[go: up one dir, main page]

CN112950972A - Parking lot map construction method, device, equipment and medium - Google Patents

Parking lot map construction method, device, equipment and medium Download PDF

Info

Publication number
CN112950972A
CN112950972A CN202110097606.1A CN202110097606A CN112950972A CN 112950972 A CN112950972 A CN 112950972A CN 202110097606 A CN202110097606 A CN 202110097606A CN 112950972 A CN112950972 A CN 112950972A
Authority
CN
China
Prior art keywords
map
parking lot
dimensional
identified
intersection
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110097606.1A
Other languages
Chinese (zh)
Other versions
CN112950972B (en
Inventor
罗飞扬
王梓里
夏志勋
温俊杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou Xiaopeng Motors Technology Co Ltd
Original Assignee
Guangzhou Xiaopeng Autopilot Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangzhou Xiaopeng Autopilot Technology Co Ltd filed Critical Guangzhou Xiaopeng Autopilot Technology Co Ltd
Priority to CN202110097606.1A priority Critical patent/CN112950972B/en
Publication of CN112950972A publication Critical patent/CN112950972A/en
Application granted granted Critical
Publication of CN112950972B publication Critical patent/CN112950972B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/09—Arrangements for giving variable traffic instructions
    • G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
    • G08G1/0967—Systems involving transmission of highway information, e.g. weather, speed limits
    • G08G1/096708—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control
    • G08G1/096725—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control where the received information generates an automatic action on the vehicle control
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/251—Fusion techniques of input or preprocessed data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G06T17/05—Geographic models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Multimedia (AREA)
  • Geometry (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Remote Sensing (AREA)
  • Computer Graphics (AREA)
  • Atmospheric Sciences (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)
  • Image Processing (AREA)

Abstract

The application discloses a parking lot map construction method, a parking lot map construction device, parking lot map construction equipment and parking lot map construction media, wherein the method comprises the following steps: acquiring parking lot data acquired by more than two sensors at a vehicle end; converting the parking lot data acquired by each sensor into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system; determining a map to be identified based on the three-dimensional map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified; and adding the intersection recognition result to the three-dimensional map to obtain the parking lot map. The method and the device improve the technical problems that in the prior art, intersection recognition is directly carried out on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced.

Description

Parking lot map construction method, device, equipment and medium
Technical Field
The present application relates to the field of map construction technologies, and in particular, to a method, an apparatus, a device, and a medium for constructing a parking lot map.
Background
In order to ensure that an automatic driving vehicle safely runs in a parking lot environment, the intersection needs to be accurately sensed in advance, so that the intersection in a parking lot map is a very important semantic element. In the prior art, intersections of a parking lot environment are usually identified based on visual images, but for a typical scene of the parking lot intersections, intersection identification of images acquired by a camera is difficult, particularly when the scene is blocked, the visual field is not complete and the like, intersection identification precision is difficult to guarantee, and further the driving safety of an automatic driving vehicle in the parking lot environment is influenced.
Disclosure of Invention
The application provides a parking lot map construction method, device, equipment and medium, which are used for solving the technical problems that in the prior art, intersection identification is directly carried out on a parking lot image collected by a camera, the identification precision is low, and the driving safety of an automatic driving vehicle in a parking lot environment is influenced.
In view of this, a first aspect of the present application provides a parking lot map building method, including:
acquiring parking lot data acquired by more than two sensors at a vehicle end;
converting the parking lot data acquired by each sensor into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system;
determining a map to be identified based on the three-dimensional map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified;
and adding the intersection identification result to the three-dimensional map to obtain a parking lot map.
Optionally, after converting the parking lot data collected by each sensor into the same coordinate system, constructing a three-dimensional map based on the parking lot data in the same coordinate system, and then further including:
carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps;
the method for determining the map to be identified based on the three-dimensional map and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified comprises the following steps:
and determining a map to be identified based on each three-dimensional sub-map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
Optionally, the performing area division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps further includes:
mapping each three-dimensional sub-map to obtain a map image;
determining a map to be identified based on each three-dimensional sub-map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified, wherein the intersection identification result comprises the following steps:
and determining a map image to be identified based on the map image, and inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
Optionally, after the parking lot data acquired by each sensor is converted into the same coordinate system, a three-dimensional map is constructed based on the parking lot data in the same coordinate system, and then the method further includes:
and mapping the three-dimensional map to obtain a map image.
Optionally, the mapping process of the map image includes:
determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped;
and mapping the three-dimensional map to be mapped into the map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped.
Optionally, the configuration process of the preset deep learning model is as follows:
determining a map to be annotated based on the three-dimensional map;
marking the accessible edges and the inaccessible edges of the intersections in the map to be marked to obtain training samples;
and training a deep learning network through the training samples to obtain the preset deep learning model.
The second aspect of the present application provides a parking lot map building apparatus, including:
the system comprises an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used for acquiring parking lot data acquired by more than two sensors at a vehicle end;
the conversion unit is used for converting the parking lot data acquired by the sensors into the same coordinate system and then constructing a three-dimensional map based on the parking lot data in the same coordinate system;
the identification unit is used for determining a map to be identified based on the three-dimensional map and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified;
and the adding unit is used for adding the intersection identification result to the three-dimensional map to obtain a parking lot map.
Optionally, the method further includes:
the dividing unit is used for carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps;
the identification unit is specifically used for determining a map to be identified based on each three-dimensional sub-map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
Optionally, the method further includes:
the first mapping unit is used for mapping each three-dimensional sub-map to obtain a map image;
the identification unit is specifically used for determining a map image to be identified based on the map image, inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
Optionally, the method further includes:
and the second mapping unit is used for mapping the three-dimensional map to obtain a map image.
Optionally, the mapping process of the map image includes:
determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped;
and mapping the three-dimensional map to be mapped into the map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped.
A third aspect of the present application provides a parking lot mapping apparatus comprising a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the parking lot map construction method according to any one of the first aspect according to instructions in the program code.
A fourth aspect of the present application provides a computer-readable storage medium for storing program code for executing the parking lot map construction method according to any one of the first aspects.
According to the technical scheme, the method has the following advantages:
the application provides a parking lot map construction method, which comprises the following steps: acquiring parking lot data acquired by more than two sensors at a vehicle end; converting the parking lot data acquired by each sensor into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system; determining a map to be identified based on the three-dimensional map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified; and adding the intersection recognition result to the three-dimensional map to obtain the parking lot map.
In the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.
Fig. 1 is a schematic flowchart of a parking lot map building method according to a first embodiment of the present disclosure;
fig. 2 is a top view of a three-dimensional map of a parking lot according to an embodiment of the present disclosure;
fig. 3 is a schematic flowchart of a second embodiment of a parking lot map construction method provided in the present application;
fig. 4 is a schematic flowchart of a third embodiment of a parking lot map construction method provided in the present application;
fig. 5 is a schematic flowchart of a fourth embodiment of a parking lot map construction method provided in the present application;
fig. 6 is a map image obtained by mapping according to an embodiment of the present application;
fig. 7 is a schematic diagram of a training sample obtained by labeling the map image in fig. 6 according to an embodiment of the present application;
fig. 8 is a top view perspective view of a parking lot map obtained after intersection recognition is performed on the three-dimensional map in fig. 2 according to an embodiment of the present application;
fig. 9 is a schematic structural diagram of a parking lot map building apparatus according to an embodiment of the present application.
Detailed Description
The application provides a parking lot map construction method, device, equipment and medium, which are used for solving the technical problems that in the prior art, intersection identification is directly carried out on a parking lot image collected by a camera, the identification precision is low, and the driving safety of an automatic driving vehicle in a parking lot environment is influenced.
In order to make the technical solutions of the present application better understood, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
For easy understanding, please refer to fig. 1, an embodiment of a parking lot map building method provided by the present application includes:
step 101, acquiring parking lot data acquired by more than two sensors at a vehicle end.
The embodiment of the application considers that intersection identification is difficult to directly carry out on the parking lot image acquired by the camera, particularly when the problems of scene shading, incomplete visual field and the like exist, the intersection identification precision is difficult to guarantee, and the driving safety of the automatic driving vehicle in the parking lot environment is further influenced.
In order to improve the above problem, in the embodiment of the present application, the parking lot data collected by two or more sensors at the vehicle end are obtained, where the sensors may include a camera, a odometer, or a radar, and the radar may include an ultrasonic radar, a millimeter wave radar, or a laser radar. When the automatic driving vehicle runs in the parking lot, the camera of the automatic driving vehicle can identify the pavement marks of the parking lot, such as lane lines, parking spaces, arrows, speed bumps and other common pavement marks; the radar of the autonomous vehicle may detect an obstacle; an odometer for an autonomous vehicle records the trajectory of the vehicle. Therefore, parking lot data including road surface identification, obstacle point cloud, and travel track, etc. may be acquired through a camera, radar, and odometer at the vehicle end. Of course, the odometer of an autonomous vehicle can also be obtained by an IMU (inertial sensor) or a combined inertial navigation system of wheel speeds or other slam algorithm calculations.
And 102, converting the parking lot data acquired by the sensors into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system.
Since the parking lot data collected by different sensors may be data of different dimensions, for example, the road surface identification obtained by a camera is two-dimensional data, and the obstacle point cloud obtained by a radar is three-dimensional data. Therefore, the parking lot data collected by each sensor needs to be converted into the same coordinate system, and then a three-dimensional map of the parking lot is constructed based on the parking lot data in the same coordinate system, where the three-dimensional map includes points and line elements such as lane lines, parking spaces, arrows, obstacle points, speed bumps, and driving tracks, and the top view perspective view of the three-dimensional map of the parking lot provided in fig. 2 can be referred to.
It can be understood that the parking lot data collected by the sensors of the plurality of autonomous vehicles may be obtained, and a three-dimensional map of the plurality of parking lots may be correspondingly constructed.
And 103, determining a map to be identified based on the three-dimensional map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
After the three-dimensional map is constructed, a plurality of three-dimensional maps can be selected from the obtained three-dimensional maps to be used as the map to be annotated, and the rest three-dimensional maps are used as the map to be identified. Configuring a preset deep learning model through a map to be marked; and then inputting the map to be recognized into a preset deep learning model for intersection recognition to obtain an intersection recognition result of the map to be recognized. The intersection identification result comprises an accessible side and an inaccessible side of the intersection in the map to be identified.
Further, the configuration process of the preset deep learning model comprises the following steps: determining a map to be annotated based on the three-dimensional map; marking the accessible edges and the inaccessible edges of the intersection in the map to be marked to obtain a training sample; and training a deep learning network through the training samples to obtain a preset deep learning model. The deep learning network in the embodiment of the application may adopt an existing deep learning network structure, for example, a residual error network, and a specific network structure may be selected according to actual needs, where the specific structure of the deep learning network is not specifically limited.
And step 104, adding the intersection identification result to the three-dimensional map to obtain the parking lot map.
And particularly, the intersection and the accessible and inaccessible sides of the intersection are added to the corresponding three-dimensional map to obtain a parking lot map, so that when an automatic driving vehicle enters a certain parking lot, the automatic driving vehicle can safely drive according to the parking lot map of the parking lot, and the driving safety of the automatic driving vehicle is ensured. After the intersection is identified on the three-dimensional map in fig. 2, the parking lot map obtained by adding the intersection identification result to the three-dimensional map is shown in fig. 8.
In the embodiment of the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
The above is a first embodiment of the parking lot map building method provided by the present application, and the following is a second embodiment of the parking lot map building method provided by the present application.
Referring to fig. 3, a method for constructing a parking lot map provided by the embodiment of the present application includes:
step 201, parking lot data collected by more than two sensors at the vehicle end are obtained.
Step 202, after the parking lot data collected by the sensors are converted into the same coordinate system, a three-dimensional map is constructed based on the parking lot data in the same coordinate system.
The specific contents of steps 201 to 202 are the same as those of steps 101 to 102, and are not described herein again.
And 203, carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps.
According to the embodiment of the application, the fact that the whole three-dimensional map of the parking lot is directly input into the preset deep learning model for intersection recognition is considered, the data volume is large, and intersection recognition speed is influenced. In order to improve the problem, after the three-dimensional map is constructed, the three-dimensional map may be further divided into a plurality of sub-areas according to the size of the three-dimensional map, so as to obtain a plurality of three-dimensional sub-maps.
And 204, determining a map to be identified based on each three-dimensional sub-map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
After the plurality of three-dimensional sub-maps are obtained, a plurality of three-dimensional sub-maps can be selected from the obtained plurality of three-dimensional sub-maps to be used as maps to be annotated, and the rest three-dimensional sub-maps are used as maps to be identified. Configuring a preset deep learning model through a map to be marked; and then inputting the map to be recognized into a preset deep learning model for intersection recognition to obtain an intersection recognition result of the map to be recognized. The intersection identification result comprises an accessible side and an inaccessible side of the intersection in the map to be identified.
Further, the configuration process of the preset deep learning model comprises the following steps: determining a map to be annotated based on the three-dimensional sub-map; marking the accessible edges and the inaccessible edges of the intersection in the map to be marked to obtain a training sample; and training a deep learning network through the training samples to obtain a preset deep learning model.
And step 205, adding the intersection recognition result to the three-dimensional map to obtain the parking lot map.
And particularly, the intersection and the accessible and inaccessible sides of the intersection are added to the corresponding three-dimensional map to obtain a parking lot map, so that when an automatic driving vehicle enters a certain parking lot, the automatic driving vehicle can safely drive according to the parking lot map of the parking lot, and the driving safety of the automatic driving vehicle is ensured.
In the embodiment of the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
Furthermore, the embodiment of the application divides the three-dimensional map into a plurality of three-dimensional sub-maps, and then inputs each three-dimensional sub-map into the preset deep learning model for intersection recognition, so that the problems that the whole three-dimensional map of the parking lot is directly input into the preset deep learning model for intersection recognition, the data size is large, and the intersection recognition speed is influenced are solved.
The above is an embodiment two of the parking lot map building method provided by the present application, and the following is an embodiment three of the parking lot map building method provided by the present application.
Referring to fig. 4, a method for constructing a parking lot map provided in an embodiment of the present application includes:
301, acquiring parking lot data acquired by more than two sensors at the vehicle end.
And 302, converting the parking lot data acquired by the sensors into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system.
The specific contents of steps 301 to 302 are the same as the specific contents of steps 101 to 102, and are not described herein again.
And 303, mapping the three-dimensional map to obtain a map image.
In the embodiment of the application, after the three-dimensional map is constructed, the three-dimensional map can be mapped into the image according to the mapping relation between the map coordinate and the image coordinate to obtain the map image, so that intersection identification of the three-dimensional map is converted into intersection identification of the two-dimensional image, and the complexity of input data of the deep learning model can be reduced.
Further, the mapping process of the map image may be: determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped; and mapping the three-dimensional map to be mapped into a map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped. And when the three-dimensional map is subjected to mapping processing to obtain a map image, the three-dimensional map to be mapped is the three-dimensional map.
When a three-dimensional map to be mapped is mapped into a map image, a mapping relation between a map coordinate and an image coordinate needs to be established, when the mapping relation between the map coordinate and the image coordinate is established, the height information of the three-dimensional map to be mapped is ignored, a certain point in the three-dimensional map to be mapped is known to be P (x, y) in a map coordinate system, and a point P (u, v) in the image coordinate system corresponding to the point P (x, y) can be calculated according to the mapping relation between the map coordinate and the image coordinate, namely:
u＝(x-minx)/s；
v＝(y-miny)/s；
in the formula (II)xIs the minimum coordinate value of the abscissa, min, of the three-dimensional map to be mappedyThe minimum coordinate value of the vertical coordinate of the three-dimensional map to be mapped is s, the preset map-image scale factor is provided, the unit is meter/pixel, and the value of s can be set as required actually, and is not specifically limited herein; the unit of the point in the map coordinate system is meter, and the unit of the point in the image coordinate system is pixel.
Conversely, if a point P (u, v) in the image coordinate system is known, a point P (x, y) in the map coordinate system corresponding to the point P (u, v) can be calculated according to the mapping relationship between the map coordinates and the image coordinates, that is:
x＝u*s+minx；
y＝v*s+miny；
according to the maximum coordinate value max of the abscissa of the three-dimensional map to be mappedxMaximum ordinate value maxyPreset map-image scale factor s and ordinate minimum coordinate value minyAnd minimum coordinate value min of abscissaxThe size w × h of the mapped map image may be calculated, that is:
w＝(maxx-minx)/s；
h＝(maxy-miny)/s；
where w is the width of the map image and h is the height of the map image.
Further, after the three-dimensional map is mapped into the map image, different types of elements (parking spaces, lane lines, speed bumps, arrows, driving tracks and the like) in the map image can be represented by different colors. Different data elements are distinguished by different colors, intersection features are more obvious, the accuracy of the preset deep learning model in identifying the intersections in the map image is improved, and passable and non-passable edges of the intersections are effectively distinguished.
And 304, determining a map image to be identified based on the map image, and inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
It can be understood that the parking lot data collected by the sensors of the plurality of autonomous vehicles may be acquired, a three-dimensional map of the plurality of parking lots may be constructed, and correspondingly, a plurality of map images may be mapped. The method comprises the steps of selecting a plurality of map images from the obtained map images as to-be-identified map images, inputting the to-be-identified map images as to-be-identified maps into a preset deep learning model for intersection identification, and obtaining intersection identification results of the to-be-identified maps, wherein the intersection identification results comprise accessible edges and inaccessible edges of intersections in the to-be-identified maps.
After a plurality of map images are selected from the obtained map images to be used as the map images to be identified, the rest map images are used as the map images to be annotated; marking intersections in the map image to be marked, specifically marking the accessible edges and the inaccessible edges of the intersections in the map image to be marked to obtain training samples; and training a deep learning network through the training samples to obtain a preset deep learning model.
In the embodiment of the application, the mapping relation between the map coordinate and the image coordinate is established, intersection identification of the three-dimensional map is converted into intersection identification of the two-dimensional image, the marking process of the map image to be marked is simplified, the complexity of deep learning network input data is reduced, the complexity of a preset deep learning model is further reduced, and the intersection identification speed is favorably improved.
And 305, adding the intersection identification result to the three-dimensional map to obtain the parking lot map.
After the intersection recognition result of the map to be recognized is obtained, the intersection characteristics of the map to be recognized, namely the intersection in the map to be recognized and the accessible edges and the inaccessible edges of the intersection can be determined, and then the polygonal intersection area can be determined; according to the mapping relation between the map coordinates and the image coordinates, the coordinates of the polygonal intersection area in the map coordinate system can be obtained through calculation, and finally the polygonal intersection area is added into the three-dimensional map according to the coordinates of the polygonal intersection area in the map coordinate system, so that the parking lot map is obtained.
In the embodiment of the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
Furthermore, the mapping relation between the map coordinate and the image coordinate is established, intersection identification of the three-dimensional map is converted into intersection identification of the two-dimensional image, the labeling process of the map image to be labeled is simplified, the complexity of the deep learning network input data is reduced, the complexity of the preset deep learning model is further reduced, and the intersection identification speed is improved.
Further, after the three-dimensional map is mapped into the map image, different colors are adopted for different types of elements (parking spaces, lane lines, speed bumps, arrows, driving tracks and the like) in the map image to represent, different data elements are distinguished by using different colors, intersection features are more obvious, further, the accuracy of the preset deep learning model for identifying intersections in the map image is improved, and passable and impassable edges of the intersections are effectively distinguished.
The above is a third embodiment of the parking lot map building method provided by the present application, and the following is a fourth embodiment of the parking lot map building method provided by the present application.
Referring to fig. 5, a method for constructing a parking lot map provided in an embodiment of the present application includes:
step 401, parking lot data collected by more than two sensors at the vehicle end are obtained.
Step 402, after the parking lot data collected by each sensor is converted into the same coordinate system, a three-dimensional map is constructed based on the parking lot data in the same coordinate system.
And 403, carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps.
The specific contents of steps 401 to 403 are the same as the specific contents of steps 201 to 203, and are not described herein again.
And step 404, mapping each three-dimensional sub-map to obtain a map image.
The method has the advantages that the whole map image of the parking lot is directly input into the preset deep learning model for intersection recognition, the data volume is large, and the intersection recognition speed is influenced. In order to improve the problem, in the embodiment of the application, before the three-dimensional map is mapped into the map image, the three-dimensional map can be further divided into areas to obtain a plurality of three-dimensional sub-maps; and mapping each three-dimensional sub-map to obtain a map image. Taking the three-dimensional map provided in fig. 2 as an example, after the three-dimensional map is divided into regions, each three-dimensional sub-map is mapped to obtain a plurality of map images, which may refer to one map image obtained by mapping provided in fig. 6.
Further, the mapping process of the map image may be: determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped; and mapping the three-dimensional map to be mapped into a map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped. And when the three-dimensional sub-map is subjected to mapping processing to obtain a map image, the three-dimensional map to be mapped is the three-dimensional sub-map. For a specific mapping process, reference may be made to step 303, which is not described herein again.
It should be noted that, after the three-dimensional map is mapped into a map image, the map image may be divided to obtain a plurality of sub-map images; and determining a map image to be identified based on each sub-map image, inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
Step 405, determining a map image to be identified based on the map image, and inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
The method comprises the steps of selecting a plurality of map images from the obtained map images as to-be-identified map images, inputting the to-be-identified map images as to-be-identified maps into a preset deep learning model for intersection identification, and obtaining intersection identification results of the to-be-identified maps, wherein the intersection identification results comprise accessible edges and inaccessible edges of intersections in the to-be-identified maps.
After a plurality of map images are selected from the obtained map images to be used as the map images to be identified, the rest map images are used as the map images to be annotated; the method comprises the steps of marking intersections in a map image to be marked, specifically marking the accessible edges and the inaccessible edges of the intersections in the map image to be marked to obtain a training sample. And taking the map image in the figure 6 as a map image to be marked, marking the passable side of the intersection by adopting a dotted line, and marking the impassable side of the intersection by adopting a solid line to obtain the training sample shown in the figure 7. And training a deep learning network through the training samples to obtain a preset deep learning model.
In the embodiment of the application, the mapping relation between the map coordinate and the image coordinate is established, intersection identification of the three-dimensional map is converted into intersection identification of the two-dimensional image, the marking process of the map image to be marked is simplified, the complexity of deep learning network input data is reduced, the complexity of a preset deep learning model is further reduced, and the intersection identification speed is favorably improved.
And step 406, adding the intersection identification result to the three-dimensional map to obtain a parking lot map.
The specific contents of steps 405 to 406 are the same as those of steps 304 to 305, and are not described herein again.
In the embodiment of the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
Furthermore, the embodiment of the application divides the three-dimensional map into a plurality of three-dimensional sub-maps, then maps each three-dimensional sub-map into a map image, inputs the map image corresponding to the three-dimensional sub-map into the preset deep learning model for intersection recognition, and solves the problems that the whole map image of the parking lot is directly input into the preset deep learning model for intersection recognition, the existing data size is large, and the intersection recognition speed is influenced.
The above is a fourth embodiment of the parking lot map building method provided by the present application, and the following is an embodiment of the parking lot map building apparatus provided by the present application.
Referring to fig. 9, an embodiment of the present application provides a parking lot map building apparatus, including:
the system comprises an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used for acquiring parking lot data acquired by more than two sensors at a vehicle end;
the conversion unit is used for converting the parking lot data acquired by each sensor into the same coordinate system and then constructing a three-dimensional map based on the parking lot data in the same coordinate system;
the identification unit is used for determining a map to be identified based on the three-dimensional map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified;
and the adding unit is used for adding the intersection recognition result to the three-dimensional map to obtain the parking lot map.
As a further improvement, the method further comprises the following steps:
the dividing unit is used for carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps;
and the identification unit is specifically used for determining a map to be identified based on each three-dimensional sub-map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
As a further improvement, the method further comprises the following steps:
the first mapping unit is used for mapping each three-dimensional sub-map to obtain a map image;
and the identification unit is specifically used for determining a map image to be identified based on the map image, inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
As a further improvement, the method further comprises the following steps:
and the second mapping unit is used for mapping the three-dimensional map to obtain a map image.
As a further improvement, the mapping process of the map image includes:
determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped;
and mapping the three-dimensional map to be mapped into a map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped.
In the embodiment of the application, the three-dimensional map is constructed by combining the parking lot data acquired by more than two sensors, so that the three-dimensional map containing richer intersection characteristics can be obtained; the intersection recognition is carried out by inputting the three-dimensional map containing richer intersection characteristics into a preset deep learning model, so that the intersection recognition precision is improved, and a more accurate and reliable intersection recognition result of intersection semantic elements is obtained; the intersection recognition result is added into the three-dimensional map to obtain a high-precision parking lot map, and the obtained high-precision parking lot map is beneficial to improving the driving safety of the automatic driving vehicle in the parking lot environment, so that the technical problems that the prior art directly performs intersection recognition on the parking lot image collected by the camera, the recognition precision is low, and the driving safety of the automatic driving vehicle in the parking lot environment is influenced are solved.
Furthermore, the mapping relation between the map coordinate and the image coordinate is established, intersection identification of the three-dimensional map is converted into intersection identification of the two-dimensional image, the labeling process of the map image to be labeled is simplified, the complexity of the deep learning network input data is reduced, the complexity of the preset deep learning model is further reduced, and the intersection identification speed is improved.
Furthermore, the embodiment of the application divides the three-dimensional map into a plurality of three-dimensional sub-maps, then maps each three-dimensional sub-map into a map image, inputs the map image corresponding to the three-dimensional sub-map into the preset deep learning model for intersection recognition, and solves the problems that the whole map image of the parking lot is directly input into the preset deep learning model for intersection recognition, the existing data size is large, and the intersection recognition speed is influenced.
The embodiment of the application also provides parking lot map construction equipment, which is characterized by comprising a processor and a memory;
the memory is used for storing the program codes and transmitting the program codes to the processor;
the processor is configured to execute the parking lot map construction method in the foregoing method embodiment according to instructions in the program code.
The embodiment of the application also provides a computer-readable storage medium, which is used for storing a program code, and the program code is used for executing the parking lot map building method in the foregoing method embodiment.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more (including two) units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for executing all or part of the steps of the method described in the embodiments of the present application through a computer device (which may be a personal computer, a server, or a network device). And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims (13)

1. A parking lot map construction method is characterized by comprising the following steps:
acquiring parking lot data acquired by more than two sensors at a vehicle end;
converting the parking lot data acquired by each sensor into the same coordinate system, and constructing a three-dimensional map based on the parking lot data in the same coordinate system;
determining a map to be identified based on the three-dimensional map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified;
and adding the intersection identification result to the three-dimensional map to obtain a parking lot map.
2. The parking lot map building method according to claim 1, wherein after converting the parking lot data collected by each sensor into the same coordinate system, building a three-dimensional map based on the parking lot data in the same coordinate system, and then further comprising:
carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps;
the method for determining the map to be identified based on the three-dimensional map and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified comprises the following steps:
and determining a map to be identified based on each three-dimensional sub-map, and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
3. The parking lot map construction method according to claim 2, wherein the three-dimensional map is divided into regions to obtain a plurality of three-dimensional sub-maps, and then the method further comprises:
mapping each three-dimensional sub-map to obtain a map image;
determining a map to be identified based on each three-dimensional sub-map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified, wherein the intersection identification result comprises the following steps:
and determining a map image to be identified based on the map image, and inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified.
4. The parking lot map building method according to claim 1, wherein after converting the parking lot data collected by each sensor into the same coordinate system, building a three-dimensional map based on the parking lot data in the same coordinate system, further comprising:
and mapping the three-dimensional map to obtain a map image.
5. The parking lot map construction method according to claim 3 or 4, wherein the map image mapping process includes:
determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped;
and mapping the three-dimensional map to be mapped into the map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped.
6. The parking lot map construction method according to claim 1, wherein the preset deep learning model is configured by the following process:
determining a map to be annotated based on the three-dimensional map;
marking the accessible edges and the inaccessible edges of the intersections in the map to be marked to obtain training samples;
and training a deep learning network through the training samples to obtain the preset deep learning model.
7. A parking lot map building apparatus, characterized by comprising:
the system comprises an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used for acquiring parking lot data acquired by more than two sensors at a vehicle end;
the conversion unit is used for converting the parking lot data acquired by the sensors into the same coordinate system and then constructing a three-dimensional map based on the parking lot data in the same coordinate system;
the identification unit is used for determining a map to be identified based on the three-dimensional map and inputting the map to be identified into a preset deep learning model for intersection identification to obtain an intersection identification result of the map to be identified;
and the adding unit is used for adding the intersection identification result to the three-dimensional map to obtain a parking lot map.
8. The parking lot map building apparatus according to claim 7, further comprising:
the dividing unit is used for carrying out regional division on the three-dimensional map to obtain a plurality of three-dimensional sub-maps;
the identification unit is specifically used for determining a map to be identified based on each three-dimensional sub-map, inputting the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
9. The parking lot map building apparatus according to claim 8, further comprising:
the first mapping unit is used for mapping each three-dimensional sub-map to obtain a map image;
the identification unit is specifically used for determining a map image to be identified based on the map image, inputting the map image to be identified as the map to be identified into a preset deep learning model for intersection identification, and obtaining an intersection identification result of the map to be identified.
10. The parking lot map building apparatus according to claim 7, further comprising:
and the second mapping unit is used for mapping the three-dimensional map to obtain a map image.
11. The parking lot map building apparatus according to claim 9 or 10, wherein the mapping process of the map image includes:
determining a minimum coordinate value of a vertical coordinate and a minimum coordinate value of an abscissa of the three-dimensional map to be mapped;
and mapping the three-dimensional map to be mapped into the map image according to a preset map-image scale factor, and the minimum coordinate value of the ordinate and the minimum coordinate value of the abscissa of the three-dimensional map to be mapped.
12. A parking lot mapping apparatus, characterized in that the apparatus comprises a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the parking lot mapping method according to any one of claims 1 to 6 according to instructions in the program code.
13. A computer-readable storage medium characterized in that the computer-readable storage medium stores program code for executing the parking lot map construction method according to any one of claims 1 to 6.
CN202110097606.1A 2021-01-25 2021-01-25 Parking lot map construction method, device, equipment and medium Active CN112950972B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110097606.1A CN112950972B (en) 2021-01-25 2021-01-25 Parking lot map construction method, device, equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110097606.1A CN112950972B (en) 2021-01-25 2021-01-25 Parking lot map construction method, device, equipment and medium

Publications (2)

Publication Number Publication Date
CN112950972A true CN112950972A (en) 2021-06-11
CN112950972B CN112950972B (en) 2022-07-01

Family

ID=76236506

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110097606.1A Active CN112950972B (en) 2021-01-25 2021-01-25 Parking lot map construction method, device, equipment and medium

Country Status (1)

Country Link
CN (1) CN112950972B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114093195A (en) * 2021-11-17 2022-02-25 深圳市豪位科技有限公司 Method and device for marking parking spaces
CN114440859A (en) * 2022-01-26 2022-05-06 广州小鹏自动驾驶科技有限公司 Map updating method and device, computer equipment and storage medium
CN114490910A (en) * 2022-01-26 2022-05-13 中国第一汽车股份有限公司 Map generation method and device, electronic equipment and storage medium
CN114743405A (en) * 2022-03-29 2022-07-12 广州小鹏自动驾驶科技有限公司 Parking space data processing method and device, electronic equipment and storage medium
CN114937369A (en) * 2022-04-06 2022-08-23 福思(杭州)智能科技有限公司 Autonomous valet parking method, system, vehicle, parking lot terminal, device and medium
CN117495847A (en) * 2023-12-27 2024-02-02 安徽蔚来智驾科技有限公司 Intersection detection method, readable storage medium and intelligent device

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9984572B1 (en) * 2017-01-16 2018-05-29 Nio Usa, Inc. Method and system for sharing parking space availability among autonomous vehicles
CN108801276A (en) * 2018-07-23 2018-11-13 奇瑞汽车股份有限公司 Accurately drawing generating method and device
CN108959321A (en) * 2017-05-25 2018-12-07 纵目科技（上海）股份有限公司 Parking lot map constructing method, system, mobile terminal and storage medium
CN109253731A (en) * 2018-08-06 2019-01-22 百度在线网络技术（北京）有限公司 Parking space drawing generating method, device, equipment and readable storage medium storing program for executing
CN109429518A (en) * 2017-06-22 2019-03-05 百度时代网络技术（北京）有限公司 Automatic Pilot traffic forecast based on map image

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9984572B1 (en) * 2017-01-16 2018-05-29 Nio Usa, Inc. Method and system for sharing parking space availability among autonomous vehicles
CN108959321A (en) * 2017-05-25 2018-12-07 纵目科技（上海）股份有限公司 Parking lot map constructing method, system, mobile terminal and storage medium
CN109429518A (en) * 2017-06-22 2019-03-05 百度时代网络技术（北京）有限公司 Automatic Pilot traffic forecast based on map image
CN108801276A (en) * 2018-07-23 2018-11-13 奇瑞汽车股份有限公司 Accurately drawing generating method and device
CN109253731A (en) * 2018-08-06 2019-01-22 百度在线网络技术（北京）有限公司 Parking space drawing generating method, device, equipment and readable storage medium storing program for executing

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114093195A (en) * 2021-11-17 2022-02-25 深圳市豪位科技有限公司 Method and device for marking parking spaces
CN114440859A (en) * 2022-01-26 2022-05-06 广州小鹏自动驾驶科技有限公司 Map updating method and device, computer equipment and storage medium
CN114490910A (en) * 2022-01-26 2022-05-13 中国第一汽车股份有限公司 Map generation method and device, electronic equipment and storage medium
CN114440859B (en) * 2022-01-26 2024-08-09 广州小鹏自动驾驶科技有限公司 Map updating method, map updating device, computer equipment and storage medium
CN114743405A (en) * 2022-03-29 2022-07-12 广州小鹏自动驾驶科技有限公司 Parking space data processing method and device, electronic equipment and storage medium
CN114743405B (en) * 2022-03-29 2023-10-31 广州小鹏自动驾驶科技有限公司 Parking space data processing method and device, electronic equipment and storage medium
CN114937369A (en) * 2022-04-06 2022-08-23 福思(杭州)智能科技有限公司 Autonomous valet parking method, system, vehicle, parking lot terminal, device and medium
CN117495847A (en) * 2023-12-27 2024-02-02 安徽蔚来智驾科技有限公司 Intersection detection method, readable storage medium and intelligent device
CN117495847B (en) * 2023-12-27 2024-03-19 安徽蔚来智驾科技有限公司 Intersection detection method, readable storage medium and intelligent device
WO2025139575A1 (en) * 2023-12-27 2025-07-03 安徽蔚来智驾科技有限公司 Intersection detection method, readable storage medium and intelligent device

Also Published As

Publication number Publication date
CN112950972B (en) 2022-07-01

Similar Documents

Publication Publication Date Title
CN112950972B (en) Parking lot map construction method, device, equipment and medium
CN110501018B (en) Traffic sign information acquisition method for high-precision map production
US11131999B2 (en) Method and apparatus for identifying laser point cloud data of autonomous vehicle
US11670087B2 (en) Training data generating method for image processing, image processing method, and devices thereof
EP3524936B1 (en) Method and apparatus providing information for driving vehicle
CN109584706B (en) Electronic map lane line processing method, device and computer readable storage medium
US10118614B2 (en) Detailed map format for autonomous driving
CN110287276A (en) High-precision map updating method, device and storage medium
CN112212874B (en) Vehicle track prediction method and device, electronic equipment and computer readable medium
US11288526B2 (en) Method of collecting road sign information using mobile mapping system
US11231283B2 (en) Localization with neural network based image registration of sensor data and map data
WO2019173009A1 (en) Automatic identification of roadside objects for localization
Agrawal et al. Localization and mapping for autonomous navigation in outdoor terrains: A stereo vision approach
CN104766058A (en) Method and device for obtaining lane line
CN111177288A (en) System for deriving autonomous vehicle enabled drivable maps
WO2022147924A1 (en) Method and apparatus for vehicle positioning, storage medium, and electronic device
CN113358125B (en) Navigation method and system based on environment target detection and environment target map
CN114518122B (en) Driving navigation method, device, computer equipment, storage medium and computer program product
CN112735253B (en) Traffic light automatic labeling method and computer equipment
CN108805930A (en) The localization method and system of automatic driving vehicle
KR20200102378A (en) Information processing method and device and storage medium
WO2022082571A1 (en) Lane line detection method and apparatus
Gressenbuch et al. Mona: The munich motion dataset of natural driving
CN114694116A (en) A road boundary detection method, device, electronic device and storage medium
US20240119740A1 (en) Method and apparatus with lane line determination

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right

Effective date of registration: 20240304

Address after: 510000 No.8 Songgang street, Cencun, Tianhe District, Guangzhou City, Guangdong Province

Patentee after: GUANGZHOU XIAOPENG MOTORS TECHNOLOGY Co.,Ltd.

Country or region after: China

Address before: Room 46, room 406, No.1, Yichuang street, Zhongxin knowledge city, Huangpu District, Guangzhou City, Guangdong Province 510000

Patentee before: Guangzhou Xiaopeng Automatic Driving Technology Co.,Ltd.

Country or region before: China

TR01 Transfer of patent right