Mobile control strategy determination method and system

Abstract

The embodiment of the invention provides a mobile control strategy determination method and system, and the method comprises the steps: determining actual measurement elements based on the mobile data of a terminal; the actual measurement elements comprise actual measurement strong semantic elements and actual measurement weak semantic elements; matching the actually measured elements with reference elements of the target area, and determining strategy auxiliary information; the reference elements comprise reference strong semantic elements and reference weak semantic elements; wherein the actually-measured strong semantic elements and / or the reference strong semantic elements comprise element objects with definite semantic features, and the actually-measured weak semantic elements and / or the reference weak semantic elements comprise abstract element objects with a constraint effect on positioning; and determining a mobile control strategy of the terminal based on the strategy auxiliary information.

Description

Technical Field

[0001] The present invention relates to the field of mobile terminal control, and in particular to a method and system for determining a mobile control strategy. Background Art

[0002] Instructing, planning, and controlling terminals to replace human work is one of the difficulties in the current field of industrial automatic control, and has received widespread attention in recent years. Automatic terminal control means that the terminal relies on the coordinated cooperation of artificial intelligence, visual computing, detection radar, monitoring devices, and positioning systems to enable the terminal to automatically and safely operate without real-time manual operation. For example, when the terminal is moving, it is necessary to combine the terminal's real-time perception and positioning of surrounding data to obtain policy auxiliary information for determining future mobile control strategies. In turn, the accuracy of the terminal's real-time perception and positioning of surrounding data will directly affect the accuracy of future mobile control strategies. Summary of the invention

[0003] The embodiments of this specification provide a method and system for determining a mobile control strategy, which can improve the real-time perception of surrounding data and the accuracy of positioning during the movement of a terminal.

[0004] One of the embodiments of the present specification provides a method for determining a mobile control strategy, the method comprising: determining measured elements based on mobile data of a terminal; the measured elements include measured strong semantic elements and measured weak semantic elements; matching the measured elements with reference elements of the target area to determine strategy auxiliary information; the reference elements include reference strong semantic elements and reference weak semantic elements; wherein the measured strong semantic elements and / or the reference strong semantic elements include element objects with clear semantic features, and the measured weak semantic elements and / or the reference weak semantic elements include abstract element objects that have a constraining effect on positioning; based on the strategy auxiliary information, determining the mobile control strategy of the terminal.

[0005] One of the embodiments of the present specification provides a mobile control strategy determination system, the system comprising: an element determination module, used to determine measured elements based on terminal movement data; the measured elements include measured strong semantic elements and measured weak semantic elements; an information determination module, used to match the measured elements with reference elements of the target area to determine strategy auxiliary information; the reference elements include reference strong semantic elements and reference weak semantic elements; wherein the measured strong semantic elements and / or the reference strong semantic elements include element objects with clear semantic features, and the measured weak semantic elements and / or the reference weak semantic elements include abstract element objects that have constraints on positioning; a strategy determination module, used to determine the mobile control strategy of the terminal based on the strategy auxiliary information.

[0006] One of the embodiments of this specification provides a mobile control strategy determination device, the device comprising: at least one storage medium storing computer instructions; at least one processor executing the computer instructions to implement a mobile control strategy determination method.

[0007] One of the embodiments of the present specification provides a storage medium storing computer instructions, and when a computer reads the computer instructions, the computer executes the mobile control strategy determination method. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0009] Figure 1 is a schematic diagram of an application scenario of a mobile control strategy determination system according to some embodiments of this specification;

[0010] Figure 2 is an exemplary module diagram of a mobile control strategy determination system according to some embodiments of this specification;

[0011] Figure 3 is an exemplary flow chart of a method for determining a mobility control strategy according to some embodiments of this specification;

[0012] Figure 4 is an exemplary schematic diagram of determining strategy auxiliary information according to some embodiments of this specification;

[0013] Figure 5 is an exemplary schematic diagram of reference elements shown in some embodiments of this specification;

[0014] Figure 6 is an exemplary schematic diagram of a confidence model according to some embodiments of the present specification. DETAILED DESCRIPTION

[0015] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0016] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0017] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0018] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0019] Figure 1 Schematic diagram of application scenarios of the mobile control strategy determination system shown in some embodiments of this specification. Figure 1 As shown, the application scenario of the mobility control strategy determination system 100 may include a mobility area 110 , strategy auxiliary information 120 , a processor 130 , and a terminal 140 .

[0020] The movement area 110 refers to the relevant area where the terminal 140 moves. Different terminals have different corresponding movement areas. For example, for a terminal with a large range of movement, such as a vehicle, the movement area 110 may be an urban area, a highway section, etc. For another example, for a terminal with a limited range of movement, such as a robot, the movement area may be a space where the robot is placed or where the robot works.

[0021] The policy assistance information 120 is information related to the mobile behavior of the terminal. For example, the policy assistance information may be the current mobile state information, the current location information, the current mobile control policy information, etc. of the terminal 140.

[0022] In some embodiments, the policy assistance information 120 may be determined based on a processor. For example, the processor may determine the measured elements based on processing the data collected by the terminal 140, and determine the policy assistance information based on the processing of the measured elements. Figure 3 Description of the contents.

[0023] The processor 130 may process data and / or information obtained from other components of the mobile control strategy determination system 100 or other external devices. The processor 130 may execute program instructions based on the data, information and / or processing results to perform one or more functions described in this specification. For example, the processor 130 may process the mobile data obtained from the terminal 140, determine the measured elements based on the mobile data, and then determine the policy auxiliary information.

[0024] In some embodiments, the processor 130 may include one or more sub-processing devices (e.g., a single-core processing device or a multi-core multi-core processing device). As an example only, the processor 130 may include one or any combination of a central processing unit (CPU), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), etc. In some embodiments, the processor 130 may be integrated in the terminal 140.

[0025] The terminal 140 refers to a device for which a mobile control strategy needs to be determined. For example, the terminal 140 may be an autonomous driving vehicle 140-1, a robot 140-2, a drone 140-3, etc. The autonomous driving vehicle 140-1 may be a purely unmanned vehicle without a driver or an auxiliary driving module in a vehicle. The auxiliary driving module in the vehicle may autonomously control the vehicle's travel or output driving prompts to the driver.

[0026] In some embodiments, the terminal 140 may collect movement data through GPS, inertial sensors, laser radar or cameras, and send it to the processor 130 for subsequent processing. In some embodiments, the terminal 140 may obtain and execute the movement control strategy issued by the processor. Figure 3 And other content.

[0027] The positioning system 150 may include one or more satellites, for example, the positioning system 150 may include satellite 150-1, satellite 150-2 and satellite 150-3. In some embodiments, the positioning system 150 may be a global positioning system (GPS), a global navigation satellite system (GLONASS), a compass navigation system (COMPASS), a Beidou navigation satellite system, a Galileo positioning system, a quasi-zenith satellite system (QZSS), etc. The positioning system 150 may send a navigation signal to a positioning component in the terminal 140 via a wireless connection.

[0028] The above scenarios are only examples, and there may be various forms of transformations in the application.

[0029] Figure 2 200 is an exemplary module diagram of a mobile control strategy determination system according to some embodiments of this specification. The mobile control strategy determination system 200 may be implemented by a control component in the processor 130 and / or the terminal 140. Figure 2 As shown, the mobility control policy determination system 200 may include a factor determination module 210 , an information determination module 220 and a policy determination module 230 .

[0030] The element determination module 210 may be configured to determine measured elements based on the movement data of the terminal.

[0031] The information determination module 220 may be used to match the measured elements with the reference elements of the target area to determine the strategy auxiliary information.

[0032] In some embodiments, the information determination module 220 can be further used to match the measured strong semantic elements with the reference strong semantic elements of the sub-region to determine the first evaluation value of the sub-region; match the measured weak semantic elements with the reference weak semantic elements of the sub-region to determine the second evaluation value of the sub-region; determine the final evaluation value of the sub-region based on the first evaluation value and the second evaluation value; determine the sub-region with the highest final evaluation value as the target sub-region; and determine the policy auxiliary information based on the target sub-region.

[0033] The policy determination module 230 may be configured to determine a mobility control policy of the terminal based on the policy auxiliary information.

[0034] In some embodiments, the final evaluation value is a weighted value of the first evaluation value and the second evaluation value; the weight of the first evaluation value is positively correlated to the richness of the reference strong semantic elements of the sub-region; the weight of the second evaluation value is positively correlated to the richness of the reference weak semantic elements of the sub-region. In some embodiments, the final evaluation value is a weighted value of the first evaluation value and the second evaluation value; the weight of the first evaluation value is positively correlated to the degree of match between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region; the weight of the second evaluation value is positively correlated to the degree of match between the richness of the measured weak semantic elements and the richness of the reference weak semantic elements of the sub-region. For more information on the richness of reference strong / weak semantic elements, the richness of measured strong / weak semantic elements and their matching degree, see Figure 4 and its related description.

[0035] It should be noted that the above description of the mobile control strategy determination system 200 and its modules is only for convenience of description and does not limit the present specification to the scope of the embodiments. It is understandable that, after understanding the principle of the system, those skilled in the art may arbitrarily combine the modules or form a subsystem to connect with other modules without deviating from the principle. In some embodiments, Figure 2The element determination module 210, information determination module 220 and strategy determination module 230 disclosed in the specification can be different modules in a system, or a module can realize the functions of two or more modules. For example, each module can share a storage module, or each module can have its own storage module. Such variations are all within the scope of protection of this specification.

[0036] Figure 3 is an exemplary flow chart of a method for determining a mobile control strategy according to some embodiments of this specification. In some embodiments, process 300 may be executed by a control component in processor 130 or terminal 140. For example, process 300 may be executed by a mobile control strategy determination system 200 implemented by a control component in processor 130 or terminal 140. Figure 3 As shown, process 300 includes the following steps:

[0037] Step 310, determining measured elements based on the mobile data of the terminal.

[0038] Mobile data refers to relevant data generated or collected by the terminal during movement. For example, mobile data may include images or videos of the moving area or laser point cloud data collected by the terminal. For another example, mobile data may include the moving route data, trajectory data, navigation data, etc. of the terminal.

[0039] The measured elements refer to the semantic elements around the terminal obtained based on the mobile data. The measured elements include measured strong semantic elements and measured weak semantic elements. In some embodiments, the measured elements can be obtained based on the processing of the mobile data by the processor. For example, the processor processes the image or laser point cloud data of the mobile area in the mobile data based on algorithms such as image recognition to obtain measured strong semantic elements and measured weak semantic elements.

[0040] The measured strong semantic elements refer to element objects with clear semantic features extracted based on the actually acquired mobile data, which can clearly represent certain element objects with clear semantic features in the mobile area. For example, the measured strong semantic elements can be elements with clear meanings on the road that are actually acquired. For example, the elements on the road can include pavement elements and roadside elements. Pavement elements can include lane lines, road signs (turn signs, sidewalks, stop lines, bus lane signs, etc.), etc. Roadside elements can include street lights, traffic lights, signs, boundaries on both sides of the road, road separators, etc.

[0041] In some embodiments, the processor may use model processing, image recognition and other technologies to process images and / or videos collected by the autonomous driving vehicle to obtain measured strong semantic elements.

[0042] In some embodiments, based on different forms of the measured strong semantic elements, the measured strong semantic elements can be divided into different types. In some embodiments, the measured strong semantic elements can be divided into texture sub-elements and geometric sub-elements.

[0043] For example, digital or textual elements in the collected mobile data (such as the vehicle spacing numbers on the highway, the "bus lane" text on the bus lane, the text or numbers on the sign, the traffic light countdown numbers, etc.) can be used as texture sub-elements; for another example, geometric elements such as road boundaries, lane lines, stop lines, crosswalks, and turn signs on the road surface in the collected mobile data can be used as geometric sub-elements. For more information about texture sub-elements and geometric sub-elements, see Figure 5 and its related description.

[0044] Measured weak semantic elements refer to abstract element objects extracted based on mobile data that can constrain positioning. Measured weak semantic elements can correspond to elements on the road that have no specific meaning or have vague meanings. For example, measured weak semantic elements can be the corners, edges, surfaces, etc. of road surface elements and roadside elements in the laser point cloud data or images collected in the mobile area, such as the corners of buildings, the edges of billboards, and the side walls of roads.

[0045] In some embodiments, the processor can use model processing, image recognition, feature extraction and other technologies to process images, videos, laser point cloud data, etc. collected by the autonomous driving vehicle to obtain measured weak semantic elements.

[0046] In some embodiments, the processor may obtain the measured weak semantic elements in a less difficult manner than obtaining the measured strong semantic elements. For example, when obtaining the measured strong semantic elements, it is necessary to identify the category of the element and / or to identify the complete element, for example, it is necessary to identify a complete left turn sign.

[0047] As can be seen from the above content, since the measured weak semantic elements do not need to have specific meanings, when extracting the measured weak semantic elements, it is not necessary to identify the category to which the corresponding element belongs (for example, whether it belongs to a road sign or a traffic light), nor is it necessary to extract the complete road element corresponding to the corresponding element. For example, the extracted measured weak semantic element can be a part of the complete outline of the corresponding object, an edge, or a part of an edge, or a part of the surface, etc., thereby reducing the difficulty of extracting the measured weak semantic elements, and a large number of weak semantic elements can also be extracted.

[0048] For the difference between reference strong semantic elements and reference weak semantic elements, please refer to the above description of the difference between measured strong semantic elements and measured weak semantic elements. In some embodiments, since more data acquisition devices or resources can be used when obtaining reference elements, more reference strong semantic elements can be obtained from the perspective of improving the accuracy and completeness of the elements, and more reference weak semantic elements can be obtained in more precise ways.

[0049] In some embodiments, based on the different forms of the measured weak semantic elements, the measured weak semantic elements can be divided into different types. In some embodiments, the measured weak semantic elements can be divided into line elements and surface elements. For example, line features without specific semantics in the collected mobile data can be used as line elements; for another example, surface features without specific semantics in the collected mobile data can be used as surface elements. For more information about line elements and surface elements, see Figure 5 and its related description.

[0050] Step 320, matching the measured elements with the reference elements of the target area to determine the strategy auxiliary information.

[0051] The target area refers to the preliminary determination result of the mobile area where the terminal is currently located.

[0052] The processor can determine the target area in a variety of ways. In some embodiments, the processor can determine the target area based on the mobile data. For example, the processor can preliminarily determine the approximate range of the mobile area where the terminal is currently located based on the navigation data in the mobile data, and then determine the target area.

[0053] In some embodiments, the processor may determine the target area based on obtaining user input.

[0054] In some embodiments, the target area may be determined based on a coarse-grained spatial matching result of the movement data.

[0055] The coarse-grained spatial matching result refers to a positioning result based on rough positioning of mobile data. For example, the coarse-grained spatial matching result may be a preliminary positioning result obtained based on a preliminary matching of mobile data with preset map data. The processor may roughly determine the area where the terminal is currently located based on the preliminary positioning result. The preset map data may be a map resource provided by a third party or a pre-cached map resource.

[0056] The processor can determine the coarse-grained spatial matching result in a variety of ways. In some embodiments, the coarse-grained spatial matching result can be determined based on the trajectory data in the mobile data. For example, the processor can collect a large amount of historical trajectory data and its corresponding historical positions, and establish a trajectory database; perform similarity matching based on the current trajectory data and the historical trajectory data, and find the historical trajectory data with the highest similarity to the current trajectory data as the reference trajectory data; determine the mobile area where the historical position corresponding to the reference trajectory data is located as the coarse-grained spatial matching result.

[0057] In some embodiments, the coarse-grained spatial matching result may be determined based on a combination of satellite positioning and inertial navigation data.

[0058] Satellite positioning refers to a location obtained using a satellite locator. In some embodiments, satellite positioning can be obtained based on a satellite locator configured on a terminal.

[0059] Inertial navigation data refers to data obtained by an inertial sensor. In some embodiments, the inertial navigation data can be obtained based on an inertial sensor configured on the terminal. An inertial sensor refers to a device that converts inertial force into an electrical signal. For example, an inertial sensor may include an accelerometer and a gyroscope.

[0060] In some embodiments, the processor can transmit the longitude, latitude and altitude data obtained based on satellite positioning to the inertial navigation system to further obtain the terminal's attitude, position, speed and other data, and use the position data as the coarse-grained spatial matching result.

[0061] In some embodiments of the present specification, the coarse-grained spatial matching result can be determined based on the combination of satellite positioning and inertial navigation data, which can achieve the advantages of combining satellite positioning and inertial navigation, reduce the failure risk of a single technology, and improve the overall matching accuracy.

[0062] In some embodiments, the processor may determine the moving area corresponding to the coarse-grained spatial matching result as the target area.

[0063] In some embodiments of the present specification, the target area is determined based on the coarse-grained spatial matching result of the mobile data, which can more quickly determine whether the terminal is in the target area, reducing calculation time and cost.

[0064] The reference elements of the target area refer to the semantic elements collected when constructing the reference standard data. For example, the reference standard data can be a regional map or a spatial model. The reference standard data corresponding to different terminals may be different. For example, for an autonomous driving vehicle, the corresponding reference standard data may be a city map, and for a logistics robot, the corresponding reference standard data may be a spatial stereo model.

[0065] Reference elements include reference strong semantic elements and reference weak semantic elements.

[0066] Reference strong semantic elements refer to those elements with clear semantic features in the mobile area that are collected when constructing reference standard data. For example, reference strong semantic elements can be speed limit numbers on the road surface, route sign frames, text in route sign frames, traffic light frames, etc. in images collected when constructing reference standard data.

[0067] In some embodiments, the processor may use model processing, image recognition and other technologies to process the data collected when constructing the reference standard data to obtain reference strong semantic elements. When constructing the reference standard data, reference elements are collected based on a richer method than collecting mobile data, for example, multiple collections or using more accurate equipment (such as high-pixel cameras, etc.) for collection, so as to improve the accuracy and comprehensiveness of the reference standard data.

[0068] In some embodiments, based on different forms of the reference strong semantic elements, the reference strong semantic elements may be divided into different types. In some embodiments, the reference strong semantic elements may be divided into texture sub-elements and geometry sub-elements.

[0069] For example, digital or textual elements in images and / or videos collected when constructing reference standard data can be used as texture sub-elements; for another example, geometric elements such as lane lines and stop lines in images and / or videos collected when constructing reference standard data can be used as geometry sub-elements. The content of texture sub-elements and geometry sub-elements of reference strong semantic elements is similar to that of measured strong semantic elements. For more information, please refer to the relevant instructions for reference strong semantic elements and Figure 5 The corresponding content.

[0070] Reference weak semantic elements refer to abstract element objects collected when constructing reference standard data that can constrain positioning. For example, reference weak semantic elements can be fences, buildings in the middle of the road, corners and edges or faces of billboards in point cloud data or images and videos collected when constructing reference standard data.

[0071] In some embodiments, the processor may use model processing, image recognition and other technologies to process the data collected when constructing the reference standard data to obtain reference weak semantic elements.

[0072] In some embodiments, based on different forms of reference weak semantic elements, different types of divisions can be performed on the reference weak semantic elements. In some embodiments, the reference weak semantic elements can be divided into at least one of line elements and surface elements. For example, line features without specific semantics in images collected when constructing reference standard data can be used as line elements; for another example, surface features without specific semantics in images collected when constructing reference standard data can be used as surface elements. The content of line elements and surface elements of reference weak semantic elements is similar to that of line elements and surface elements of measured weak semantic elements. For more information, please refer to the relevant instructions on reference weak semantic elements and Figure 5 The corresponding content.

[0073] Policy assistance information refers to reference information used to formulate a terminal's mobility control policy. For example, policy assistance information may include precise positioning information of the terminal's current mobility area, attribute information of the current mobility area, and information about adjacent areas of the current mobility area.

[0074] In some embodiments, the target area includes a plurality of sub-areas. A sub-area refers to an area obtained by further dividing the target area. The division of the sub-areas can be performed according to various criteria such as geographical location. For example, when the target area is a highway, different sub-areas can be divided according to different road sections. For another example, when the target area is a factory building, different sub-areas can be divided according to different processing workshops.

[0075] In some embodiments, the policy assistance information includes a target sub-area to which the terminal belongs. The target sub-area refers to a sub-area in which the terminal is currently located among multiple sub-areas included in the target area.

[0076] In some embodiments, the processor may match the measured element with the reference element of the target area to obtain a corresponding matching result. The matching result of the measured element with the reference element of the target area refers to the degree of conformity between the measured element and the reference element of the target area.

[0077] In some embodiments, the processor can use a variety of methods to obtain the matching results of the measured elements and the reference elements of the target area. For example, the processor can express the matching results of the measured elements and the reference elements of multiple sub-areas based on the intersection-and-union ratio, find the sub-area corresponding to the reference element with the largest intersection-and-union ratio of the measured elements, and determine it as the target sub-area, and then determine the policy auxiliary information. For example, the location information of the target sub-area is used as the policy auxiliary information. Among them, the intersection-and-union ratio refers to the ratio of the number of elements in the intersection of two sets to the number of elements in the union.

[0078] In some embodiments, the processor may also determine the target sub-region based on the final evaluation values ​​of the measured elements and the reference elements of each sub-region. For more information about the final evaluation values, see Figure 4 and its related contents.

[0079] Step 330: Determine the mobility control policy of the terminal based on the policy auxiliary information.

[0080] The mobile control strategy refers to the driving strategy of the terminal during the movement. For example, the control strategy may include operations such as perception, positioning, prediction, decision-making, and planning related to the movement of the terminal.

[0081] In some embodiments, the processor can plan the movement route based on the strategy auxiliary information and determine the movement control strategy based on the actual road conditions. For example, the strategy auxiliary information indicates that the autonomous driving vehicle is in the parking space and the autonomous driving vehicle is in the reversing state. The processor can determine the reversing direction and reversing distance based on the strategy auxiliary information and the map. For another example, the strategy auxiliary information indicates that the transport robot is in the raw material workshop and the raw materials have been loaded. Based on the current production plan, the transport robot needs to transport the raw materials to the injection molding workshop. The processor can plan the subsequent transportation route of the transport robot based on the pre-downloaded factory map.

[0082] In some embodiments of this specification, by matching the measured elements with the reference elements of the target area, the mobile state and situation of the terminal in the target area can be quickly and accurately identified and judged, thereby providing corresponding strategy auxiliary information. At the same time, dividing the measured elements and the reference elements into strong and weak elements can reduce the difficulty of extraction, and by complementing and comparing the strong and weak elements, the robustness of positioning can be improved, thereby improving the accuracy of the mobile control strategy.

[0083] It should be noted that the above description of the relevant process is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the process under the guidance of this specification. However, these modifications and changes are still within the scope of this specification.

[0084] Figure 4 This is an exemplary schematic diagram of determining strategy auxiliary information according to some embodiments of this specification.

[0085] In some embodiments, for each sub-region, the processor may match the measured strong semantic element 410 with the reference strong semantic element 420 of the sub-region to determine a first evaluation value 450 of the sub-region; match the measured weak semantic element 430 with the reference weak semantic element 440 of the sub-region to determine a second evaluation value 460 of the sub-region; determine a final evaluation value 470 of the sub-region based on the first evaluation value 450 and the second evaluation value 460; determine the sub-region with the highest final evaluation value 470 as the target sub-region 480; and determine the policy auxiliary information 490 based on the target sub-region 480.

[0086] In some embodiments, the processor may match the measured strong semantic element 410 with the reference strong semantic element 420 of the sub-region to determine the corresponding matching degree; and match the measured weak semantic element 430 with the reference weak semantic element 440 of the sub-region to determine the corresponding matching degree. The matching degree refers to the degree of conformity.

[0087] In some embodiments, the processor can determine the degree of matching based on a variety of methods. Taking the determination of the degree of matching between the measured strong semantic element 410 and the reference strong semantic element 420 of the sub-region as an example, the processor can construct a measured strong semantic feature vector based on the number of elements of each subdivision category of the strong semantic element contained in the measured strong semantic element, and correspondingly, the processor can construct a reference strong semantic feature vector based on the number of elements of each subdivision category of the strong semantic element contained in the reference strong semantic element 420 of the sub-region, and the processor can further calculate the vector distance (such as a cosine vector, etc.) between the measured strong semantic feature vector and the reference strong semantic feature vector, and determine the degree of matching based on the vector distance, such as the larger the vector distance, the smaller the degree of matching.

[0088] For more information about the various subcategories of strong semantic elements, see Figure 5 and its related description.

[0089] The matching degree between the measured weak semantic element 430 and the reference weak semantic element 440 of the sub-region may be determined in the same manner as determining the matching degree between the measured strong semantic element 410 and the reference strong semantic element 420 of the sub-region.

[0090] The first evaluation value 450 of the sub-region refers to a quantitative value of the degree of matching between the measured strong semantic element 410 and the reference strong semantic element 420 of the sub-region. In some embodiments, the processor may preset first evaluation values ​​corresponding to different degrees of matching, and determine the first evaluation value 450 by looking up a table or the like based on the degree of matching between the measured strong semantic element 410 and the reference strong semantic element 420 of the sub-region.

[0091] The second evaluation value 460 of the sub-region refers to a quantitative value of the degree of matching between the measured weak semantic element 430 and the reference weak semantic element 440 of the sub-region. The method of determining the second evaluation value based on the degree of matching between the measured weak semantic element 430 and the reference weak semantic element 440 of the sub-region can refer to the method of determining the first evaluation value described above.

[0092] The final evaluation value 470 of the sub-region refers to a quantitative value of the degree of matching between the measured element and the reference element of the sub-region. In some embodiments, the processor can determine the final evaluation value 470 based on the first evaluation value 450 and the second evaluation value 460 in a variety of ways. For example, the final evaluation value 470 can be the average of the first evaluation value 450 and the second evaluation value 460, or the final evaluation value 470 is the maximum value of the first evaluation value 450 and the second evaluation value 460.

[0093] In some embodiments, the final evaluation value 470 is a weighted value of the first evaluation value 450 and the second evaluation value 460; the weight of the first evaluation value 450 is positively correlated to the richness of the reference strong semantic elements of the sub-region; the weight of the second evaluation value 460 is positively correlated to the richness of the reference weak semantic elements of the sub-region.

[0094] The richness of the reference strong semantic elements of the sub-region refers to the richness of the information contained in the reference strong semantic elements of the sub-region. In some embodiments, the processor can calculate the richness of the reference strong semantic elements of the sub-region based on the strong semantic elements collected in the sub-region when constructing the reference standard data, and pre-store it in the map.

[0095] In some embodiments, the processor may calculate the richness of the reference strong semantic elements of the sub-region based on a variety of methods. For example, the processor may use the product of the average number of each type of strong semantic elements collected in the sub-region when constructing the reference standard data and the number of types of the collected strong semantic elements as the richness of the reference strong semantic elements of the sub-region.

[0096] In some embodiments, the richness of the reference strong semantic elements can be determined based on at least one of the number of types of reference strong semantic elements, the number of elements of at least one type of reference strong semantic elements, and the semantic coverage of at least one type of reference strong semantic elements.

[0097] The number of types of reference strong semantic elements is related to the type classification of reference strong semantic elements. In some embodiments, the processor can further classify strong semantic elements (i.e., the collective term for reference strong semantic elements and measured strong semantic elements, the same below) and weak semantic elements (i.e., the collective term for reference weak semantic elements and measured weak semantic elements, the same below).

[0098] Figure 5 This is an exemplary schematic diagram of the division of reference element types according to some embodiments of the present specification.

[0099] In some embodiments, the reference element 510 includes a reference strong semantic element 420 and a reference weak semantic element 440. In some embodiments, the reference strong semantic element 420 includes at least one of a texture sub-element 520 and a geometry sub-element 530. The reference weak semantic element 440 includes at least one of a line element 540 and a surface element 550.

[0100] Figure 5 The reference element classification is only illustrated as an example. As mentioned above, the measured elements also include measured strong semantic elements and / or measured weak semantic elements, and the measured strong semantic elements also include at least one of texture sub-elements and geometric sub-elements. The measured weak semantic elements include at least one of line elements and surface elements.

[0101] The difference between measured elements and reference elements is that measured elements are elements obtained based on the environmental data actually obtained online during the movement of the terminal, while reference elements are elements obtained based on the processing of field environmental data when the reference standard data is constructed in advance (for example, offline). The following uses the classification of reference elements as an example for explanation, and the classification of measured elements is the same as that of reference elements.

[0102] The texture sub-element 520 refers to numbers or words in the environment. For example, the texture sub-element 520 can be speed limit numbers on the ground, height limit numbers in tunnels, vehicle spacing numbers on highways, "bus lane" words on bus lanes, and words on various signboards in the environment. The numbers or words in the texture sub-element have clear semantic information.

[0103] In some embodiments, the processor may further divide the texture sub-element 520 into various sub-categories. For example, the processor may treat the texture sub-element 520 as a first-level category, which includes multiple second-level categories, such as a second-level category including a digital texture category and a text texture category. In some embodiments, the processor may further divide the second-level categories of the texture sub-element to obtain one or more third-level categories corresponding to each second-level category. For example, the text texture category may include multiple third-level categories, such as a landmark text category, a warning text category, etc. The above division description is only an example, and the specific division can be determined based on actual conditions.

[0104] The geometric sub-element 530 refers to a geometric element with a clear meaning in the environment, for example, a lane line, a stop line, an arrow, a crosswalk, a pole, a sign frame, a traffic light frame, etc.

[0105] In some embodiments, the processor may further divide the geometry sub-element 530 into sub-categories. For example, the processor may treat the geometry sub-element 530 as a primary category, which includes multiple secondary categories, such as a secondary category including a line geometry category and a polygon geometry category. In some embodiments, the processor may further divide the secondary categories of the geometry sub-element to obtain one or more tertiary categories corresponding to each secondary category. For example, the line geometry category may include multiple tertiary categories, such as a landmark line category, a rod line category, etc. The above division description is only an example, and the specific division can be determined based on actual conditions.

[0106] Line elements 540 refer to straight lines detected in an image and / or line features detected in point cloud data based on curvature or intensity values. For example, line elements 540 may include corners and edges of road surface elements and roadside elements, such as corners of buildings, edges of billboards, local contours of street lamps, and the like.

[0107] The surface element 550 refers to a plane or curved surface detected in an image, and / or a surface-like feature detected in point cloud data. For example, the surface element 550 may include the surface of a road surface element and a roadside element, such as the surface of a building, a billboard, a fence, etc.

[0108] In some embodiments, the processor may also perform multi-level division of line elements and surface elements in accordance with the aforementioned multi-level division of texture sub-elements and geometric sub-elements. For example, the surface elements may be subdivided into wall surface elements, billboard surface elements, etc. The specific division method may be set based on actual conditions.

[0109] The number of reference strong semantic elements refers to the number of reference strong semantic elements collected in the sub-region environment when constructing the reference standard data. For example, if the strong semantic elements collected in a sub-region include the warning text of "accident-prone road ahead", the text of "bus lane" and the crosswalk when constructing the reference standard data, the number of reference strong semantic elements in the sub-region is 3, which are warning text, landmark text and landmark line.

[0110] The number of elements of a type of reference strong semantic elements refers to the number of reference strong semantic elements contained in the type corresponding to each level of the reference strong semantic elements. Continuing with the previous example, the strong semantic elements collected in a sub-area include the text "accident-prone road ahead", the text "bus lane" and the crosswalk; then the number of elements of the warning text class under the third-level category of the sub-area is 8, the number of elements of the landmark text class is 6, and the number of elements of the landmark line class is the number of horizontal lines of the crosswalk; the number of elements of the text texture class under the second-level category of the sub-area is 14, and the number of elements of the line geometry class is the number of horizontal lines of the crosswalk; the number of elements of the texture class under the first-level category of the sub-area is 14, and the number of elements of the geometry class is the number of horizontal lines of the crosswalk.

[0111] The semantic coverage of a type of reference strong semantic elements refers to the degree of consistency between the information contained in the reference strong semantic elements of a type and the information that the reference strong semantic elements of this type should contain under ideal circumstances.

[0112] The ideal reference strong semantic elements can refer to the reference strong semantic elements of the corresponding type that can be collected in the corresponding sub-area under ideal collection conditions (such as sufficient equipment, high equipment accuracy and good collection environment). For example, a road is 100 meters long, 10 meters wide, and has three lanes corresponding to four lane lines. The ideal reference strong semantic elements should include at least four landmark line strong semantic elements corresponding to the four lane lines with a length of 100 meters.

[0113] In some embodiments, the processor may calculate the semantic coverage of each type of reference strong semantic elements based on a variety of methods. For example, the processor may use the similarity between each type of reference strong semantic elements actually obtained and the reference strong semantic elements in an ideal situation as the semantic coverage of each type of reference strong semantic elements. For another example, the processor may use the ratio of the number of each type of reference strong semantic elements actually obtained to the number of reference strong semantic elements in an ideal situation as the semantic coverage of each type of reference strong semantic elements.

[0114] In some embodiments, the processor can determine the richness of the reference strong semantic elements in a variety of ways based on the number of types of reference strong semantic elements, the number of elements of each type of reference strong semantic elements, and the semantic coverage of each reference strong semantic element. For example, the processor can perform a weighted summation of the sum of the product of the number of elements of each type of reference strong semantic elements contained in the reference strong semantic element and the semantic coverage of each type of reference strong semantic elements and the number of types of the reference strong semantic elements, and use the value of the weighted summation as the richness of the reference strong semantic elements of the reference strong semantic elements. The weight can be obtained by preset.

[0115] In some embodiments, the processor may pre-save normalized values ​​corresponding to different weighted sum values, and the processor may normalize the weighted sum value to a value between 0 and 1 by looking up a table or the like.

[0116] In some embodiments, the richness of the reference strong semantic element may be determined based on formula (1):

[0117] Among them, E is the richness of the reference strong semantic elements, n is the number of reference strong semantic elements, y i is the number of elements in the reference strong semantic element category i, k i The weight corresponding to the reference strong semantic element type i can be a percentage, and the weight can be determined based on a variety of methods. For example, the weight corresponding to the reference strong semantic element type i is positively correlated with the semantic coverage of the reference strong semantic element type i. For another example, the technician can preset the weights corresponding to different types of reference strong semantic elements in advance.

[0118] In some embodiments of the present specification, the richness of the reference strong semantic elements can be the weighted sum of the number of elements in each type of reference strong semantic elements, so that different types of strong semantic elements contribute to the final positioning to different degrees. Subdividing the types of different reference strong semantic elements can improve the reliability of the richness of the reference strong semantic elements.

[0119] In some embodiments of the present specification, the richness of reference strong semantic elements is determined based on at least one of the number of types of reference strong semantic elements, the number of elements of at least one type of reference strong semantic elements, and the semantic coverage of at least one type of reference strong semantic elements. A detailed and comprehensive calculation method is provided, thereby improving the accuracy and reliability of the calculation of the richness of reference strong semantic elements, having strong adaptability and being able to meet the needs of different application scenarios.

[0120] The richness of the reference weak semantic elements of the sub-region refers to the richness of the information contained in the reference weak semantic elements of the sub-region. In some embodiments, the processor can calculate the reference weak semantic richness of the sub-region based on the weak semantic elements collected in the sub-region when constructing the reference standard data, and pre-store it in the reference standard data.

[0121] In some embodiments, the processor may calculate the richness of the reference weak semantic elements of the sub-region based on a variety of methods. For example, the processor may use the product of the average number of each type of weak semantic elements and the number of types of weak semantic elements collected in the sub-region when constructing the reference standard data as the richness of the reference weak semantic elements of the sub-region.

[0122] In some embodiments, the richness of the reference weak semantic elements may be determined based on at least one of the number of categories of the reference weak semantic elements, the number of at least one category of reference weak semantic elements, and the confidence of at least one category of reference weak semantic elements.

[0123] The number of reference weak semantic element categories refers to the number of types of reference weak semantic elements collected in the sub-region environment when constructing the reference standard data. If the weak semantic elements collected in a sub-region include the edge of a billboard and the local outline of a street lamp when constructing the reference standard data, the number of reference weak semantic element categories in the sub-region is 1, and the category is line element.

[0124] The number of a class of reference weak semantic elements refers to the number of reference weak semantics contained in the corresponding class. Continuing with the previous example, the weak semantic elements collected in a sub-region include the edge of a billboard and the local outline of a street lamp, so the number of line elements in this sub-region is 2.

[0125] Confidence refers to the confidence level of the detected semantic elements. In some embodiments, the confidence level of the reference weak semantic elements can be obtained when the reference weak semantic elements are identified based on the image detection algorithm. For example, after the processor detects the laser point cloud data, it outputs the confidence level based on information such as the point cloud changes and texture refinement of the image.

[0126] The confidence of at least one type of reference weak semantic elements may include reference line element confidence and reference surface element confidence. In some embodiments, the reference line element confidence and the reference surface element confidence may be determined based on a confidence model. For more information about the confidence model, see Figure 6 and its related description.

[0127] In some embodiments, the processor may determine the richness of the reference weak semantic elements based on multiple methods. For example, the richness of the reference weak semantic elements may be the product of the number of reference weak semantic elements in each category and the average number of reference weak semantic elements in each category.

[0128] In some embodiments, the richness of the reference weak semantic elements may also be determined based on formula (2): F=m1×z1+m2×z2 (2)

[0129] Among them, F is the richness of the reference weak semantic elements, m1 is the reference line element confidence, z1 is the number of line elements, m2 is the reference surface element confidence, and z2 is the number of surface elements. For more information about the reference line element confidence and reference surface element confidence, see Figure 6 and its related description.

[0130] In some embodiments of the present specification, the richness of the reference weak semantic elements may be the product of the reference line element confidence and the number of line elements, and the sum of the product of the reference surface element confidence and the number of surface elements. This may more comprehensively consider various factors and improve the reliability of the richness of the reference weak semantic elements.

[0131] In some embodiments of the present specification, the richness of reference weak semantic elements is determined based on at least one of the number of categories of reference weak semantic elements, the number of at least one category of reference weak semantic elements, and the confidence of at least one category of reference weak semantic elements, providing a detailed and comprehensive calculation method that can be adaptively adjusted for different scenarios to further improve the reliability of the calculation.

[0132] In some embodiments of the present specification, different weights are dynamically assigned to different semantic matching degrees according to the richness. The richness of strong semantic elements and the richness of weak semantic elements in the scene richness can be weighed and used to achieve robust positioning in scenes with different scene richness.

[0133] In some embodiments, the final evaluation value 470 is a weighted value of the first evaluation value 450 and the second evaluation value 460; the weight of the first evaluation value 450 is positively correlated to the degree of match between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-area; the weight of the second evaluation value 460 is positively correlated to the degree of match between the richness of the measured weak semantic elements and the richness of the reference weak semantic elements of the sub-area.

[0134] The richness of the measured strong semantic elements refers to the richness of the information contained in the measured strong semantic elements. In some embodiments, the processor can calculate the richness of the measured strong semantic elements at the corresponding position based on the situation of the strong semantic elements obtained by the terminal when moving in real time. The process of calculating the richness of the measured strong semantic elements is similar to the process of calculating the richness of the reference strong semantic elements, which can be seen above.

[0135] The richness of the measured weak semantic elements refers to the richness of the information contained in the measured weak semantic elements. In some embodiments, the processor can calculate the richness of the measured weak semantic elements at the corresponding position based on the situation of the weak semantic elements obtained by the terminal when moving in real time. The process of calculating the measured weak semantic richness is similar to the process of calculating the richness of the reference weak semantic elements, which can be seen above.

[0136] In some embodiments, the degree of matching between the richness of the measured strong semantic element and the richness of the reference strong semantic element of the sub-region refers to the degree of similarity between the two richnesses.

[0137] In some embodiments, the processor may determine the degree of match between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region based on a variety of methods. For example, the degree of match between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region may be the ratio of the richness of the measured strong semantic elements to the richness of the reference strong semantic elements of the sub-region.

[0138] In some embodiments, the processor may use formula (3) to determine the degree of match between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region:

[0139] Among them, G is the matching degree between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region, H is the richness of the measured strong semantic elements, and E is the richness of the reference strong semantic elements of the sub-region.

[0140] The matching degree between the richness of the measured weak semantic elements and the richness of the reference weak semantic elements of the sub-region is determined similarly to the matching degree between the richness of the measured strong semantic elements and the richness of the reference strong semantic elements of the sub-region. For details, please refer to the above content.

[0141] In some embodiments of the present specification, the first evaluation value and the second evaluation value are used as indicators for evaluating the accuracy of policy auxiliary information. The first evaluation value and the second evaluation value are positively correlated to the degree of matching between the richness of the corresponding measured semantic elements and the richness of the reference semantic elements. When determining the policy auxiliary information, priority is given to the scenario when constructing the reference standard data, which can further improve the accuracy of determining the target sub-area.

[0142] In some embodiments, when the processor determines the final evaluation value based on the weighted value of the first evaluation value and the second evaluation value, the weight corresponding to the first evaluation value may be the first weight, and the weight corresponding to the second evaluation value may be the second weight.

[0143] In some embodiments, the processor may preset a first initial weight and a second initial weight; determine a first difference based on the richness of measured strong semantic elements and the richness of reference strong semantic elements of the sub-area; determine a second difference based on the richness of measured weak semantic elements and the richness of reference weak semantic elements of the sub-area; determine a target difference based on the first difference and the second difference; determine a target difference degree based on the target difference based on a first preset algorithm; determine a weight of the richness of the measured semantic elements corresponding to the target difference based on a second preset algorithm; determine a second weight and / or the first weight based on the weight of the richness of the measured semantic elements corresponding to the target difference.

[0144] The first initial weight refers to the initial weight of the first evaluation value. The second initial weight refers to the initial weight of the second evaluation value. In some embodiments, the first initial weight and the second initial weight can be preset by a technician based on historical experience.

[0145] The first difference refers to the difference between the richness of the measured strong semantic element and the richness of the reference strong semantic element of the sub-region. In some embodiments, the first difference can be the difference between the richness of the measured strong semantic element and the richness of the reference strong semantic element of the sub-region.

[0146] The second difference refers to the difference between the richness of the measured weak semantic element and the richness of the reference weak semantic element of the sub-region. In some embodiments, the second difference can be the difference between the richness of the measured weak semantic element and the richness of the reference weak semantic element of the sub-region.

[0147] The target difference refers to the difference used to calculate the target difference degree. In some embodiments, the target difference can be the maximum difference between the first difference and the second difference.

[0148] In some embodiments, the target difference degree may be determined based on a first preset algorithm.

[0149] In some embodiments, the first preset algorithm may be formula (4): Among them, C is the target difference degree, A is the first difference, B is the second difference, z1 is the reference strong semantic richness of the sub-region, and z2 is the reference weak semantic richness of the sub-region. For an explanation of how to determine the richness of the reference strong semantic elements and the richness of the reference weak semantic elements, see above.

[0150] The weight of the richness of the measured semantic elements corresponding to the target difference refers to the first weight or the second weight. For example, when the target difference is the first difference, the weight of the richness of the measured semantic elements corresponding to the target difference is the first weight, and when the target difference is the second difference, the weight of the richness of the measured semantic elements corresponding to the target difference is the second weight.

[0151] In some embodiments, the weight of the richness of the measured semantic elements corresponding to the target difference may be determined based on a second preset algorithm.

[0152] In some embodiments, the second preset algorithm may be formula (5): Where D is the weight of the richness of the measured semantic elements corresponding to the target difference, x is the initial weight corresponding to the target difference, and C is the degree of the target difference. When the target difference is the first difference, the initial weight x corresponding to the target difference is the first initial weight; when the target difference is the second difference, the initial weight x corresponding to the target difference is the second initial weight.

[0153] In some embodiments, the sum of the first weight and the second weight may be 1. After determining any one of the two weights based on Formula 5, the value of the other weight may be obtained.

[0154] Some embodiments of the present specification provide an effective way to process and analyze the richness of semantic elements to determine the weights of the first evaluation value and the second evaluation value, thereby significantly improving the efficiency of determining the final evaluation value and, at the same time, enhancing the accuracy of the final evaluation value.

[0155] In some embodiments, the processor may determine the sub-region with the largest final evaluation value as the target sub-region 480 .

[0156] In some embodiments, the processor may determine the target sub-region 480 as the policy assistance information 490. For more information about the target sub-region 480 and the policy assistance information 490, see Figure 3 and its related contents.

[0157] In some embodiments of the present specification, each sub-region is evaluated based on the degree of matching between the measured and reference strong semantic elements and weak semantic elements, and the sub-region with the highest final evaluation value is determined as the target sub-region, so as to achieve accurate and efficient positioning of the vehicle.

[0158] In some embodiments, as the measured elements of each sub-region are accumulated, the processor may update the reference elements of the sub-region based on the measured elements of the sub-region.

[0159] In some embodiments, in response to the measured element satisfying a preset condition, the processor may update the reference element of the target area based on the measured element.

[0160] The preset condition refers to the condition that the measured element needs to meet when updating the reference element. For example, the preset condition can be that the same measured element is obtained in the same target sub-area for multiple consecutive times (such as the number of times exceeds the preset threshold), and the element content corresponding to the measured element does not exist in the reference element of the target sub-area. The preset threshold in the preset condition can be obtained based on manual settings.

[0161] In some embodiments, in response to the measured element satisfying a preset condition, the processor may add the measured element to the reference element of the corresponding category of the target sub-area. For example, if the terminal obtains a stop line at the same intersection for multiple consecutive times, and the stop line does not exist in the reference strong semantic element corresponding to the intersection, it means that after the reference standard data corresponding to the intersection is constructed, a new stop line is added to the intersection. The processor may add the reference strong semantic element of the landmark line class corresponding to the stop line to the reference strong semantic element corresponding to the intersection in the reference standard data.

[0162] In some embodiments of the present specification, by updating the reference elements of the target area based on the measured elements, the reference elements can be corrected and adjusted in a timely manner, thereby improving the accuracy and reliability of the reference elements and providing more reliable data support for subsequent decision-making and control.

[0163] Figure 6 is an exemplary schematic diagram of a confidence model according to some embodiments of the present specification.

[0164] In some embodiments, the processor may determine a reference line feature confidence 650 and a reference surface feature confidence 660 based on the confidence model 640 .

[0165] The confidence model 640 may be a machine learning model. For example, the confidence model 640 may be a deep neural network (DNN) model, a convolutional neural network (CNN) model, a recurrent neural network (RNN) model, or any combination thereof.

[0166] In some embodiments, the input of the confidence model 640 may include weather data 610 , terrain data 620 of the sub-area, and building layout data 630 of the sub-area, and the output may be reference line element confidence 650 and reference surface element confidence 660 .

[0167] Weather data 610 may characterize environment and weather information. For example, weather data 610 may include temperature, humidity, rainfall, etc. In some embodiments, weather data 610 may be obtained based on a meteorological website.

[0168] In some embodiments, the terrain data 620 of the sub-region and the building layout data 630 of the sub-region may be acquired through three-dimensional modeling based on map data acquired from a navigation website.

[0169] In some embodiments, the confidence model can be trained based on labeled training samples. Multiple labeled training samples can be input into the initial confidence model, and a loss function can be constructed by using the labels and the results of the initial confidence model, and the parameters of the initial confidence model can be iteratively updated based on the loss function. When the loss function of the initial confidence model meets the preset conditions, the model training is completed, and a trained confidence model is obtained. The preset conditions may be that the loss function converges, the number of iterations reaches a threshold, etc.

[0170] In some embodiments, the training samples may include sample weather data, sample sub-region terrain data, and sample building layout data, and the labels may be sample reference line element confidence and sample reference surface element confidence.

[0171] In some embodiments, the training samples can be obtained based on historical data, and the labels can be obtained based on the accuracy of the line category features and the surface category features extracted from the images captured in the sample sub-area under the sample weather data, or from the laser point cloud data detected from the laser sensor. For example, if 5 lines and 2 surfaces are extracted from the environment in history, and there are indeed 5 lines and 2 surfaces in the actual environment, then the line category confidence label is 1, and the surface category confidence label is also 1; if no lines or surfaces are extracted, then the line category confidence label is 0, and the surface category confidence label is also 0; if the extraction of lines and surfaces is not comprehensive (for example, 2 lines and 1 surface), then the accuracy of the line category feature is calculated: 2÷5×100%=40%, then the line category confidence label is 40%; the accuracy of the surface category feature is calculated: 1÷2×100%=50%, then the surface category confidence label is 50%.

[0172] In some embodiments of the present specification, weather data and terrain and building layout data are processed through a confidence model. The self-learning ability of the machine learning model can be used to find patterns from a large amount of terrain and building layout data, and the correlation between the reference line element confidence, the reference surface element confidence and the terrain and building layout data is obtained, thereby improving the efficiency of determining the reference line element confidence and the reference surface element confidence, and at the same time, improving the accuracy of the reference line element confidence and the reference surface element confidence.

[0173] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0174] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0175] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0176] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0177] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0178] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0179] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A method for determining a mobile control strategy, characterized in that: The method comprises: Determine measured elements based on the mobile data of the terminal; the measured elements include measured strong semantic elements and measured weak semantic elements; Matching the measured elements with reference elements of the target area to determine strategy auxiliary information; the reference elements include reference strong semantic elements and reference weak semantic elements; wherein the measured strong semantic elements and / or the reference strong semantic elements include element objects with clear semantic features, and the measured weak semantic elements and / or the reference weak semantic elements include abstract element objects that have a constraining effect on positioning; Based on the policy auxiliary information, a mobility control policy of the terminal is determined.

2. The method according to claim 1, characterized in that The target area includes a plurality of sub-areas; and matching the measured elements with reference elements of the target area to determine the strategy auxiliary information includes: For each of the sub-regions, Matching the measured strong semantic element with the reference strong semantic element of the sub-region to determine a first evaluation value of the sub-region; Matching the measured weak semantic element with the reference weak semantic element of the sub-region to determine a second evaluation value of the sub-region; Determining a final evaluation value of the sub-region based on the first evaluation value and the second evaluation value; Determine the sub-region with the highest final evaluation value as the target sub-region; Based on the target sub-region, the policy auxiliary information is determined.

3. The method according to claim 2, characterized in that The final evaluation value is a weighted value of the first evaluation value and the second evaluation value; The weight of the first evaluation value is positively correlated to the richness of the reference strong semantic element of the sub-region; The weight of the second evaluation value is positively correlated to the richness of the reference weak semantic element of the sub-region.

4. The method according to claim 2, characterized in that: The final evaluation value is a weighted value of the first evaluation value and the second evaluation value; The weight of the first evaluation value is positively correlated to the matching degree between the richness of the measured strong semantic element and the richness of the reference strong semantic element of the sub-region; The weight of the second evaluation value is positively correlated to the degree of matching between the richness of the measured weak semantic element and the richness of the reference weak semantic element of the sub-region.

5. The method according to claim 3 or 4, characterized in that: The richness of the reference strong semantic elements is determined based on at least one of the number of types of the reference strong semantic elements, the number of elements of the at least one type of reference strong semantic elements, and the semantic coverage of the at least one type of reference strong semantic elements.

6. The method according to claim 3 or 4, characterized in that: The richness of the reference weak semantic elements is determined based on at least one of the number of categories of the reference weak semantic elements, the number of the at least one category of reference weak semantic elements, and the confidence of the at least one category of reference weak semantic elements.

7. The method according to claim 1, characterized in that The target area is determined based on a coarse-grained spatial matching result of the movement data.

8. The method according to claim 1, characterized in that The method further comprises: In response to the measured element satisfying a preset condition, the reference element of the target area is updated based on the measured element.

9. The method according to claim 1, characterized in that: The reference strong semantic element and / or the measured strong semantic element includes at least one of a texture sub-element and a geometry sub-element.

10. The method according to claim 1, characterized in that The reference weak semantic elements and / or the measured weak semantic elements include at least one of line elements and surface elements.

11. An automatic driving control system, characterized in that: The system comprises: An element determination module, used to determine measured elements based on the mobile data of the terminal; the measured elements include measured strong semantic elements and measured weak semantic elements; An information determination module is used to match the measured elements with reference elements of the target area to determine the strategy auxiliary information; the reference elements include reference strong semantic elements and reference weak semantic elements; wherein the measured strong semantic elements and / or the reference strong semantic elements include element objects with clear semantic features, and the measured weak semantic elements and / or the reference weak semantic elements include abstract element objects that have a constraining effect on positioning; A policy determination module is used to determine the mobility control policy of the terminal based on the policy auxiliary information.

12. A mobile control strategy determination device, characterized in that: The device comprises: at least one storage medium storing computer instructions; At least one processor executes the computer instructions to implement the mobile control strategy determination method according to any one of claims 1 to 10.

13. A computer-readable storage medium storing computer instructions, wherein when a computer reads the computer instructions, the computer executes the mobile control strategy determination method according to any one of claims 1 to 10.

Images