Map construction method and device, electronic equipment and vehicle

Abstract

The invention provides a map construction method and device, electronic equipment and a vehicle. The map construction method comprises the following steps: fusing and acquiring multiple groups of positioning and motion data of a vehicle by using a fusion positioning algorithm to obtain a vehicle pose of the vehicle in a world coordinate system; target information of the target object is acquired based on the acquired image by using a deep learning model, and the target information comprises coordinates of preset feature points of the target object in a vehicle coordinate system; according to the vehicle pose and the target information, obtaining a target pose of the target object in a world coordinate system and a size parameter of the target object, and taking the target pose and the size parameter as observation information of the target object; and on the basis of the observation information, filtering and updating the state information of the target object to obtain updated state information of the target object, and constructing a map. According to the method and the device, the mapping precision and the mapping efficiency are improved, real-time mapping at the vehicle end is realized, the hardware cost and the labor cost of mapping are reduced, and the influence of human factors is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and in particular to a map construction method, device, electronic equipment and vehicle. Background Art

[0002] At present, intelligent driving technology is widely used in various vehicles. In intelligent driving technology, the construction technology of high-precision maps is directly related to the safety, accuracy and real-time performance of the autonomous driving system.

[0003] The traditional method of building high-precision maps mainly uses high-precision laser radar as the main means of data collection. The point cloud data obtained by high-precision laser radar needs to go through complex computer processing steps, including point cloud denoising, feature extraction, object recognition and classification, etc. to extract object information before a preliminary three-dimensional model of the environment can be formed. In order to ensure the accuracy of the map, a large amount of manual participation is often required for fine annotation and verification. The traditional method of building high-precision maps relies on expensive high-precision sensors and requires a large amount of manual participation and offline processing, resulting in low efficiency and high cost of map construction, which is difficult to meet the urgent need for real-time map updates in intelligent driving technology. Summary of the invention

[0004] In view of this, the present application provides a map construction method, device, electronic device and vehicle to solve the technical problem that traditional high-precision map construction methods rely on expensive high-precision sensors, require a lot of manual participation and offline processing, resulting in low map construction efficiency and high cost, and are difficult to meet the urgent demand for real-time map updates in intelligent driving technology.

[0005] In a first aspect, an embodiment of the present application provides a map construction method, comprising:

[0006] The fusion positioning algorithm is used to fuse multiple sets of positioning and motion data of the vehicle to obtain the vehicle's position and posture in the world coordinate system;

[0007] Acquire target information of a target object based on the collected image using a deep learning model, wherein the target information includes coordinates of preset feature points of the target object in a vehicle coordinate system;

[0008] According to the vehicle posture and the target information, a target posture of the target object in the world coordinate system and a size parameter of the target object are acquired, and the target posture and the size parameter are used as observation information of the target object;

[0009] A map is constructed based on the updated state information of the target object obtained by filtering and updating the state information of the target object based on the observation information.

[0010] The map construction method provided in the above embodiment first uses a fusion positioning algorithm to fuse multiple sets of positioning and motion data of the vehicle to obtain the vehicle posture in the world coordinate system; further uses a deep learning model to obtain the target information of the target object based on the collected image, wherein the target information includes the coordinates of the preset feature points of the target object in the vehicle coordinate system; further obtains the target posture and size parameters of the target object in the world coordinate system according to the vehicle posture and target information, and uses the target posture and size parameters as the observation information of the target object; finally, based on the observation information, the state information of the target object is filtered and updated to obtain the updated state information of the target object, and the map is constructed. Based on this, the present application can obtain a more accurate and effective vehicle posture by fusing multiple sets of positioning and motion data collected by GNSS, IMU and inertial measurement unit, without the need to use high-precision sensors, reducing the high hardware cost; obtain the observation information of the target object according to the vehicle posture and target information, without the need for manual participation in annotation, reducing the influence of human factors, improving the efficiency and accuracy of mapping, and reducing labor costs; constructing a map based on the updated state information of the target object, realizing real-time mapping on the vehicle side, further improving the real-time and accuracy of mapping.

[0011] In some embodiments of the present application, the vehicle posture includes the coordinates of the vehicle in the world coordinate system, and acquiring the target posture of the target object in the world coordinate system and the size parameters of the target object according to the vehicle posture and the target information includes:

[0012] Constructing the target coordinate system corresponding to the target object based on the coordinates of the preset feature points in the vehicle coordinate system, and obtaining the coordinates of the preset feature points in the target coordinate system;

[0013] Calculating a target posture conversion relationship between the target coordinate system and the world coordinate system according to the vehicle posture, the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the target coordinate system;

[0014] Using the target pose conversion relationship, the coordinates of the target object in the target coordinate system are converted to coordinates in the world coordinate system to obtain the target pose;

[0015] The size parameter of the target object is calculated according to the coordinates of the preset feature point in the target coordinate system.

[0016] In some embodiments of the present application, calculating the target pose conversion relationship between the target coordinate system and the world coordinate system according to the vehicle pose, the coordinates of the preset feature point in the vehicle coordinate system, and the coordinates of the preset feature point in the target coordinate system includes:

[0017] Determine a first posture conversion relationship between the vehicle coordinate system and the world coordinate system according to the vehicle posture;

[0018] Calculating a second posture conversion relationship between the target coordinate system and the vehicle coordinate system according to the coordinates of the preset feature point in the target coordinate system and the coordinates of the preset feature point in the vehicle coordinate system;

[0019] The target posture conversion relationship between the target coordinate system and the world coordinate system is calculated according to the first posture conversion relationship and the second posture conversion relationship.

[0020] In some embodiments of the present application, the preset feature points include a plurality of contour points, and calculating the size parameter of the target object according to the coordinates of the preset feature points in the target coordinate system includes:

[0021] Determine a first parameter according to the coordinates of the first dimension of the plurality of contour points in the target coordinate system, wherein the first parameter includes an absolute value of a maximum difference between the coordinates of the first dimension of the plurality of contour points;

[0022] Determine a second parameter according to the coordinates of the second dimension of the plurality of contour points in the target coordinate system, wherein the second parameter includes an absolute value of a maximum difference between the coordinates of the second dimension of the plurality of contour points;

[0023] The size parameter is determined according to the first parameter and the second parameter.

[0024] In some embodiments of the present application, the map construction method further includes:

[0025] Based on the target posture conversion relationship and the size parameter, a state vector of a Kalman filter is constructed, and an observation vector is determined;

[0026] According to the state vector at the k-1th moment, predict the state vector at the kth moment, where k is a positive integer greater than 1;

[0027] According to the prediction error covariance matrix at the k-1th moment, predict the error covariance matrix at the kth moment;

[0028] Determine the Kalman filter gain at the kth moment according to the error covariance matrix and the observation noise covariance matrix at the kth moment;

[0029] According to the state vector at the kth moment, the Kalman filter gain and the observation vector, the updated state vector at the kth moment is determined, and the updated state information of the target object is obtained according to the updated state vector.

[0030] In some embodiments of the present application, the map construction method further includes:

[0031] The observation noise covariance matrix is ​​determined according to the positioning confidence of the fusion positioning algorithm, the detection confidence and the detection distance of the deep learning model.

[0032] In some embodiments of the present application, the map construction method further includes:

[0033] Determining whether the target object is located behind the vehicle according to the coordinates of the preset feature points of the target object in the vehicle coordinate system;

[0034] When the target object is located behind the vehicle, filtering and updating the state information of the target object is stopped, and the updated state information of the target object is output.

[0035] In the second aspect, an embodiment of the present application also provides a map construction device, including: a fusion positioning module, which is used to use a fusion positioning algorithm to fuse multiple sets of positioning and motion data of the vehicle to obtain the vehicle posture in the world coordinate system; a deep learning module, which is used to use a deep learning model to obtain target information of the target object based on the collected image, wherein the target information includes the coordinates of preset feature points of the target object in the vehicle coordinate system; an information acquisition module, which is used to obtain the target posture of the target object in the world coordinate system and the size parameters of the target object according to the vehicle posture and the target information, and use the target posture and the size parameters as the observation information of the target object; a map construction module, which is used to construct a map based on the updated status information of the target object obtained after filtering and updating the status information of the target object based on the observation information.

[0036] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a memory storing computer-readable instructions; and a processor executing the computer-readable instructions stored in the memory to implement the map construction method as described in the above embodiment.

[0037] In a fourth aspect, an embodiment of the present application further provides a vehicle, which includes the electronic device as described in the above embodiment.

[0038] It can be understood that the map construction device of the second aspect, the electronic device of the third aspect and the vehicle of the fourth aspect provided above all correspond to the map construction method of the first aspect. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding map construction method provided above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the steps of a map construction method provided in one embodiment of the present application.

[0040] Figure 2 It is a schematic diagram of the hardware structure of a vehicle provided in one embodiment of the present application.

[0041] Figure 3 This is a schematic diagram of the steps of a map construction method provided in yet another embodiment of the present application.

[0042] Figure 4 It is a schematic diagram of a target coordinate system provided in yet another embodiment of the present application.

[0043] Figure 5 This is a schematic diagram of the steps of a map construction method provided in yet another embodiment of the present application.

[0044] Figure 6 It is a schematic diagram of functional modules of a map construction device provided in one embodiment of the present application.

[0045] Main component symbols

[0046] Vehicle 100

[0047] GNSS 110

[0048] Inertial Measurement Unit 120

[0049] Wheel speed sensor 130

[0050] Image sensor 140

[0051] Electronic equipment 150

[0052] Memory 151

[0053] Processor 152

[0054] Map construction device 10

[0055] Fusion positioning module 11

[0056] Deep Learning Module 12

[0057] Information acquisition module 13

[0058] Map building module 14

[0059] Filter update module 15 DETAILED DESCRIPTION

[0060] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application is described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the implementation methods of the present application and the features in the implementation methods can be combined with each other without conflict.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present application. The described implementations are only part of the implementations of the present application, rather than all the implementations.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0063] It should be further noted that, in this article, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0064] In this application, "at least one" means one or more, and "more than one" means two or more than two. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and drawings of this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0065] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0066] See also Figure 1 , Figure 1 A schematic diagram of the steps of a map construction method provided in one embodiment of the present application.

[0067] The map construction method provided in the embodiment of the present application is applied to one or more electronic devices 150 (such as Figure 2 As shown in the figure, the electronic device 150 is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0068] In some embodiments of the present application, the vehicle 100 includes an electronic device 150. For example, the electronic device 150 may be an onboard device of the vehicle 100.

[0069] In other embodiments, the electronic device 150 may also be any electronic product that is communicatively connected to the vehicle 100, such as a personal computer, a tablet computer, etc. The electronic device 150 may interact with the user through a keyboard, a mouse, a remote control, a touch pad, or a voice control device.

[0070] In some embodiments of the present application, the network where the electronic device 150 is located includes but is not limited to: the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), etc.

[0071] Specifically, the map construction method includes the following steps. According to different requirements, the order of some steps in the step diagram can be changed, and some steps can be omitted.

[0072] Step S10, using a fusion positioning algorithm to fuse multiple sets of positioning and motion data of the vehicle to obtain the vehicle's position and posture in the world coordinate system.

[0073] Combination Figure 2As shown, in some embodiments of the present application, multiple sensors installed on the vehicle 100 can be used to collect multiple sets of positioning and motion data of the vehicle 100. The multiple sensors may include, but are not limited to, a global navigation satellite system 110 (Global Navigation Satellite System, GNSS), an inertial measurement unit 120 (Inertial Measurement Unit, IMU) and a wheel speed sensor 130. Specifically, the position information of the vehicle 100 in the world coordinate system, such as longitude, latitude, altitude, etc., is collected through the GNSS; the speed information of the vehicle 100, such as acceleration, heading angular velocity, steering angular velocity, is collected through the IMU; the wheel speed of the vehicle 100 is collected through the wheel speed sensor 130, for example, the wheel speed sensor 130 may be a magnetoelectric wheel speed sensor, a Hall wheel speed sensor, etc., and the present application does not impose any restrictions on this.

[0074] In some embodiments of the present application, the GNSS, IMU and inertial measurement unit 120 may collect data at a preset collection frequency, for example, collecting a set of data every 0.5 seconds.

[0075] In some embodiments of the present application, the fusion positioning algorithm may include but is not limited to a Kalman filter algorithm, a particle filter algorithm, and a more complex optimization algorithm. Based on the fusion positioning algorithm, the vehicle posture of the vehicle 100 in the world coordinate system can be obtained, and the vehicle posture includes the position (longitude, latitude, altitude) and attitude (pitch angle, roll angle, yaw angle) of the vehicle 100 in the world coordinate system.

[0076] Specifically, the step of obtaining the vehicle posture of the vehicle 100 in the world coordinate system may include: first, since the acquisition frequency and acquisition time of different sensors may be different, it is necessary to perform time synchronization processing on the multiple sets of positioning and motion data collected by the GNSS, IMU and inertial measurement unit 120 to ensure the temporal consistency of the positioning and motion data. Then, the multiple sets of positioning and motion data collected by the GNSS, IMU and inertial measurement unit 120 are filtered to remove noise and outliers. Then, the speed information collected by the IMU is used to perform short-term prediction of the vehicle posture. Then, the prediction result is corrected by combining the position information collected by the GNSS and the wheel speed information collected by the wheel speed sensor 130, wherein the data collected by the GNSS is used as an observation value to update the position of the vehicle 100 in the world coordinate system; the data collected by the wheel speed sensor 130 is used to optimize the longitudinal speed and displacement estimation of the vehicle 100. Then, the Kalman filter algorithm is used for data fusion, the calculated vehicle posture is converted to the world coordinate system, and the converted vehicle posture is output.

[0077] In the above embodiment, the estimation accuracy of the vehicle's posture is improved by fusing the data collected by GNSS, IMU and wheel speed sensor 130. In particular, when the GNSS signal is interfered with or lost, the data collected by IMU and wheel speed sensor 130 can maintain a certain positioning capability and can cope with complex and changeable environmental conditions, such as tunnels, high-rise building obstructions, and uneven road conditions, thereby ensuring the stability and reliability of the positioning of the vehicle 100.

[0078] Step S11, using a deep learning model to obtain target information of the target object based on the collected image.

[0079] In some embodiments of the present application, the target information includes coordinates of preset feature points of the target object in a vehicle coordinate system, wherein the vehicle coordinate system is a coordinate system established with the vehicle 100 as a reference.

[0080] In some embodiments of the present application, the target objects may include but are not limited to traffic signs, such as speed limit signs, no turn signs, stop signs, etc.; road markings, such as lane lines (solid lines, dashed lines, double yellow lines, etc.), stop lines, zebra crossings, etc.; static obstacles, such as guardrails, curbs, buildings, trees, etc. on both sides of the road, speed bumps on the road, etc.

[0081] In some embodiments of the present application, an image sensor 140 installed on the vehicle 100 can be used to capture images of the surrounding environment of the vehicle 100, and by identifying the target object in the captured image, the position information of the target object relative to the vehicle 100, that is, the coordinates of the preset feature points of the target object in the vehicle coordinate system, can be determined, and a high-precision map can be subsequently constructed based on the position information of the target object relative to the vehicle 100.

[0082] In some embodiments of the present application, the image sensor 140 may include but is not limited to a periscopic camera and an surround camera. The image sensor 140 may also capture images at a preset capture frame rate. For example, an image is captured every 0.5 seconds. In one example, the positional relationship between the image sensor 140 and the vehicle 100 is known. For example, the positional relationship may include but is not limited to the installation position of the image sensor 140 in the vehicle coordinate system, the installation angle of the image sensor 140, etc., and further, the conversion relationship between the image sensor 140 coordinate system and the vehicle coordinate system may be determined based on the positional relationship between the image sensor 140 and the vehicle 100. In one example, the conversion relationship between the image sensor 140 coordinate system and the vehicle coordinate system is known.

[0083] Specifically, the target object in the image captured by the image sensor 140 can be first identified based on the deep learning model, and the coordinates of the preset feature points of the target object in the image coordinate system can be determined. Then, according to the conversion relationship between the image coordinate system and the image sensor 140 coordinate system (e.g., the first conversion relationship), the coordinates of the preset feature points in the image coordinate system are converted to the image sensor 140 coordinate system to obtain the coordinates of the preset feature points in the image sensor 140 coordinate system. According to the conversion relationship between the image sensor 140 coordinate system and the vehicle coordinate system (e.g., the second conversion relationship), the coordinates of the preset feature points in the image sensor 140 coordinate system are converted to the coordinates in the vehicle coordinate system to obtain the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the vehicle coordinate system are used as the target information of the target object.

[0084] Specifically, the step of obtaining the target information of the target object based on the captured image using the deep learning model may include: first preprocessing the image, such as image cleaning, image enhancement, image standardization, etc. Then, according to the characteristics and recognition requirements of the target object, select a suitable deep learning model, such as a convolutional neural network (CNN), a recurrent neural network (RNN), etc. Then, the deep learning model is trained using a preset image data set, wherein the image data set includes multiple images with annotation information. For example, the annotation information includes but is not limited to a combination of one or more of the following information: a labeling box of the target object in the image, a boundary contour of the target object, a feature point of the target object, etc. Then, the image captured by the image sensor 140 is input into the trained deep learning model, and the deep learning model performs target detection on the input image, identifies the target object in the image, and further uses the segmentation capability or detection capability of the deep learning model to extract the preset feature points of each target object, and calculates the coordinates of the preset feature points in the image coordinate system. Next, a first conversion relationship between the image coordinate system and the image sensor 140 coordinate system (e.g., an intrinsic parameter matrix of the image sensor 140) is determined based on the intrinsic parameters of the image sensor 140, and the coordinates of the preset feature points in the image coordinate system are converted to coordinates in the image sensor 140 according to the first conversion relationship. According to a second conversion relationship between the image sensor 140 and the vehicle coordinate system (e.g., an extrinsic parameter matrix of the image sensor 140), the coordinates of the preset feature points in the image sensor 140 are converted to coordinates in the vehicle coordinate system, thereby obtaining target information of the target object.

[0085] In the above embodiments, the automatic extraction of target information is achieved through the application of the deep learning model, which greatly improves the processing efficiency and accuracy of extracting target information.

[0086] Step S12, according to the vehicle posture and target information, obtain the target posture and size parameters of the target object in the world coordinate system, and use the target posture and size parameters as the observation information of the target object.

[0087] Specifically, the step of obtaining the target pose of the target object and the size parameters of the target object in the world coordinate system according to the vehicle pose and target information includes: constructing a target coordinate system corresponding to the target object based on the coordinates of the preset feature points in the vehicle coordinate system, and obtaining the coordinates of the preset feature points in the target coordinate system; calculating the target pose conversion relationship between the target coordinate system and the world coordinate system according to the vehicle pose, the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the target coordinate system; using the target pose conversion relationship, converting the coordinates of the target object in the target coordinate system to the coordinates in the world coordinate system, and obtaining the target pose; calculating the size parameters of the target object according to the coordinates of the preset feature points in the target coordinate system.

[0088] It should be noted that how to obtain the target pose and size parameters of the target object in the world coordinate system based on the vehicle pose and target information will be discussed in the following section. Figure 3 The steps shown are described in detail and will not be repeated here to avoid repetition.

[0089] Step S13: filtering and updating the state information of the target object based on the observation information.

[0090] Specifically, the Kalman filter can be used to filter and update the state information of the target object. The steps of filtering and updating the state information of the target object based on the observation information using the Kalman filter include: constructing a state vector of the Kalman filter based on the target posture conversion relationship and size parameters, and determining the observation vector; predicting the state vector at the kth moment based on the state vector at the k-1th moment, where k is a positive integer greater than 1; predicting the error covariance matrix at the kth moment based on the predicted error covariance matrix at the k-1th moment; determining the Kalman filter gain at the kth moment based on the error covariance matrix at the kth moment and the observation noise covariance matrix; determining the updated state vector at the kth moment based on the state vector at the kth moment, the Kalman filter gain and the observation vector, and obtaining the updated state information of the target object based on the updated state vector.

[0091] It should be noted that how to use the Kalman filter to filter and update the state information of the target object based on the observation information will be discussed in the following section. Figure 5 The steps shown are described in detail and will not be repeated here to avoid repetition.

[0092] Step S14: determining whether the target object is located behind the vehicle based on the coordinates of the preset feature points of the target object in the vehicle coordinate system.

[0093] Specifically, when the forward coordinate value of the center point of the target object in the vehicle coordinate system is less than 0, it indicates that the target object is located behind the vehicle 100. When the forward coordinate value of the center point of the target object in the vehicle coordinate system is greater than or equal to 0, it indicates that the target object is not located behind the vehicle 100. The forward coordinate value refers to the coordinate value of the center point of the target object in the positive direction of the x-axis or the positive direction of the y-axis corresponding to the forward direction of the vehicle 100 in the vehicle coordinate system.

[0094] In some embodiments of the present application, when the target object is located behind the vehicle 100, the filtering and updating of the observation information of the target object is stopped, and the updated state information of the target object is output, and then step S15 is executed.

[0095] In some embodiments of the present application, when the target object is not located behind the vehicle 100, the process returns to step S13 and continues to filter and update the observation information of the target object using the filter until the target object is located behind the vehicle 100, at which time the process stops filtering and updating the observation information of the target object.

[0096] Step S15, stop filtering and updating the state information of the target object, output the updated state information of the target object, and construct a map based on the updated state information of the target object.

[0097] Specifically, the SLAM (Simultaneous Localization and Mapping) algorithm can be used to construct and update map information in real time based on the updated state information of the target object.

[0098] In the above embodiment, high-precision positioning of the vehicle 100 is achieved through SLAM technology, and the map information can be continuously updated as the vehicle 100 moves, ensuring the timeliness and accuracy of the map. In addition, the state information of the target object is filtered based on the observation information. The observation information in the update includes the target posture and size parameters, which can further improve the accuracy of the map, and can build a more complete and three-dimensional map, which helps to improve the perception and decision-making ability of intelligent driving. Accurate environmental perception and map construction can help the vehicle 100 better understand and respond to complex traffic scenes, thereby improving the safety and reliability of intelligent driving.

[0099] The map construction method provided in the embodiment of the present application can obtain a more accurate and effective vehicle posture by fusing multiple sets of positioning and motion data collected by GNSS, IMU and inertial measurement unit 120, without the need to use high-precision sensors, thereby reducing the high hardware cost; the observation information of the target object is obtained according to the vehicle posture and target information, without the need for manual labeling, thus reducing the influence of human factors, improving mapping efficiency and accuracy, and reducing labor costs; the map is constructed based on the updated status information of the target object, thereby realizing real-time mapping at the vehicle 100 end, further improving the real-time and accuracy of mapping.

[0100] See also Figure 3 , Figure 3 This is a schematic diagram of the steps of a map construction method provided in yet another embodiment of the present application.

[0101] This embodiment is a specific description of step S12 in the above embodiment, and further describes how to obtain the target pose and size parameters of the target object in the world coordinate system according to the vehicle pose and target information. Specifically, the following steps are included:

[0102] Step S121 : constructing a target coordinate system corresponding to the target object based on the coordinates of the preset feature points in the vehicle coordinate system, and obtaining the coordinates of the preset feature points in the target coordinate system.

[0103] In some embodiments of the present application, the preset feature points may include multiple feature points, such as contour points of the target object, for example, the contour points may include the upper right vertex lu, the upper left vertex ru, the lower left vertex rd, and the lower right vertex ld.

[0104] Specifically, the target coordinate system is a coordinate system established with the target object as a reference. The steps of constructing the target coordinate system may include: first, determining the coordinates of the center point O of the target object based on the coordinates of multiple contour points in the vehicle coordinate system, for example, the coordinates of lu, ru, rd, and ld in the vehicle coordinate system are divided into lu_1, ru_1, rd_1, and ld_1. Then, taking the center point O of the target object as the coordinate origin, determine the x-axis direction according to the long side direction of the image of the target object captured by the image sensor 140, and determine the y-axis direction according to the wide side direction of the image of the target object captured by the image sensor 140. Then, using the right-hand rule, determine the z-axis direction based on the x-axis direction and the y-axis direction. The right-hand rule means that the thumb of the right hand points to the positive direction of the x-axis, and the index finger points to the positive direction of the y-axis, then the direction pointed by the middle finger is the positive direction of the z-axis. Then, based on the above-mentioned coordinate axis directions, construct the target coordinate system corresponding to the target object, such as Figure 4As shown, the target coordinate system takes the center point O as the origin, the x-axis points from lu to ru, and the y-axis points from ld to lu. Finally, the coordinates of the feature points in the target coordinate system are determined, for example, the coordinates of lu, ru, rd, and ld in the target coordinate system are divided into lu_2, ru_2, rd_2, and ld_2.

[0105] Step S122, calculating the target posture conversion relationship between the target coordinate system and the world coordinate system according to the vehicle posture, the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the target coordinate system.

[0106] Specifically, the steps of calculating the target posture transformation relationship include:

[0107] 1. Determine the first position conversion relationship between the vehicle coordinate system and the world coordinate system based on the vehicle position.

[0108] The specific composition of the first posture transformation relationship is as follows:

[0109]

[0110] Among them, w represents the world coordinate system, b represents the vehicle coordinate system, is a 4×4 matrix, representing t k The first position conversion relationship at the moment, that is, t k The position and posture of the vehicle coordinate system in the world coordinate system at this moment, Represented as a 3×3 matrix, representing t k The rotation from the vehicle coordinate system to the world coordinate system at the moment, is a 3×1 vector, representing t k Translation from the vehicle coordinate system to the world coordinate system at time instant.

[0111] 2. According to the coordinates of the preset feature points in the target coordinate system and the coordinates of the preset feature points in the vehicle coordinate system, the second posture conversion relationship between the target coordinate system and the vehicle coordinate system is calculated.

[0112] The specific composition of the second posture transformation relationship is as follows:

[0113]

[0114] Where p represents the target coordinate system, is a 4×4 matrix, representing t k The second posture transformation relationship at time t k The position and posture of the target coordinate system in the vehicle coordinate system at each moment, Represented as a 3×3 matrix, representing t k The rotation from the target coordinate system to the vehicle coordinate system at the moment, is a 3×1 vector, representing tk The translation from the target coordinate system to the vehicle coordinate system at time .

[0115] Specifically, the step of calculating the second posture transformation relationship includes:

[0116] 2.1. Determine a first vector and a second vector according to the coordinates of the plurality of contour points in the vehicle coordinate system, wherein the first vector and the second vector are not parallel to each other.

[0117] Specifically, the first vector may be along the horizontal direction of the vehicle 100, specifically: The second vector may be along the vertical direction of the vehicle 100, specifically:

[0118] 2.2. Determine a first unit vector corresponding to the first vector in the vehicle coordinate system, and a second unit vector corresponding to the second vector in the vehicle coordinate system.

[0119] Specifically, the first unit vector may be along the horizontal direction of the vehicle 100, specifically: The second unit vector may be along the vertical direction of the vehicle 100, specifically:

[0120] 2.3. Determine a rotation matrix between the target coordinate system and the vehicle coordinate system according to the first unit vector, the second unit vector, and the cross product of the first unit vector and the second unit vector.

[0121] Specifically, Perpendicular to and The plane where it is located can be obtained by the cross product of the first unit vector and the second unit vector, specifically:

[0122] The steps of determining the rotation matrix between the target coordinate system and the vehicle coordinate system include:

[0123] 2.3.1 Definition is the unit orthogonal basis of the target coordinate system, specifically:

[0124]

[0125] 2.3.2, Assume yes The expression in the vehicle coordinate system can be obtained:

[0126]

[0127] 2.3.3. Due to is the identity matrix, specifically: From this we can get:

[0128]

[0129] 2.3.4. Since the coordinates of multiple contour points lu, ru, rd, and ld in the vehicle coordinate system are known, namely lu_1, ru_1, rd_1, and ld_1, respectively, the above formula (1) can be used to calculate

[0130] 2.4. According to the coordinates of the center points of multiple feature points in the vehicle coordinate system, determine the translation vector between the target coordinate system and the vehicle coordinate system.

[0131] Specifically, the coordinates of the center point O in the vehicle coordinate system can be obtained from the coordinates of ru and rd in the vehicle coordinate system, specifically:

[0132]

[0133] 2.5. Determine the second posture transformation relationship based on the rotation matrix and translation vector between the target coordinate system and the vehicle coordinate system.

[0134] Specifically, the second posture conversion relationship can be obtained from the above formula (1) and formula (2).

[0135] 3. According to the first pose transformation relationship and the second pose transformation relationship, the target pose transformation relationship between the target coordinate system and the world coordinate system is calculated.

[0136] Specifically, the target posture transformation relationship is calculated by the following formula:

[0137]

[0138] in, is a 4×4 matrix, representing t k The target posture transformation relationship at the moment, that is, t k The position and posture of the target coordinate system in the world coordinate system at each moment, Represented as a 3×3 matrix, representing t k The rotation from the target coordinate system to the world coordinate system at that moment, is a 3×1 vector, representing t k The translation from the target coordinate system to the world coordinate system at that moment.

[0139] Step S123, using the target pose conversion relationship, convert the coordinates of the target object in the target coordinate system to the coordinates in the world coordinate system to obtain the target pose.

[0140] Specifically, the coordinates of the target object in the target coordinate system are converted to the coordinates in the world coordinate system by the following formula:

[0141]

[0142] Among them, (x', y', z') represents the coordinates of the target object in the target coordinate system, and (x, y, z) represents the coordinates of the target object in the world coordinate system.

[0143] Step S124, calculating the size parameters of the target object according to the coordinates of the preset feature points in the target coordinate system.

[0144] Specifically, the step of calculating the size parameters of the target object includes:

[0145] 1. Determine a first parameter according to the coordinates of the first dimension of the plurality of contour points in the target coordinate system, wherein the first parameter includes an absolute value of a maximum difference between the coordinates of the first dimension of the plurality of contour points.

[0146] In some embodiments of the present application, the first dimension may be the x-axis direction of the target coordinate system, and the first parameter represents the length in the size parameter. Since the coordinates of lu, ru, rd, and ld in the target coordinate system are lu_2, ru_2, rd_2, and ld_2, the length is defined as l, and l=|ru_2-lu_2|.

[0147] 2. Determine a second parameter according to the coordinates of the second dimension of the plurality of contour points in the target coordinate system, where the second parameter includes an absolute value of a maximum difference between the coordinates of the second dimension of the plurality of contour points.

[0148] In some embodiments of the present application, the second dimension may be the y-axis direction of the target coordinate system, and the second parameter represents the width in the size parameter. The width is defined as w, then w=|ru_2-rd_2|.

[0149] 3. Determine the size parameters of the target object based on the first parameter and the second parameter.

[0150] In other embodiments, the coordinates of the preset feature points in the vehicle coordinate system can be converted to coordinates in the world coordinate system according to the first pose conversion relationship to obtain the coordinates of the preset feature points in the world coordinate system, and then the size parameters of the target object can be calculated according to the coordinates of the preset feature points in the world coordinate system.

[0151] See also Figure 5 , Figure 5 This is a schematic diagram of the steps of a map construction method provided in yet another embodiment of the present application.

[0152] This embodiment is a specific description of step S13 in the above embodiment, and further describes how to use the Kalman filter to filter and update the state information of the target object based on the observation information. Specifically, it includes the following steps:

[0153] Step S131, constructing the state vector of the Kalman filter based on the target posture conversion relationship and size parameters, and determining the observation vector.

[0154] Specifically, the system state vector of the Kalman filter is set to:

[0155] X=(x,y,z,θ x ,θ y ,θ z ,l,w) T

[0156] Among them, the translation vector in the target pose transformation relationship is (θ x ,θ y ,θ z ) T is the rotation matrix in the target pose transformation relationship The corresponding Lie algebraic form, (l,w) T The length and width of the target object's size parameters.

[0157] Since the system state vector is at t k-1 The filtering result at time t is k The predicted value of the filter at the moment, so the state transfer matrix F is I 8×8 Matrix, that is, the unit matrix. The observation vector can also be transformed into the observation vector corresponding to the system state vector, specifically:

[0158]

[0159] Since the state vector and the observation vector are consistent, the observation matrix H is also I 8×8 Matrix, that is, the identity matrix.

[0160] Step S132, predicting the state vector at the kth moment based on the state vector at the k-1th moment.

[0161] Wherein, k is a positive integer greater than 1.

[0162] Specifically, since the system state vector is a static quantity, it can be considered that there is no input disturbance and process noise, and the prediction equation of the state vector is:

[0163]

[0164] in, represents the state vector at the kth moment, Represents the state vector at the k-1th moment.

[0165] Step S133: predicting the error covariance matrix at the kth moment according to the error covariance matrix at the k-1th moment.

[0166] Specifically, the prediction equation of the error covariance matrix is:

[0167] P k / k-1 =P k-1

[0168] Among them, P k / k-1 represents the error covariance matrix at the kth moment, P k-1 represents the error covariance matrix at the k-1th moment.

[0169] Step S134, determining the Kalman filter gain at the kth moment according to the error covariance matrix and the observation noise covariance matrix at the kth moment.

[0170] Specifically, the calculation equation of the Kalman filter gain is:

[0171] K k =P k / k-1 (P k / k-1 +R k ) -1

[0172] Among them, K k represents the Kalman filter gain at the kth moment, R k represents the observation noise covariance matrix at the kth moment.

[0173] In some embodiments of the present application, the observation noise covariance matrix is ​​determined based on the positioning confidence of the fusion positioning algorithm, the detection confidence of the deep learning model, and the detection distance.

[0174] Specifically, the step of determining the observation noise covariance matrix includes: determining a first observation parameter according to the detection confidence, the first observation parameter is inversely proportional to the detection confidence; determining a second observation parameter according to the positioning confidence, the second observation parameter is inversely proportional to the positioning confidence; determining a third observation parameter according to the detection distance, the third observation parameter is proportional to the detection distance; determining a weighted sum of the first observation parameter, the second observation parameter and the third observation parameter; determining a comprehensive observation parameter according to the weighted sum and the maximum value of the preset hyperparameters; and determining the observation noise covariance matrix based on the comprehensive observation parameters.

[0175] In some embodiments of the present application, the value range of the detection confidence may be [0, 100], the value range of the positioning confidence may be [0, 255], and the value range of the detection distance may be [0, 15]. The present application does not impose any restrictions on this.

[0176] In some embodiments of the present application, the weights of the first observation parameter, the second observation parameter, and the third observation parameter may be set according to actual needs, for example, all of them are 1, and the present application does not impose any limitation on this.

[0177] In some embodiments of the present application, the preset hyperparameter may be a smaller value greater than 0, such as 0.0001, and the present application does not impose any limitation on this.

[0178] Specifically, the observation noise covariance matrix is ​​calculated by combining the above content and the following calculation formula, which is:

[0179]

[0180] k=max(a+β+γ,0.0001)

[0181] R=k·I 8×8

[0182] Among them, α represents the first observation parameter, β represents the second observation parameter, γ represents the third observation parameter, k represents the comprehensive observation parameter, and R represents the observation noise covariance matrix.

[0183] Step S135, determining the updated state vector at the kth moment according to the state vector at the kth moment, the Kalman filter gain and the observation vector, and obtaining the updated state information of the target object according to the updated state vector.

[0184] Specifically, the update equation of the state vector is:

[0185]

[0186] in, represents the state vector updated at the kth moment, Z k Represents the observation vector at the kth moment.

[0187] Step S136, determining the updated error covariance matrix at the kth moment according to the error covariance matrix at the kth moment and the Kalman filter gain.

[0188] Specifically, the update equation of the error covariance matrix is:

[0189] P k =(IK k ) k / k-1

[0190] Among them, P k Represents the error covariance matrix after the update at the kth moment.

[0191] In the above embodiment, the updated error covariance matrix is ​​used to indicate the uncertainty of the estimation of the obtained state information.

[0192] See also Figure 6 , Figure 6 A schematic diagram of the functional modules of the map construction device 10 provided in an embodiment of the present application.

[0193] In this embodiment, based on the same concept as the map construction method in the above embodiment, the present application also provides a map construction device 10, which can be used to execute the above map construction method. For the convenience of explanation, the composition diagram of the embodiment of the map construction device 10 only shows the parts related to the embodiment of the present application. Those skilled in the art can understand that the illustrated structure does not constitute a limitation on the map construction device 10, and it can include more or less components than the illustrated one, or combine certain components, or arrange the components differently.

[0194] Specifically, the map construction device 10 provided in the embodiment of the present application includes a fusion positioning module 11, a deep learning module 12, an information acquisition module 13 and a map construction module 14. The fusion positioning module 11 is used to obtain multiple sets of positioning and motion data of the vehicle 100 by using a fusion positioning algorithm to obtain the vehicle posture of the vehicle 100 in the world coordinate system. The deep learning module 12 is used to obtain the target information of the target object based on the collected image using a deep learning model, wherein the target information includes the coordinates of the preset feature points of the target object in the vehicle coordinate system. The information acquisition module 13 is used to obtain the target posture of the target object in the world coordinate system and the size parameters of the target object according to the vehicle posture and target information, and use the target posture and size parameters as the observation information of the target object. The map construction module 14 is used to construct a map based on the updated state information of the target object obtained by filtering and updating the state information of the target object based on the observation information.

[0195] In some embodiments of the present application, the map construction device 10 further includes a filter update module 15. The filter update module 15 is used to filter and update the state information of the target object based on the observation information using a Kalman filter.

[0196] Combination Figure 2 As shown, in some embodiments of the present application, the electronic device 150 includes, but is not limited to, a memory 151 storing computer-readable instructions; and a processor 152 executing the computer-readable instructions stored in the memory 151 to implement the map construction method of the above embodiment.

[0197] Those skilled in the art will appreciate that the schematic diagram is merely an example of the electronic device 150 and does not constitute a limitation on the electronic device 150 , and may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the electronic device 150 may also include an input device, an output device, a network access device, a bus, etc.

[0198] The processor 152 obtains the operating system of the electronic device 150 and various installed applications. The processor 152 obtains the application to implement the steps in the above-mentioned various map construction method embodiments, for example Figure 1 , Figure 3 , Figure 5 Steps shown.

[0199] Exemplarily, the computer readable instructions may be divided into one or more modules / units, one or more modules / units are stored in the memory 151, and are obtained by the processor 152 to complete the present application. One or more modules / units may be a series of computer instruction segments capable of completing a specific function, and the instruction segment is used to describe the acquisition process of the computer readable instructions in the electronic device 150.

[0200] In some embodiments of the present application, the memory 151 is used to store program codes and various data, such as the map construction device 10 installed in the electronic device 150, and to achieve high-speed and automatic access to programs or data during the operation of the electronic device 150. The memory 151 may include a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0201] In some embodiments of the present application, the memory 151 may also be an external memory and / or an internal memory of the electronic device 150. Furthermore, the memory 151 may be a memory in a physical form, such as a memory stick, a TF card (Trans-flash Card), and the like.

[0202] In some embodiments of the present application, the processor 152 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 152 is the computing core and control center of the electronic device 150, and uses various interfaces and lines to connect various parts of the entire electronic device 150, and calls the data stored in the memory 151 to execute various functions of the electronic device 150 and process data, such as executing the function of map construction.

[0203] In some embodiments of the present application, the processor 152 is used to obtain the operating system of the electronic device 150 and various installed applications. For example, the processor 152 obtains a map construction program to implement the map construction method of the above embodiment. Figure 1 , Figure 3 , Figure 5 Steps shown.

[0204] In some embodiments of the present application, the electronic device 150 may also include a power supply (not shown) for supplying power to each component. Preferably, the power supply may be logically connected to the processor 152 through a power management device, so that the power management device can manage charging, discharging, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 150 may also include a Bluetooth module, a Wi-Fi module, etc., which will not be described in detail here.

[0205] In some embodiments of the present application, if the modules / submodules integrated in the electronic device 150 are implemented in the form of software functional submodules and sold or used as independent artifacts, they can be stored in a storage medium. Based on this understanding, the present application implements all or part of the steps in the above-mentioned embodiment method, and can also instruct the relevant hardware to complete through computer-readable instructions. The computer-readable instructions can be stored in a storage medium. When the computer-readable instructions are obtained by the processor 152, the above-mentioned Figure 1 , Figure 3 , Figure 5 The steps of each map construction method embodiment are shown.

[0206] In some embodiments of the present application, the computer-readable instructions may include computer-readable instruction codes, which may be in source code form, object code form, obtainable files, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer-readable instruction code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM).

[0207] The memory 151 in the electronic device 150 stores a plurality of instructions to implement a map construction method, and the processor 152 can obtain the plurality of instructions to implement the map construction method of the above embodiment. Specifically, the specific implementation method of the above instructions by the processor 152 can refer to Figure 1 , Figure 3 , Figure 5 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0208] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0209] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0210] Each functional module in each embodiment of the present application can be integrated into a processing submodule, or each submodule can exist physically separately, or two or more submodules can be integrated into one submodule. The above-mentioned integrated submodule can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0211] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present application is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present application. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0212] In addition, it is obvious that the word "comprising" does not exclude other submodules or steps, and the singular does not exclude the plural. The multiple submodules or devices stated in this application can also be implemented by one submodule or device through software or hardware.

[0213] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present application.

Claims

1. A map construction method, characterized in that: Applied to electronic equipment, the map construction method includes: The fusion positioning algorithm is used to fuse multiple sets of positioning and motion data of the vehicle to obtain the vehicle's position and posture in the world coordinate system; Acquire target information of a target object based on the collected image using a deep learning model, wherein the target information includes coordinates of preset feature points of the target object in a vehicle coordinate system; According to the vehicle posture and the target information, a target posture of the target object in the world coordinate system and a size parameter of the target object are acquired, and the target posture and the size parameter are used as observation information of the target object; A map is constructed based on the updated state information of the target object obtained by filtering and updating the state information of the target object based on the observation information.

2. The map construction method according to claim 1, characterized in that: The acquiring the target pose of the target object in the world coordinate system and the size parameters of the target object according to the vehicle pose and the target information comprises: Constructing the target coordinate system corresponding to the target object based on the coordinates of the preset feature points in the vehicle coordinate system, and obtaining the coordinates of the preset feature points in the target coordinate system; Calculating a target posture conversion relationship between the target coordinate system and the world coordinate system according to the vehicle posture, the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the target coordinate system; Using the target pose conversion relationship, the coordinates of the target object in the target coordinate system are converted to coordinates in the world coordinate system to obtain the target pose; The size parameter of the target object is calculated according to the coordinates of the preset feature point in the target coordinate system.

3. The map construction method according to claim 2, characterized in that: The calculating the target posture conversion relationship between the target coordinate system and the world coordinate system according to the vehicle posture, the coordinates of the preset feature points in the vehicle coordinate system, and the coordinates of the preset feature points in the target coordinate system comprises: Determine a first posture conversion relationship between the vehicle coordinate system and the world coordinate system according to the vehicle posture; Calculating a second posture conversion relationship between the target coordinate system and the vehicle coordinate system according to the coordinates of the preset feature point in the target coordinate system and the coordinates of the preset feature point in the vehicle coordinate system; The target posture conversion relationship between the target coordinate system and the world coordinate system is calculated according to the first posture conversion relationship and the second posture conversion relationship.

4. The map construction method according to claim 2, characterized in that: The preset feature points include a plurality of contour points, and calculating the size parameters of the target object according to the coordinates of the preset feature points in the target coordinate system includes: Determine a first parameter according to the coordinates of the first dimension of the plurality of contour points in the target coordinate system, wherein the first parameter includes an absolute value of a maximum difference between the coordinates of the first dimension of the plurality of contour points; Determine a second parameter according to the coordinates of the second dimension of the plurality of contour points in the target coordinate system, the second parameter comprising an absolute value of a maximum difference between the coordinates of the second dimension of the plurality of contour points; The size parameter is determined according to the first parameter and the second parameter.

5. The map construction method according to claim 2, characterized in that: The map construction method further includes: Based on the target posture conversion relationship and the size parameter, a state vector of a Kalman filter is constructed, and an observation vector is determined; According to the state vector at the k-1th moment, predict the state vector at the kth moment, where k is a positive integer greater than 1; According to the prediction error covariance matrix at the k-1th moment, predict the error covariance matrix at the kth moment; Determine the Kalman filter gain at the kth moment according to the error covariance matrix and the observation noise covariance matrix at the kth moment; According to the state vector at the kth moment, the Kalman filter gain and the observation vector, the updated state vector at the kth moment is determined, and the updated state information of the target object is obtained according to the updated state vector.

6. The map construction method according to claim 5, characterized in that: The map construction method further includes: The observation noise covariance matrix is ​​determined according to the positioning confidence of the fusion positioning algorithm, the detection confidence and the detection distance of the deep learning model.

7. The map construction method according to claim 1, characterized in that: The map construction method further includes: Determining whether the target object is located behind the vehicle according to the coordinates of the preset feature points of the target object in the vehicle coordinate system; When the target object is located behind the vehicle, filtering and updating the state information of the target object is stopped, and the updated state information of the target object is output.

8. A map construction device, characterized in that: The map construction device comprises: The fusion positioning module is used to obtain multiple sets of positioning and motion data of the vehicle by using the fusion positioning algorithm to obtain the vehicle's position and posture in the world coordinate system; A deep learning module, used to obtain target information of a target object based on the collected image using a deep learning model, wherein the target information includes coordinates of preset feature points of the target object in a vehicle coordinate system; An information acquisition module, used to acquire the target posture of the target object in the world coordinate system and the size parameters of the target object according to the vehicle posture and the target information, and use the target posture and the size parameters as observation information of the target object; A map construction module is used to construct a map based on the updated state information of the target object obtained by filtering and updating the state information of the target object based on the observation information.

9. An electronic device, characterized in that: The electronic device comprises: a memory storing computer-readable instructions; and a processor executing the computer-readable instructions stored in the memory to implement the map construction method according to any one of claims 1 to 7.

10. A vehicle, characterized in that: The vehicle comprises the electronic device as claimed in claim 9.

Images