Point cloud map construction method, device, equipment, storage medium and computer program
By chunking the map construction area of the vehicle system and optimizing the position information with local point cloud maps, the problem of combined inertial guide being affected by weather and object occlusion is solved, and the accuracy and composition consistency of high-precision maps are improved.
Patent Information
- Application Number
- CN202111314754.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-08
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-11-08
AI Technical Summary
The vehicle position information output by the combined inertial navigation is easily affected by factors such as weather and object occlusion, resulting in a decrease in the accuracy of high-precision maps.
By chunking the entire map construction area, a blocked point cloud map is built for each block area, and the position information is optimized in combination with the local point cloud maps that have been constructed in the historical time period to ensure the accuracy of the position information.
It improves the accuracy of high-precision maps, avoids the inaccuracy of pose information under the influence of combined inertial navigation factors, and ensures global consistency and local smoothness.
Smart Images

Figure CN114092638B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of map construction technology, and in particular to a point cloud map construction method, device, equipment, storage medium and computer program. Background Art
[0002] High-precision maps have broad application value in application areas such as autonomous driving, smart cities, the Internet of Things, and other small but precise vertical applications. They can be generated by collecting point cloud data from vehicle-mounted laser radar.
[0003] For example, when a vehicle is driving, the position and posture information of the vehicle at each time is recorded through the combined inertial navigation, and the three-dimensional point cloud data at each time is collected through the lidar. Then, the three-dimensional point cloud data are spliced based on the above position and posture information to generate a high-precision map.
[0004] However, the vehicle's position information output by the combined inertial navigation is easily affected by factors such as weather and object occlusion, resulting in reduced accuracy of high-precision maps. Summary of the invention
[0005] The embodiment of the present application provides a point cloud map construction method, device, equipment, storage medium and computer program, which can avoid the problem that the combined inertial navigation is easily affected by factors such as weather and object occlusion, and the output posture information is inaccurate. The technical solution is as follows.
[0006] According to one aspect of the present application, a point cloud map construction method is provided, the method comprising:
[0007] Acquire 3D point cloud data. The 3D point cloud data is obtained by scanning the environment in the block area by a map scanning device. The block area is a sub-area obtained by dividing the entire mapping area;
[0008] Associating the 3D point cloud data with the position information of the map scanning device according to the timestamp, the position information is used to represent the motion posture of the map scanning device during the 3D point cloud data collection process;
[0009] When the confidence of the pose information is less than the confidence threshold, the pose information is optimized through the block point cloud map and the local point cloud map to obtain the optimized pose information; the block point cloud map refers to the point cloud map of the block area, and the local point cloud map is obtained by splicing part or all of the constructed block point cloud maps;
[0010] Based on the optimized pose information, the 3D point cloud data is stitched into block point cloud maps and local point cloud maps.
[0011] According to another aspect of the present application, a point cloud map construction device is provided, the device comprising:
[0012] An acquisition module is used to acquire three-dimensional point cloud data. The three-dimensional point cloud data is obtained by scanning the environment in the block area by a map scanning device. The block area is a sub-area obtained by dividing the entire mapping area;
[0013] An association module, used to associate the three-dimensional point cloud data with the posture information of the map scanning device according to the timestamp, and the posture information is used to represent the motion posture of the map scanning device during the three-dimensional point cloud data collection process;
[0014] The optimization module is used to optimize the pose information through the block point cloud map and the local point cloud map to obtain the optimized pose information when the confidence of the pose information is less than the confidence threshold; the block point cloud map refers to the point cloud map of the block area, and the local point cloud map is obtained by splicing part or all of the constructed block point cloud maps;
[0015] A construction module is used to stitch 3D point cloud data into block point cloud maps and local point cloud maps based on optimized pose information.
[0016] According to another aspect of the present application, a computer device is provided, comprising: a processor and a memory, wherein the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the point cloud map construction method as described above.
[0017] According to another aspect of the present application, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is loaded and executed by a processor to implement the point cloud map construction method as described above.
[0018] According to another aspect of the present application, a computer program (or computer program product) is provided, wherein the computer program (or computer program product) includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the point cloud map construction method as described above.
[0019] The beneficial effects brought by the technical solution provided by the embodiment of the present application include at least:
[0020] In the method, the construction of the point cloud map first divides the entire mapping area into blocks, and then constructs a block point cloud map for each block area; and while constructing the block point cloud map, a local point cloud map is also constructed. The local point cloud map here is spliced by part or all of the block point cloud maps that have been constructed in the historical time period, and is not exactly the same as each block point cloud map. In the process of constructing the point cloud map, if the confidence of the pose information is less than the confidence threshold, the pose information is optimized by the block point cloud map and the local point cloud map, so that the accuracy of the pose information can be guaranteed, thereby ensuring the accuracy of the composition according to the pose information, so as to avoid the combined inertial navigation being affected by factors such as weather and object occlusion, and the output pose information is inaccurate, which in turn affects the accuracy of the high-precision map constructed based on its output pose information; wherein the combined inertial navigation is composed of a global positioning system (Global Positioning System, GPS) and an inertial sensor (Inertial Measurement Unit, IMU).
[0021] Secondly, by mapping the entire mapping area in blocks, we can avoid the error accumulation caused by using only local point cloud map construction method, and ensure the global consistency of the entire composition; combining local point cloud map can ensure the smoothness of local composition. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0023] Figure 1 A schematic diagram of the structure of a vehicle-mounted system provided by an exemplary embodiment of the present application is shown;
[0024] Figure 2 A flowchart of a point cloud map construction method provided by an exemplary embodiment of the present application is shown;
[0025] Figure 3 A flowchart of a point cloud map construction method provided by another exemplary embodiment of the present application is shown;
[0026] Figure 4 A flowchart of a point cloud map construction method provided by another exemplary embodiment of the present application is shown;
[0027] Figure 5 A flowchart of a point cloud map construction method provided by another exemplary embodiment of the present application is shown;
[0028] Figure 6 A flowchart of a point cloud map construction method provided by another exemplary embodiment of the present application is shown;
[0029] Figure 7 A flowchart of a point cloud map construction method provided by another exemplary embodiment of the present application is shown;
[0030] Figure 8 A block diagram of a point cloud map construction device provided by an exemplary embodiment of the present application is shown;
[0031] Fig. 9 A schematic diagram of the structure of a computer device provided by an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION
[0032] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.
[0033] First, several terms involved in this application are introduced.
[0034] 3D point cloud data refers to a set of vectors in a 3D coordinate system. These vectors are usually expressed in the form of x, y, z 3D coordinates and are generally used to represent the outer surface shape of an object. For example, P k ={x k ,y k ,z k} represents the kth point in space, Point Cloud = {P 1 ,P 2 ,P 3 ,…,P v} represents a set of point cloud data, k and v are positive integers. Exemplarily, the above three-dimensional point cloud data is generated by a three-dimensional scanning device, such as a lidar, a stereo camera, or a time-of-flight camera.
[0035] The map scanning device can obtain three-dimensional point cloud data by scanning the environment, and then the computer device accumulates the three-dimensional point cloud data frame by frame to form a point cloud map. Among them, the map scanning device is a three-dimensional scanning device. For each frame of three-dimensional point cloud data, the position of the map scanning device when it is collected is the basis for splicing this frame of three-dimensional point cloud data with other point cloud data.
[0036] An object has six degrees of freedom in space, namely the freedom of movement along the three rectangular coordinate axes of x, y, and z and the freedom of rotation around these three coordinate axes. The degrees of freedom of rotation are heading, pitch, and roll. Usually, the on-board laser radar on the map collection vehicle directly uses GPS+IMU to obtain the six degrees of freedom of the laser radar. This method is highly dependent on GPS, which is easily affected by factors such as weather and objects, resulting in inaccurate posture output by GPS+IMU, and then the splicing of three-dimensional point cloud data is ghosted; there may also be GPS positioning jumps, resulting in faults in the splicing of three-dimensional point cloud data; all of these will reduce the accuracy of the constructed point cloud map.
[0037] There is also a composition method that builds maps based on local point cloud data. This method is prone to error accumulation. In the case of large-scale mapping, it is even impossible to form a closed loop, making it impossible to perform global corrections.
[0038] In order to solve the above problems and improve the accuracy of the constructed point cloud map, the present application provides a point cloud map construction method. For detailed implementation details of the method, please refer to the following embodiments.
[0039] For example, the above point cloud map construction method can be applied to a vehicle-mounted system. Figure 1 , shows a schematic diagram of a vehicle-mounted system involved in an embodiment of the present application. The vehicle-mounted system includes: a global navigation satellite system (GNSS) 110, a laser radar 120, an IMU 130, a camera 140 and a host computer 150. Among them, the host computer 150 is a computer device.
[0040] For the collection of raw data, GNSS 110 is used to record the location information of the map collection vehicle and the coordinates of the current collection point; LiDAR 120 is used to collect three-dimensional point cloud data and can scan the surrounding environment within a preset radius; IMU 130 is used to capture the angle and acceleration information of the map collection vehicle to correct the position and angle of the map collection vehicle; camera 140 is used to collect road signs, lane lines, etc.
[0041] The above-mentioned GNSS 110, laser radar 120, IMU 130 and camera 140 are respectively connected to the host computer 150, and the collected raw data is transmitted to the host computer 150 through the above-mentioned communication connection. The host computer 150 builds a high-precision map based on the raw data collected by the GNSS 110, the laser radar 120, the IMU 130 and the camera 140. Exemplarily, in the process of building a high-precision map, the host computer 150 executes the point cloud map construction method provided in this application to build a high-precision point cloud map based on the posture information output by the GNSS 110+IMU 130 and the three-dimensional point cloud data output by the laser radar 120, and then builds a high-precision map on the basis of the point cloud map to improve the accuracy of the high-precision map.
[0042] Figure 2 A flowchart of a point cloud map construction method provided by an exemplary embodiment of the present application is shown. Figure 1 In the computer device shown, the method comprises the following steps:
[0043] Step 201, obtaining three-dimensional point cloud data, the three-dimensional point cloud data is obtained by scanning the environment in the block area by a map scanning device, and the block area is a sub-area obtained by dividing the entire mapping area.
[0044] In the process of constructing the point cloud map, the computer device is provided with multiple block areas for dividing the entire mapping area. For each block area, the map collection vehicle drives in a block area, and the map scanning device installed on the map collection vehicle scans the environment to obtain the three-dimensional point cloud data of the block area; at the same time, the combined inertial navigation system installed on the map collection vehicle outputs the position and posture information of the map collection vehicle, and this position and posture information is used as the position and posture information of the map scanning device. The computer device receives the three-dimensional point cloud data transmitted by the map scanning device, that is, obtains the three-dimensional point cloud data of the block area; and obtains the position and posture information of the map scanning device from the combined inertial navigation system.
[0045] Exemplarily, the map scanning device can store the collected three-dimensional point cloud data to a first designated storage area of the memory, and the combined inertial navigation can output the posture information to a second designated storage area of the memory; the computer device obtains the three-dimensional point cloud data within a specified time period from the first designated storage area, and obtains the posture information within the above-specified time period from the second designated storage area.
[0046] Step 202 , associating the three-dimensional point cloud data with the position and posture information of the map scanning device according to the timestamp.
[0047] The pose information is used to represent the motion posture of the map scanning device during the three-dimensional point cloud data collection process.
[0048] Each frame of 3D point cloud data is marked with a first timestamp, which is used to indicate the acquisition time of the frame of 3D point cloud data; each pose information is marked with a second timestamp, which is used to indicate the acquisition time of the pose information. The computer device associates the 3D point cloud data and pose information whose first timestamp is consistent with the second timestamp.
[0049] Step 203, when the confidence of the pose information is less than or equal to the confidence threshold, the pose information is optimized by using the block point cloud map and the local point cloud map to obtain optimized pose information.
[0050] The block point cloud map refers to the point cloud map of the block area. The block point cloud map here can refer to a constructed block point cloud map or an unconstructed block point cloud map.
[0051] A local point cloud map is a point cloud map of a local area in the entire mapping area. The local point cloud map here can be obtained by splicing part or all of the constructed block point cloud maps; for each constructed block point cloud map, part or all of them can be used to splice to form a local point cloud map.
[0052] The constructed block point cloud map includes: the block point cloud map constructed on the historical time period adjacent to the timestamp, which is the timestamp of the 3D point cloud data to be spliced. For example, if the timestamp of the 3D point cloud data is 2021 / 11 / 02 / 12:03:55 (i.e. 12:03:55 on November 2, 2021), the historical time period can be 11:03:55 to 12:03:54.
[0053] The computer device splices each frame of three-dimensional point cloud data in a time sequence. When splicing a frame of three-dimensional point cloud data, it first determines whether the confidence of the pose information associated with the frame of three-dimensional point cloud data is greater than the confidence threshold, wherein the confidence of the pose information is used to characterize the accuracy of the pose information; only when the confidence of the pose information is greater than the confidence threshold, can it be confirmed that the pose information is accurate, thereby ensuring that the frame of three-dimensional point cloud data can be spliced at the correct position based on the pose information; when the confidence of the pose information is less than or equal to the confidence threshold, the pose information is optimized through the block point cloud map and the local point cloud map to obtain optimized pose information, that is, the pose information is corrected to maximize the accuracy of the pose information, thereby ensuring that the frame of three-dimensional point cloud data can be spliced at the correct position based on the pose information.
[0054] For posture optimization, the computer equipment can use a simultaneous localization and mapping (SLAM) closed-loop algorithm to optimize the posture information associated with the three-dimensional point cloud data to obtain optimized posture information; or, the BA optimization algorithm can be used to optimize the posture information associated with the three-dimensional point cloud data to obtain optimized posture information. BA is also known as the bundle adjustment method (Bundle Adjustment).
[0055] For example, a computer device calls a SLAM closed-loop algorithm to optimize the pose information based on the three-dimensional point cloud data and the point cloud data in the block point cloud map to obtain the first pose information; optimizes the pose information based on the three-dimensional point cloud data and the point cloud data in the local point cloud map to obtain the second pose information; and then determines the optimized pose information based on the first pose information and the second pose information.
[0056] Step 204 , stitching the three-dimensional point cloud data into a block point cloud map and a local point cloud map based on the optimized posture information.
[0057] For each frame of three-dimensional point cloud data, the computer device synchronously stitches the three-dimensional point cloud data to the block point cloud map and the local point cloud map according to the optimized posture information; or stitches the three-dimensional point cloud data to the block point cloud map according to the optimized posture information, and then stitches the three-dimensional point cloud data to the local point cloud map; or stitches the three-dimensional point cloud data to the local point cloud map according to the optimized posture information, and then stitches the three-dimensional point cloud data to the block point cloud map.
[0058] To sum up, the point cloud map construction method provided in this embodiment first divides the entire mapping area into blocks, and then constructs a block point cloud map for each block area; and while constructing the block point cloud map, a local point cloud map is also constructed. The local point cloud map here is spliced by part or all of the block point cloud maps that have been constructed in the historical time period, and is not exactly the same as each block point cloud map. In the process of constructing the point cloud map, if the confidence of the pose information is less than the confidence threshold, the pose information is optimized together by the block point cloud map and the local point cloud map, which can ensure the accuracy of the pose information, thereby ensuring the accuracy of the composition according to the pose information, thereby avoiding the combined inertial navigation being affected by factors such as weather and object occlusion, and the output pose information is inaccurate, which in turn affects the accuracy of the high-precision map constructed based on its output pose information.
[0059] Secondly, by mapping the entire mapping area in blocks, we can avoid the error accumulation caused by using only local point cloud map construction method, and ensure the global consistency of the entire composition; combining local point cloud map can ensure the smoothness of local composition.
[0060] Exemplarily, before associating the three-dimensional point cloud data with the pose information, the computer device pre-processes the acquired pose information, determines the confidence of the pose information, and marks the confidence of the pose information. Optionally, the confidence includes the signal quality of the pose information acquisition, and accordingly, the confidence threshold includes the signal quality threshold. Figure 3 , Figure 2 Step 203 can be implemented by step 301 as follows:
[0061] Step 301, when the signal quality corresponding to the posture information is less than the signal quality threshold, the posture information is optimized through the block point cloud map and the local point cloud map to obtain optimized posture information.
[0062] The computer device obtains the posture information and the signal quality corresponding to the posture information, and then associates the posture information with the signal quality, or generates the signal quality as marking information carried by the posture information.
[0063] The signal quality here may be the signal quality of the GNSS (or GPS) positioning signal when the posture information is collected. For example, it may be the signal strength and / or signal to noise ratio (SNR) of the GNSS (or GPS) positioning signal when the posture information is collected.
[0064] There is also a situation where the computer device pre-processes the posture information, determines that the confidence of the position information is greater than the confidence threshold, or is less than or equal to the confidence threshold, and marks the posture information with a confidence greater than the confidence threshold as a fixed posture. Exemplarily, when the confidence of the posture information is greater than the confidence threshold, the computer device marks the posture information as a fixed posture. Then, step 203 can also be implemented through step 302, such as Figure 4 As shown:
[0065] Step 302: When there is no fixed pose mark in the pose information, the pose information is optimized by using the block point cloud map and the local point cloud map to obtain optimized pose information.
[0066] After obtaining the posture information, the computer device determines whether the posture information carries a mark of a fixed posture (fix); if not, the step of "optimizing the posture information through the block point cloud map and the local point cloud map to obtain optimized posture information" is executed.
[0067] In this way, the above method can ensure the accuracy of the pose information of each frame of three-dimensional point cloud data, so that each frame of three-dimensional point cloud data can be accurately spliced into the block point cloud map and the local point cloud map.
[0068] like Figure 5 , the optimization of the pose information associated with each frame of three-dimensional point cloud data can also be implemented by using steps 401 to 403, as shown below:
[0069] Step 401, when the confidence of the pose information is less than or equal to the confidence threshold, the first pose information of the map scanning device at the timestamp is estimated based on the three-dimensional point cloud data and the point cloud data in the block point cloud map.
[0070] Exemplarily, a computer device randomly generates at least two candidate poses based on the original pose information (i.e., the pose information associated with the three-dimensional point cloud data); calculates the distribution probability of each of the at least two candidate poses based on the three-dimensional point cloud data and the point cloud data in the block point cloud map, and obtains a first distribution probability for each candidate pose, wherein the first distribution probability of the candidate pose represents the probability that the candidate pose is the standard pose information of the three-dimensional point cloud data in the block point cloud map; determines the candidate pose corresponding to the largest first distribution probability as the first pose information, and records the first distribution probability corresponding to the first pose information, i.e., the first estimated probability, which refers to the distribution probability of the first pose information during pose optimization.
[0071] Step 402, when the confidence of the posture information is less than or equal to the confidence threshold, the second posture information of the map scanning device at the timestamp is estimated based on the three-dimensional point cloud data and the point cloud data in the local point cloud map.
[0072] The computer device calculates the distribution probability of each of at least two candidate postures based on the three-dimensional point cloud data and the point cloud data in the local point cloud map, and obtains the second distribution probability of each candidate posture, and the second distribution probability of the candidate posture represents the probability that the candidate posture is the standard posture information of the above three-dimensional point cloud data in the local point cloud map; the candidate posture corresponding to the largest second distribution probability is determined as the second posture information, and the second distribution probability corresponding to the second posture information is recorded, that is, the second estimated probability, which refers to the distribution probability of the second posture information during posture optimization.
[0073] Step 403: Determine optimized posture information based on the first posture information and the second posture information.
[0074] When the first estimated probability is greater than the second estimated probability, the first pose information is determined as the optimized pose information; or, when the second estimated probability is greater than the first estimated probability, the second pose information is determined as the optimized pose information; or, when the first estimated probability is equal to the second estimated probability, the first pose information or the second pose information is determined as the optimized pose information.
[0075] Optionally, after obtaining the first pose information and the second pose information, the computer device may also perform pose fusion on the first pose information and the second pose information to obtain optimized pose information.
[0076] The method provides a pose information optimization method that uses a block point cloud map and a local point cloud map to jointly optimize the pose information, so that the optimized pose information can be more accurate; in particular, the fusion of the first pose information and the second pose information ensures global consistency while also ensuring the smoothness of the local composition.
[0077] The map collection vehicle is usually driven by staff. Therefore, for each block area, which includes an open area and an obstructed area, the map scanning device can be manually controlled to first collect the 3D point cloud data of the open area in the block area, that is, to obtain the first point cloud data, and then collect the 3D point cloud data of the obstructed area in the block area, that is, to obtain the second point cloud data; for example, for common fast and wide motor vehicle lanes and non-motor vehicle lanes obstructed by trees, the 3D point cloud data of the motor vehicle lane is collected first, and then the 3D point cloud data of the non-motor vehicle lane is collected. Therefore, the first collection time of the first point cloud data takes precedence over the collection time of the second point cloud data.
[0078] In this way, during the mapping process, the 3D point cloud data in the open area is spliced first to complete a part of the block point cloud map, and then the 3D point cloud data in the occluded area is spliced to complete the other part of the block point cloud map. This composition method can give priority to obtaining 3D point cloud data corresponding to high-quality (i.e., high-confidence) posture information, accurately construct a part of the block point cloud map, and ensure the accuracy of this part of the map; and when the confidence of the posture information is high enough, there is no need to perform posture optimization, which also improves the efficiency of building the block point cloud map.
[0079] Since the construction of local point cloud maps is prone to error accumulation, in order to avoid this, the computer device detects the movement distance of the map scanning device (that is, the movement distance of the map collection vehicle) in real time during the construction of the local point cloud map; when the movement distance is greater than the distance threshold, the local point cloud map is truncated to construct a new local point cloud map. The movement distance refers to the cumulative movement distance from the start of the construction of the local point cloud map. The movement distance here can include the linear movement distance of the map scanning device from the start of the construction of the local point cloud map.
[0080] In another embodiment, the computer device can estimate in real time the plane map area corresponding to the local point cloud map during the construction of the local point cloud map; when the plane map area is larger than the area threshold, the local point cloud map is truncated to construct a new local point cloud map.
[0081] For the construction of a new local point cloud map, the computer device can obtain the constructed block point cloud map adjacent to the geographical location of the map scanning device based on the geographical location of the map scanning device, wherein the constructed block point cloud map can be a completely constructed block point cloud map or a partially constructed block point cloud map; a new local point cloud map is generated based on the splicing of adjacent constructed block point cloud maps.
[0082] Alternatively, the computer device determines the target timestamp closest to the current time in the timestamp of the spliced three-dimensional point cloud map (or directly determines the current time as the target timestamp), obtains the constructed block point cloud map generated in a specified time period adjacent to the target timestamp, and generates a new local point cloud map based on the splicing of the constructed block point cloud maps in the above specified time period.
[0083] Alternatively, the computer device determines the target timestamp closest to the current time in the timestamp of the spliced three-dimensional point cloud map (or directly determines the current time as the target timestamp), determines the partial local point cloud map generated in a specified time period adjacent to the target timestamp, and removes the other partial local point cloud maps except the partial local point cloud maps in the local point cloud map to obtain a new local point cloud map, that is, retains the partial local point cloud map as the new local point cloud map.
[0084] This method can combine the local point cloud map to ensure the local smoothness of the composition while reducing the error accumulation of the local point cloud map.
[0085] In the process of composing, it is possible to repeatedly enter the same block area. At this time, the computer device can correct the block point cloud map corresponding to the block area, such as Figure 6 , the steps are as follows:
[0086] Step 501, when entering a block area, traverse the list of block areas for which the point cloud map has been constructed.
[0087] A list of block areas for which point cloud maps have been constructed is set in the computer device; during the composition process, the computer device determines that it has entered a new block area based on the geographical location of the map scanning device, and then traverses the list of block areas for which point cloud maps have been constructed to determine whether there is a block point cloud map for the block area.
[0088] Step 502: if there are block areas in the list, load the block point cloud map of the block areas from historical data.
[0089] When there is a block area in the list, the computer device searches for data of the block point cloud map of the block area from the historical data stored in the memory, and loads the block point cloud map of the block area.
[0090] Step 503: After acquiring the three-dimensional point cloud data, calibrate the block point cloud map based on the three-dimensional point cloud data.
[0091] After loading the block point cloud map of the block area, the computer device also obtains three-dimensional point cloud data and corrects the block point cloud map based on the three-dimensional point cloud data.
[0092] There is another situation. The loaded block point cloud map may be a block point cloud map that has been partially constructed for the block area, that is, this block point cloud map is only partially constructed, not the entire block area. At this time, the computer device corrects the block point cloud map based on the three-dimensional point cloud data, and also executes the point cloud map construction method provided in the above embodiment to continue to construct the block point cloud map based on other three-dimensional point cloud data.
[0093] Step 504: if there is no block area in the list, construct a block point cloud map of the block area based on the three-dimensional point cloud data.
[0094] When there is no block area in the list, the computer device executes the point cloud map construction method provided in the above embodiment to construct the block point cloud map based on the three-dimensional point cloud data.
[0095] This method can avoid repeatedly constructing the block point cloud map for the composed block area.
[0096] There is also pose information associated with the 3D point cloud data, and its confidence is greater than the confidence threshold. In this case, you can execute Figure 7 Step 205 in is as follows:
[0097] Step 205 , when the confidence level is greater than the confidence level threshold, the three-dimensional point cloud data is spliced into a block point cloud map and a local point cloud map based on the posture information.
[0098] When the signal quality of the posture information is greater than the confidence threshold, the computer device performs motion compensation on the posture information and then directly splices the three-dimensional point cloud data into the block point cloud map and the local point cloud map based on the posture information.
[0099] Exemplarily, when the posture information is marked as a fixed posture, the computer device splices the three-dimensional point cloud data into a block point cloud map and a local point cloud map based on the posture information.
[0100] This point cloud map construction method can directly use the pose information to stitch the point cloud data when the pose information is credible enough, thus saving the pose optimization step. The more pose information there is, the faster the point cloud map can be constructed.
[0101] Figure 8 A block diagram of a point cloud map construction device provided by an exemplary embodiment of the present application is shown. The device can be implemented as part or all of a computer device through software, hardware, or a combination of both.
[0102] The device includes:
[0103] An acquisition module 601 is used to acquire three-dimensional point cloud data, wherein the three-dimensional point cloud data is obtained by scanning the environment in a block area by a map scanning device, and the block area is a sub-area obtained by dividing the entire mapping area;
[0104] An association module 602 is used to associate the three-dimensional point cloud data with the posture information of the map scanning device according to the timestamp, wherein the posture information is used to represent the motion posture of the map scanning device during the three-dimensional point cloud data collection process;
[0105] The optimization module 603 is used to optimize the pose information by using a block point cloud map and a local point cloud map to obtain optimized pose information when the confidence of the pose information is less than a confidence threshold; the block point cloud map refers to a point cloud map of the block area, and the local point cloud map is obtained by splicing part or all of the constructed block point cloud maps;
[0106] The construction module 604 is used to splice the three-dimensional point cloud data into the block point cloud map and the local point cloud map based on the optimized posture information.
[0107] In some embodiments, the optimization module 603 is used to:
[0108] According to the three-dimensional point cloud data and the point cloud data in the block point cloud map, the first position information of the map scanning device at the timestamp is estimated; and
[0109] estimating second position information of the map scanning device at the timestamp based on the three-dimensional point cloud data and the point cloud data in the local point cloud map;
[0110] The optimized posture information is determined based on the first posture information and the second posture information.
[0111] In some embodiments, the optimization module 603 is used to:
[0112] When the first estimated probability is greater than the second estimated probability, the first posture information is determined as the optimized posture information; or, when the second estimated probability is greater than the first estimated probability, the second posture information is determined as the optimized posture information; wherein the first estimated probability refers to the distribution probability of the first posture information when the posture is optimized, and the second estimated probability refers to the distribution probability of the second posture information when the posture is optimized;
[0113] Alternatively, the first posture information and the second posture information are posture fused to obtain the optimized posture information.
[0114] In some embodiments, the block area includes an open area and an obstructed area, and accordingly, the three-dimensional point cloud data includes first point cloud data collected from the open area and second point cloud data collected from the obstructed area;
[0115] A first acquisition time of the first point cloud data is earlier than a second acquisition time of the second point cloud data.
[0116] In some embodiments, the apparatus further includes: a detection module 605 and a truncation module 606;
[0117] A detection module 605 is used to detect the movement distance of the map scanning device in real time during the construction of the local point cloud map; wherein the movement distance refers to the cumulative movement distance from the start of the construction of the local point cloud map;
[0118] The truncation module 606 is used to truncate the local point cloud map to construct a new local point cloud map when the movement distance is greater than a distance threshold.
[0119] In some embodiments, the apparatus further includes: a detection module 605 and a truncation module 606;
[0120] A detection module 605 is used to estimate in real time the plane map area corresponding to the local point cloud map during the construction of the local point cloud map;
[0121] The truncation module 606 is used to truncate the local point cloud map to construct a new local point cloud map when the area of the plane map is greater than the area threshold.
[0122] In some embodiments, the apparatus further includes: a detection module 605, a loading module 607, and a correction module 608;
[0123] A detection module 605 is used for traversing a list of block areas for which a point cloud map has been constructed when entering the block area;
[0124] A loading module 607 is used to load the block point cloud map of the block area from historical data when the block area exists in the list;
[0125] The correction module 608 is used to correct the block point cloud map based on the three-dimensional point cloud data after acquiring the three-dimensional point cloud data.
[0126] In some embodiments, the construction module 604 is used to splice the three-dimensional point cloud data to the block point cloud map and the local point cloud map based on the pose information when the confidence is greater than the confidence threshold.
[0127] In some embodiments, the confidence level includes the signal quality of the posture information acquisition, and accordingly, the confidence level threshold includes a signal quality threshold.
[0128] To sum up, the point cloud map construction device provided in this embodiment divides the entire mapping area into blocks, and then constructs a block point cloud map for each block area; and while constructing the block point cloud map, it also constructs a local point cloud map, where the local point cloud map is spliced by part or all of the block point cloud maps that have been constructed in the historical time period, and is not exactly the same as each block point cloud map. In the process of constructing the point cloud map, if the confidence of the pose information is less than the confidence threshold, the pose information is optimized together by the block point cloud map and the local point cloud map, which can ensure the accuracy of the pose information, thereby ensuring the accuracy of the composition according to the pose information, thereby avoiding the combined inertial navigation being affected by factors such as weather and object occlusion, and the output pose information is inaccurate, which in turn affects the accuracy of the high-precision map constructed based on its output pose information.
[0129] Secondly, by mapping the entire mapping area in blocks, we can avoid the error accumulation caused by using only local point cloud map construction method, and ensure the global consistency of the entire composition; combining local point cloud map can ensure the smoothness of local composition.
[0130] Fig. 9 A schematic diagram of the structure of a computer device provided by an exemplary embodiment of the present application is shown. The computer device may be a device for executing the point cloud map construction method provided by the present application.
[0131] The computer device 700 includes a central processing unit (CPU) 701, a system memory 704 including a random access memory (RAM) 702 and a read only memory (ROM) 703, and a system bus 705 connecting the system memory 704 and the central processing unit 701. The computer device 700 also includes a basic input / output system (I / O system) 706 for facilitating information transmission between various components in the computer, and a large-capacity storage device 707 for storing an operating system 713, application programs 714, and other program modules 715.
[0132] The basic input / output system 706 includes a display 708 for displaying information and an input device 709 such as a mouse and a keyboard for user inputting information. The display 708 and the input device 709 are connected to the central processing unit 701 through an input / output controller 710 connected to the system bus 705. The basic input / output system 706 may also include an input / output controller 710 for receiving and processing inputs from a plurality of other devices such as a keyboard, a mouse, or an electronic stylus. Similarly, the input / output controller 710 also provides output to a display screen, a printer, or other types of output devices.
[0133] The mass storage device 707 is connected to the central processing unit 701 through a mass storage controller (not shown) connected to the system bus 705. The mass storage device 707 and its associated computer readable media provide non-volatile storage for the computer device 700. That is, the mass storage device 707 may include a computer readable medium (not shown) such as a hard disk or a compact disc read only memory (CD-ROM) drive.
[0134] Computer readable media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer readable instructions, data structures, program modules or other data. Computer storage media include RAM, ROM, Erasable Programmable Read Only Memory (EPROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash memory or other solid-state storage technologies, CD-ROM, Digital Versatile Disc (DVD) or Solid State Drive (SSD), other optical storage, cassettes, tapes, disk storage or other magnetic storage devices. Among them, random access memory may include resistance random access memory (ReRAM) and dynamic random access memory (DRAM). Of course, those skilled in the art will know that computer storage media are not limited to the above. The above-mentioned system memory 704 and large-capacity storage device 707 can be collectively referred to as memory.
[0135] According to various embodiments of the present application, the computer device 700 can also be connected to a remote computer on the network through a network such as the Internet. That is, the computer device 700 can be connected to the network 712 through the network interface unit 711 connected to the system bus 705, or the network interface unit 711 can be used to connect to other types of networks or remote computer systems (not shown).
[0136] The above-mentioned memory also includes one or more programs, and the one or more programs are stored in the memory and configured to be executed by the CPU.
[0137] In an optional embodiment, a computer device is provided, which includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the point cloud map construction method as described above.
[0138] In an optional embodiment, a computer-readable storage medium is provided, in which at least one instruction, at least one program, code set or instruction set is stored, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the point cloud map construction method as described above.
[0139] Optionally, the computer readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a solid state drive (SSD), or an optical disk. Among them, the random access memory may include a resistance random access memory (ReRAM) and a dynamic random access memory (DRAM). The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.
[0140] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.
[0141] The present application also provides a computer-readable storage medium, which stores at least one instruction, at least one program, code set or instruction set. The at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the point cloud map construction method provided by the above-mentioned method embodiments.
[0142] The present application also provides a computer program product, the computer program product includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the point cloud map construction method as described above.
[0143] It should be understood that the "plurality" mentioned in this article refers to two or more. "And / or" describes the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.
[0144] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.
[0145] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. A point cloud map construction method, characterized in that: The method comprises: acquiring three-dimensional point cloud data, wherein the three-dimensional point cloud data is obtained by scanning an environment in a block area by a map scanning device, and the block area is a sub-area obtained by dividing the entire mapping area; Associating the three-dimensional point cloud data with the position information of the map scanning device according to the timestamp, wherein the position information is used to represent the motion posture of the map scanning device during the three-dimensional point cloud data collection process; When the confidence of the pose information is less than or equal to the confidence threshold, the pose information is optimized by using a block point cloud map and a local point cloud map to obtain optimized pose information; the block point cloud map refers to a point cloud map of the block area, the block point cloud map refers to a constructed block point cloud map or an unconstructed block point cloud map, and the local point cloud map is spliced by part or all of the constructed block point cloud maps in a historical time period; Based on the optimized posture information, the three-dimensional point cloud data is spliced to the block point cloud map and the local point cloud map. When the confidence level is greater than the confidence level threshold, the three-dimensional point cloud data is spliced to the block point cloud map and the local point cloud map based on the posture information.
2. The method according to claim 1, characterized in that The optimized posture information is optimized by using a block point cloud map and a local point cloud map, including: estimating the first posture information of the map scanning device at the timestamp based on the three-dimensional point cloud data and the point cloud data in the block point cloud map; and estimating the second posture information of the map scanning device at the timestamp based on the three-dimensional point cloud data and the point cloud data in the local point cloud map; and determining the optimized posture information based on the first posture information and the second posture information.
3. The method according to claim 2, characterized in that The step of determining the optimized posture information based on the first posture information and the second posture information includes: when the first estimated probability is greater than the second estimated probability, determining the first posture information as the optimized posture information; or, when the second estimated probability is greater than the first estimated probability, determining the second posture information as the optimized posture information; wherein the first estimated probability refers to the distribution probability of the first posture information when the posture is optimized, and the second estimated probability refers to the distribution probability of the second posture information when the posture is optimized; or, performing posture fusion on the first posture information and the second posture information to obtain the optimized posture information.
4. The method according to any one of claims 1 to 3, characterized in that: The block area includes an open area and an obstructed area. Accordingly, the three-dimensional point cloud data includes first point cloud data collected from the open area and second point cloud data collected from the obstructed area; a first collection time of the first point cloud data is prior to a second collection time of the second point cloud data.
5. The method according to any one of claims 1 to 3, characterized in that: The method also includes: during the construction of the local point cloud map, detecting the movement distance of the map scanning device in real time; when the movement distance is greater than a distance threshold, truncating the local point cloud map to construct a new local point cloud map; wherein the movement distance refers to the cumulative movement distance from the start of construction of the local point cloud map.
6. The method according to any one of claims 1 to 3, characterized in that: The method also includes: during the construction process of the local point cloud map, real-time estimating the plane map area corresponding to the local point cloud map; when the plane map area is greater than an area threshold, truncating the local point cloud map to construct a new local point cloud map.
7. The method according to any one of claims 1 to 3, characterized in that: The method also includes: when entering the block area, traversing the list of block areas for which the point cloud map has been constructed; if the block area exists in the list, loading the block point cloud map of the block area from historical data; after acquiring the three-dimensional point cloud data, correcting the block point cloud map based on the three-dimensional point cloud data.
8. The method according to any one of claims 1 to 3, characterized in that: The confidence level includes the signal quality of the posture information collection, and accordingly, the confidence level threshold includes a signal quality threshold.
9. A point cloud map construction device, characterized in that: The device comprises: An acquisition module, used to acquire three-dimensional point cloud data, wherein the three-dimensional point cloud data is obtained by scanning the environment in a block area by a map scanning device, and the block area is a sub-area obtained by dividing the entire mapping area; an associating module, used to associate the three-dimensional point cloud data with the position information of the map scanning device according to the timestamp, wherein the position information is used to represent the motion posture of the map scanning device during the three-dimensional point cloud data collection process; An optimization module, used for optimizing the pose information through a block point cloud map and a local point cloud map to obtain optimized pose information when the confidence of the pose information is less than a confidence threshold; the block point cloud map refers to a point cloud map of the block area, the block point cloud map refers to a constructed block point cloud map or an unconstructed block point cloud map, and the local point cloud map is spliced by part or all of the constructed block point cloud maps in a historical time period; A construction module is used to stitch the three-dimensional point cloud data to the block point cloud map and the local point cloud map based on the optimized posture information, and when the confidence level is greater than the confidence level threshold, stitch the three-dimensional point cloud data to the block point cloud map and the local point cloud map based on the posture information.
10. A computer device, characterized in that: The computer device comprises: a processor and a memory, wherein the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the point cloud map construction method according to any one of claims 1 to 8.
11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is loaded and executed by a processor to implement the point cloud map construction method according to any one of claims 1 to 8.
12. A computer program product, characterized in that The computer program product includes computer instructions, which are stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the point cloud map construction method as described in any one of claims 1 to 8.
