map is the preset area map, the confidence level is determined according to the number of matched and unmatched observation points in the first map and the second map. In the case where the second map is the adjacent area map, the confidence level is a preset value; The cloud server is further configured to construct a map according to the registration relationship and the confidence level of at least one of the third maps.

12. A computer-readable storage medium, Characterized in that, The computer-readable storage medium includes a stored program, where when the program runs, it controls the device where the computer-readable storage medium is located to execute the map construction method according to any one of claims 1 to 9.

13. A processor, Characterized in that, The processor is used to run a program, wherein, when the program runs, it executes the map construction method described in any one of claims 1 to 9.

Images