CN115356747A - Multi-line laser radar obstacle identification method and device - Google Patents
Multi-line laser radar obstacle identification method and device Download PDFInfo
- Publication number
- CN115356747A CN115356747A CN202211279561.0A CN202211279561A CN115356747A CN 115356747 A CN115356747 A CN 115356747A CN 202211279561 A CN202211279561 A CN 202211279561A CN 115356747 A CN115356747 A CN 115356747A
- Authority
- CN
- China
- Prior art keywords
- obstacle
- longitudinal
- interval
- transverse
- data point
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
- G06V10/225—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition based on a marking or identifier characterising the area
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Electromagnetism (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The invention provides a method and a device for identifying obstacles by a multi-line laser radar, which comprises the steps of preprocessing laser radar data to obtain a longitudinal interval and a transverse interval; extracting longitudinal gradient obstacles in the longitudinal interval to obtain a longitudinal obstacle data point set; performing transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set; processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set; the processing amount and the calculation amount of radar data are greatly reduced, and the identification efficiency is improved.
Description
Technical Field
The invention relates to the technical field of laser radars, in particular to a method and a device for identifying obstacles by using a multi-line laser radar.
Background
When intelligent mobile equipment such as an unmanned driving device, an outdoor robot and the like works on an outdoor structured road or a semi-structured road, the identification of obstacles on a forward route is a very important problem, and the accuracy and timeliness of the detection directly determine whether normal driving can be realized and the driving efficiency. The intelligent mobile equipment such as the unmanned driving equipment and the outdoor robot makes driving decisions depending on the obstacle recognition result, so that the obstacle is avoided in normal driving, and the intelligent mobile equipment does not rub or collide with the obstacle. The existing multi-line laser radar has high data measurement precision and long measuring range, and contains rich three-dimensional information, so that the method can cover most of use scenes. However, the existing obstacle identification method based on the multiline laser radar basically adopts methods such as road surface extraction, deep learning or reinforcement learning, clustering processing and the like, and the methods have high calculation force requirements on a calculation platform, so that the identification efficiency is low.
In view of this, the present specification provides a method and an apparatus for recognizing an obstacle of a multiline laser radar, so as to greatly reduce the processing amount and the calculation amount of radar data and improve the recognition efficiency.
Disclosure of Invention
The invention aims to provide a multiline laser radar obstacle identification method which comprises the steps of preprocessing laser radar data to obtain a longitudinal interval and a transverse interval; extracting longitudinal gradient obstacles in the longitudinal interval to obtain a longitudinal obstacle data point set; performing transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set; and processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set.
Further, the laser radar data are preprocessed, and the preprocessing comprises dividing the laser radar data into a plurality of subintervals along the radius by taking the laser radar as the circle center; dividing each subinterval into a plurality of subinterval blocks based on the laser beams of different lines; wherein the plurality of subinterval blocks of each subinterval constitute the longitudinal interval; sub-interval blocks containing laser beams of the same line among the plurality of sub-intervals constitute the transverse interval; a center of gravity of a set of data points within each subinterval block is calculated, by which center of gravity the position of the set of data points is represented.
Further, the longitudinal gradient obstacle extraction is carried out on the longitudinal interval to obtain a longitudinal obstacle data point set, and the longitudinal gradient of the adjacent longitudinal interval is calculated; when the longitudinal gradient is larger than the maximum value of a longitudinal preset threshold, marking a raised obstacle interval on the two corresponding subinterval blocks; when the longitudinal gradient is smaller than the minimum value of a longitudinal preset threshold value, marking a sunken barrier interval on the corresponding two subinterval blocks; extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of longitudinal obstacle data points.
Further, the step of extracting the transverse gradient barrier from the transverse interval to obtain a transverse barrier data point set includes calculating the transverse gradient of adjacent transverse intervals; when the transverse gradient is larger than the maximum value of a transverse preset threshold, marking a raised obstacle interval on the two corresponding sub-interval blocks; when the transverse gradient is smaller than the minimum value of a transverse preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks; extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of lateral obstacle data points.
Further, processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set, wherein the obstacle data point set comprises removing repeated obstacle data point sets in the transverse obstacle data point set and the longitudinal obstacle data point set to obtain an initial obstacle data point set; splicing the data point sets belonging to the same obstacle in the initial obstacle data point set; obtaining a plurality of sub-obstacle data point sets; the plurality of sub-set of obstacle data points constitute the set of obstacle data points.
The invention aims to provide a multi-line laser radar obstacle identification device which comprises a preprocessing module, a longitudinal obstacle data point set acquisition module, a transverse obstacle data point set acquisition module and an obstacle data point set acquisition module, wherein the preprocessing module is used for acquiring a plurality of obstacle data points; the preprocessing module is used for preprocessing laser radar data to obtain a longitudinal interval and a transverse interval; the longitudinal obstacle data point set acquisition module is used for extracting longitudinal gradient obstacles in the longitudinal interval to obtain a longitudinal obstacle data point set; the transverse obstacle data point set acquisition module is used for carrying out transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set; the obstacle data point set acquisition module is used for processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set.
Further, the preprocessing module is further configured to divide the lidar data into a plurality of subintervals along a radius with the lidar as a center of a circle; dividing each subinterval into a plurality of subinterval blocks based on the laser beams of different lines; wherein the plurality of subinterval blocks of each subinterval constitute the longitudinal interval; sub-interval blocks containing laser beams of the same line among the plurality of sub-intervals constitute the transverse interval; a center of gravity of a set of data points within each subinterval block is calculated, by which center of gravity the position of the set of data points is represented.
Further, the longitudinal obstacle data point set acquisition module is further configured to calculate a longitudinal gradient of adjacent longitudinal intervals; when the longitudinal gradient is larger than the maximum value of a longitudinal preset threshold, marking a raised obstacle interval on the two corresponding subinterval blocks; when the longitudinal gradient is smaller than the minimum value of a longitudinal preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks; extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of longitudinal obstacle data points.
Further, the transverse obstacle data point set acquisition module is further configured to calculate a transverse gradient of adjacent transverse intervals; when the transverse gradient is larger than the maximum value of a transverse preset threshold, marking a raised obstacle interval on the two corresponding sub-interval blocks; when the transverse gradient is smaller than the minimum value of a transverse preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks; extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of lateral obstacle data points.
Further, the obstacle data point set acquisition module is further configured to remove repeated obstacle data point sets in the transverse obstacle data point set and the longitudinal obstacle data point set to obtain an initial obstacle data point set; splicing the data point sets belonging to the same obstacle in the initial obstacle data point set; obtaining a plurality of sub-obstacle data point sets; the plurality of sub-sets of obstacle data points constitute the set of obstacle data points.
The technical scheme of the embodiment of the invention at least has the following advantages and beneficial effects:
the method and the device for identifying the obstacle of the multiline laser radar can greatly reduce the processing amount and the calculation amount of radar data.
The method and the device for identifying the barrier of the multi-line laser radar can detect the barrier in a 360-degree range or a specific direction, and have the advantages of wide identification space range and high controllability. And the obstacles are extracted from two dimensions, so that the tiny obstacles can be accurately identified, the obstacles are effective to raised and depressed obstacles, and the form range of the identifiable obstacles is wide.
Drawings
FIG. 1 is an exemplary flow chart of a multiline lidar obstacle identification method according to some embodiments of the present disclosure;
FIG. 2 is an exemplary diagram of partitioning lidar data provided by some embodiments of the invention;
FIG. 3 is an exemplary diagram of the identification of obstacles in a longitudinal interval provided by some embodiments of the present invention;
fig. 4 is an exemplary diagram of an obstacle not recognized but actually present in a longitudinal section according to some embodiments of the present invention;
FIG. 5 is an exemplary diagram of a multiline lidar system configured to identify obstacles according to some embodiments of the present invention;
fig. 6 is a block diagram illustrating an exemplary multiline lidar obstacle identification apparatus according to some embodiments of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
In the present specification, the multiline lidar obstacle recognition method is exemplified by a 16-line lidar, and it is understood that the method of the present invention covers 16-line radars as well as radars above 16-line radars.
Fig. 1 is an exemplary flowchart of a multiline lidar obstacle identification method according to some embodiments of the present invention. In some embodiments, the process 100 shown in fig. 1 may be performed by the apparatus 600. As shown in fig. 1, the process 100 may include the following steps:
and step 110, preprocessing the laser radar data to obtain a longitudinal interval and a transverse interval.
Lidar data may include multi-line lidar data that may be acquired by multi-line lidar sensors that may include laser rotating range radars that emit and receive multiple lasers simultaneously. The multiline laser radar can measure and acquire 3D information of the surrounding environment. Such as 16-line, 32-line, 64-line, and 128-line lidar, etc., where each line represents a beam of laser light emitted by one laser transmitter. The laser light returns to the laser receiver after encountering the object through diffuse reflection, and the radar module calculates the distance between the transmitter and the object according to the time interval of the transmitted and received signals. Taking a 16-line radar as an example, the total number of 16 laser beams (-15 degrees) is from bottom to top by taking an upper 2 degrees and a lower 2 degrees as intervals. Wherein, can take-15 as line 1, from bottom to top respectively numbering 1 st line to 16 th line.
The laser radar in the specification can be a rotary range radar, and the measurement range is 0-360 degrees. The preprocessing may refer to segmenting the lidar data to obtain a longitudinal interval along the lidar data radius and a transverse interval along the lidar perimeter. It will be appreciated that the pre-processing may be performed for the entire 360 ° range of data, or for portions of the angular range of data.
In some embodiments, the lidar data may be divided into a plurality of subintervals along a radius centered on the lidar. Taking 360 ° as an example, the 360 ° data interval of each line of laser data is divided into 360 sub-intervals (or divided into 180 sub-intervals according to other resolutions). As shown in fig. 2, the laser radar data of 360 ° is divided into 16 subintervals. The circle shown by 202 is lidar data of a certain line of the lidar, and the lidar data of each line can be divided by a dividing line 201 to obtain subintervals shown as 203 and 204, each subinterval includes multi-line lidar data, and the lidar data may include a 3D data point set. Illustratively, the set of 3D data points may be point cloud data of a radar. Illustratively, a certain subinterval may include 3D data point sets 205 and 206.
Each subinterval is divided into a plurality of subinterval blocks based on the laser beams of different lines. As shown in fig. 2, a plurality of subinterval blocks 207 and 208 containing 3D data point sets can be obtained by dividing the 3D data point sets of the laser beams belonging to different lines for each subinterval block. In some embodiments, the subinterval block may also include only a set of 3D data points. It will be appreciated that for a 16-line lidar, there may be 16 subinterval blocks per subinterval, and a 3D data point set may be contained in each subinterval block. Wherein the plurality of subinterval blocks of each subinterval form a longitudinal interval. As shown in fig. 2, the subinterval block 207 and subinterval block 208 belonging to a certain subinterval constitute a longitudinal interval, and it will be appreciated that each subinterval may include 16 longitudinal intervals for a 16-line lidar. The sub-interval blocks between the plurality of sub-intervals containing the laser beams of the same line constitute the transverse interval. As shown in fig. 2, the subinterval blocks 207 and 209 for the lidar data in the same line form a horizontal interval. It will be appreciated that 16 transverse intervals may be included for lidar data divided into 16 intervals.
And calculating the gravity center of the data point set in each subinterval block, and representing the position of the data point set through the gravity center.
The set of data points may be a set of 3D data points having three-dimensional coordinates. The multiple 3D data point sets in each subinterval block may be processed in various ways to obtain a center of gravity of the data point set, the center of gravity represents the 3D data point set, and a coordinate of the center of gravity is used as a position of the 3D data point set in the subinterval block for subsequent analysis and processing.
Some embodiments in this specification perform preprocessing on the lidar data, and represent the gravity center of the 3D data point set to multiple point cloud data of the subinterval block, so that on the premise of not affecting the recognition effect, the data throughput is reduced. Therefore, the efficiency of data processing is improved, and real-time identification is realized.
And step 120, extracting longitudinal gradient obstacles in the longitudinal interval to obtain a longitudinal obstacle data point set.
Longitudinal gradient obstacle extraction may refer to determining whether an obstacle exists in a longitudinal section using a gradient of the longitudinal section. For example, a gradient of a corresponding subinterval block may be determined based on location information of a set of adjacent data points within the subinterval, and based on a value of the gradient, a determination may be made as to whether an obstacle is present within the adjacent subinterval block. And if so, extracting a data point set in the adjacent interval, and extracting the data points of the obstacles in the data point set to be used as a longitudinal obstacle data point set.
In some embodiments, longitudinal gradients of adjacent longitudinal intervals may be calculated. As shown in fig. 2, the sub-interval block 207 and the sub-interval block 208 can be considered adjacent vertical intervals. The longitudinal gradient may refer to a gradient formed between adjacent longitudinal intervals. In some embodiments, the weighting may be based on a set of data pointsThe heart calculates the longitudinal gradient. For example, the barycentric coordinate of the subinterval block 208 is G1, and the barycentric coordinate of the subinterval block 207 is G0. Can be calculated by gradient
Wherein Z1 and X1 are the values of the Z axis and X axis of G1, respectively, and Z0 and X0 are the values of the Z axis and X axis of G0, respectively. Taking fig. 2 as an example, the positive axis of the X axis is the positive north direction of fig. 2, the positive axis of the Y axis is the positive west direction of fig. 2, and the positive axis of the Z axis is the outer direction of the drawing. And traversing and calculating each adjacent longitudinal interval to obtain a plurality of longitudinal gradients.
And when the longitudinal gradient is greater than the maximum value of the longitudinal preset threshold, marking the corresponding two subinterval blocks with a raised obstacle interval.
The longitudinal preset threshold may refer to a preset gradient threshold of adjacent longitudinal sections when there is no obstacle. In some embodiments, the longitudinal preset threshold may be a range of values within which the two adjacent longitudinal intervals may be considered to be free of obstructions when the longitudinal gradient is within the range. For example, the longitudinal preset threshold may range between
Including the endpoints. The maximum value of the longitudinal preset threshold may refer to a maximum value within a range of the longitudinal preset threshold. For example, the maximum value of the longitudinal preset threshold may be
When the longitudinal gradient is greater than
Indicating the presence of raised obstacles within the adjacent longitudinal extent. The longitudinal section with the raised barrier can be marked with the raised barrier section for subsequent treatment. For example, in FIG. 3, when the lidar data detects a raised obstacle, the lidar data may have an abrupt position, e.g., a gap may exist in a corresponding line of the lidar data that strikes raised obstacle 301. Adjacent rails of a completely flat groundGradient in the interval of
Where a denotes the angle of inclination between the lidar and the ground, as shown in fig. 5. When the laser is applied to the raised barrier, the gradient of the laser beams of the two adjacent lines is higher.
And when the longitudinal gradient is smaller than the minimum value of the longitudinal preset threshold, marking the interval of the sunken obstacles on the corresponding two subinterval blocks.
The minimum value of the longitudinal preset threshold value may refer to a minimum value within a longitudinal preset threshold value range. For example, the minimum value of the longitudinal preset threshold may be
When the longitudinal gradient is less than
Indicating the presence of a recessed barrier within the adjacent longitudinal extent. The longitudinal section with the sunken barrier can be marked with the sunken barrier section for subsequent processing.
And extracting data point sets of the obstacles in the convex obstacle interval and the concave obstacle interval into a longitudinal obstacle data point set.
When an obstacle is identified in a subinterval block, the three-dimensional coordinates of the set of data points for the obstacle within the subinterval block may be extracted. Taking fig. 5 as an example, the Z-axis coordinate in the three-dimensional coordinates of the data points hitting the raised obstacle is greater than 0. Therefore, a data point in which the Z-axis coordinate of the three-dimensional coordinate within the subinterval block is greater than 0 may be taken as a convex obstacle data point. Similarly, a data point in which the Z-axis coordinate of the three-dimensional coordinate within the subinterval block is smaller than 0 may be regarded as a recessed obstacle data point. Similarly, the obstacle data points may also be extracted based on the X-axis and the Y-axis. In some embodiments, the obstacle data points of each two adjacent longitudinal intervals may be treated as one set of obstacle data points. The longitudinal set of obstacle data points may include a plurality of adjacent intervals of obstacle data point sets.
For some short obstacles, or when the lidar is far away from the obstacle, only a single laser beam scans the obstacle and cannot be identified using the method of longitudinal gradient obstacle extraction, as shown in fig. 4, a cylindrical obstacle 401 is measured by only a single laser beam scan. This case can be identified using a method of lateral gradient obstacle extraction. Therefore, to avoid missing detection, step 130 is also included.
And step 130, performing transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set.
The transverse gradient obstacle extraction may refer to determining whether an obstacle exists in a transverse section using a gradient of the transverse section. The method for extracting the transverse obstacle data point set is similar to the method for extracting the longitudinal obstacle data point set, and for more details of obtaining the transverse obstacle data point set, refer to step 120 and the related description thereof.
In some embodiments, the lateral gradient of adjacent lateral intervals may be calculated. As shown in fig. 2, sub-interval block 207 and sub-interval block 209 may be considered as adjacent lateral intervals. A lateral gradient may refer to a gradient formed between adjacent lateral intervals. The content of calculating the transverse gradient is similar to that of calculating the longitudinal gradient, and for more about calculating the transverse gradient, see step 120 and its associated description.
And when the transverse gradient is greater than the maximum value of the transverse preset threshold, marking the corresponding two sub-interval blocks with a mark of a raised obstacle interval.
The lateral preset threshold may refer to a preset gradient threshold of adjacent lateral sections when there is no obstacle. In some embodiments, the lateral preset threshold may be a range of values within which the two adjacent lateral intervals may be considered free of obstacles when the lateral gradient is within the range. For example, the lateral preset threshold may range between
Including the endpoints. The maximum value of the lateral preset threshold may refer to a maximum value within a range of the lateral preset threshold. For example, transverselyThe maximum value of the threshold value may be set to
When the transverse gradient is greater than
Indicating the presence of raised obstacles within the adjacent lateral extent. The transverse section with the raised barrier can be marked with the raised barrier section for subsequent processing.
And when the transverse gradient is smaller than the minimum value of the transverse preset threshold, marking the interval of the sunken barrier on the corresponding two sub-interval blocks.
The minimum value of the lateral preset threshold value may refer to a minimum value within a range of the lateral preset threshold value. For example, the minimum value of the lateral preset threshold may be
When the transverse gradient is less than
Indicating the presence of a recessed barrier within the adjacent lateral extent. The transverse section with the sunken barrier can be marked with the sunken barrier section for subsequent processing.
And extracting the obstacle data point set in the convex obstacle interval and the concave obstacle interval as a transverse obstacle data point set.
The content of the extracted transverse obstacle data point set is similar to the content of the extracted longitudinal obstacle data point set, and for more about extracting the transverse obstacle data point set, see step 120 and its related description.
The set of obstacle data points may include a plurality of sub-sets of data points, each sub-set of data points corresponding to an obstacle, each sub-set of data points including all data points for each obstacle.
In some embodiments, the repeated obstacle data point sets in the transverse obstacle data point set and the longitudinal obstacle data point set may be removed to obtain an initial obstacle data point set. For example, for some obstacles, there may be both a longitudinal obstacle data point set and a lateral obstacle data point set (i.e., an obstacle that is extracted both longitudinally and laterally), and therefore, duplicate data points in the longitudinal obstacle data point set and the lateral obstacle data point set may be deleted and the two sets merged to yield an initial obstacle data point set.
Splicing data point sets belonging to the same obstacle in the initial obstacle data point set; a plurality of sub-obstacle data point sets are obtained. For example, for some larger obstacles, the range may occupy multiple subinterval blocks, and thus, data points belonging to the same obstacle may be identified based on their three-dimensional coordinates and treated as a set of sub-obstacle data points.
The plurality of sub-set of obstacle data points constitute a set of obstacle data points. The set of obstacle data points includes data points for all obstacles detected. I.e., the set of obstacle data points includes all of the obstacles identified.
Fig. 6 is a block diagram of an exemplary multi-line lidar obstacle identification apparatus according to some embodiments of the present invention. As shown in fig. 6, the apparatus 600 may include a preprocessing module 610, a longitudinal obstacle data point set acquisition module 620, a lateral obstacle data point set acquisition module 630, and an obstacle data point set acquisition module 640.
The preprocessing module 610 is configured to preprocess the lidar data to obtain a longitudinal interval and a transverse interval. The preprocessing module 610 is further configured to divide the lidar data into a plurality of subintervals along a radius with the lidar as a center of a circle; dividing each subinterval into a plurality of subinterval blocks based on the laser beams of different lines; wherein, a plurality of subinterval blocks of each subinterval form a longitudinal interval; the sub-interval blocks containing the laser beams of the same line among the plurality of sub-intervals form a transverse interval; and calculating the gravity center of the data point set in each subinterval block, and representing the position of the data point set through the gravity center. For more of the preprocessing module 610, refer to fig. 1 and its associated description.
The longitudinal obstacle data point set obtaining module 620 is configured to perform longitudinal gradient obstacle extraction on the longitudinal interval to obtain a longitudinal obstacle data point set. The longitudinal obstacle data point set acquisition module 620 is further configured to calculate a longitudinal gradient of adjacent longitudinal intervals; when the longitudinal gradient is larger than the maximum value of a longitudinal preset threshold value, marking a raised obstacle interval on the corresponding two subinterval blocks; when the longitudinal gradient is smaller than the minimum value of a longitudinal preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks; and extracting data point sets of the obstacles in the convex obstacle interval and the concave obstacle interval into a longitudinal obstacle data point set. For more on the longitudinal obstacle data point set acquisition module 620, refer to fig. 1 and its associated description.
The transverse obstacle data point set obtaining module 630 is configured to perform transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set. The transverse obstacle data point set acquisition module 630 is further configured to calculate a transverse gradient of adjacent transverse intervals; when the transverse gradient is larger than the maximum value of a transverse preset threshold value, marking a raised barrier interval on the corresponding two subinterval blocks; when the transverse gradient is smaller than the minimum value of a transverse preset threshold value, marking a sunken barrier interval on the corresponding two subinterval blocks; the data point sets of the obstacles in the convex obstacle interval and the concave obstacle interval are extracted as transverse obstacle data point sets. For more on the lateral obstacle data point set acquisition module 630, see fig. 1 and its associated description.
The obstacle data point set acquisition module 640 is configured to process the longitudinal obstacle data point set and the lateral obstacle data point set to obtain an obstacle data point set. The obstacle data point set obtaining module 640 is further configured to remove a repeated obstacle data point set in the transverse obstacle data point set and the longitudinal obstacle data point set to obtain an initial obstacle data point set; splicing data point sets belonging to the same obstacle in the initial obstacle data point set; obtaining a plurality of sub-obstacle data point sets; the plurality of sub-set of obstacle data points constitute a set of obstacle data points. For more on the obstacle data point set acquisition module 640, refer to fig. 1 and its associated description.
The multi-line laser radar obstacle identification method provided by some embodiments in the specification is high in efficiency and small in calculation amount, and can realize real-time obstacle identification. By preprocessing the laser radar measurement data, the gravity center of the 3D point set in the subinterval block represents a plurality of data points in the subinterval block, so that the data processing amount is reduced on the premise of not influencing the identification effect. Taking 16-line lidar as an example, each laser beam is divided into 180 subintervals by 1800 3D points for one circle, and the data volume is reduced to one tenth of the original data. Therefore, the method has high efficiency, can simultaneously identify the obstacles by a plurality of laser radars and really realizes real-time identification.
Some embodiments in this specification provide a method for identifying obstacles by using a multi-line lidar, which can detect obstacles in a 360-degree range or a specific direction, and thus the spatial range of the identification is wide. And by extracting the obstacle from two dimensions, the shape and the trend of the protrusion or the depression of the obstacle can be reflected from the sign of the longitudinal or transverse gradient value, so that the form range of the recognizable object is wide. And the longitudinal gradient threshold value and the transverse gradient threshold value are adjustable, so that the size of the obstacle can be adjusted, and the obstacle with small volume can be accurately identified.
In some embodiments, the installation tilt angle of the lidar is adjustable, i.e., the 3D space for obstacle identification is adjustable, and multiple lidar combinations installed at different positions and tilt angles may cover a wider 3D space.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A multiline laser radar obstacle identification method is characterized by comprising the following steps,
preprocessing laser radar data to obtain a longitudinal interval and a transverse interval;
extracting longitudinal gradient obstacles in the longitudinal interval to obtain a longitudinal obstacle data point set;
extracting transverse gradient obstacles in the transverse interval to obtain a transverse obstacle data point set;
and processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set.
2. The multiline lidar barrier identification method of claim 1 wherein the pre-processing of lidar data includes,
dividing laser radar data into a plurality of subintervals along a radius by taking the laser radar as a circle center;
dividing each subinterval into a plurality of subinterval blocks based on the laser beams of different lines; wherein the plurality of subinterval blocks of each subinterval constitute the longitudinal interval; sub-interval blocks containing laser beams of the same line among the plurality of sub-intervals constitute the transverse interval;
a center of gravity of the set of data points within each subinterval is calculated, by which center of gravity the position of the set of data points is represented.
3. The multiline lidar barrier identification method of claim 1 wherein the longitudinal gradient barrier extraction for the longitudinal interval results in a set of longitudinal barrier data points comprising,
calculating the longitudinal gradient of the adjacent longitudinal intervals;
when the longitudinal gradient is larger than the maximum value of a longitudinal preset threshold value, marking a raised obstacle interval on the corresponding two subinterval blocks;
when the longitudinal gradient is smaller than the minimum value of a longitudinal preset threshold value, marking a sunken barrier interval on the corresponding two subinterval blocks;
extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of longitudinal obstacle data points.
4. The multiline lidar barrier identification method of claim 1 wherein the transverse gradient barrier extraction for the transverse interval results in a transverse barrier data point set comprising,
calculating the transverse gradient of the adjacent transverse interval;
when the transverse gradient is larger than the maximum value of a transverse preset threshold value, marking a raised obstacle interval on the corresponding two sub-interval blocks;
when the transverse gradient is smaller than the minimum value of a transverse preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks;
extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of lateral obstacle data points.
5. The multiline lidar obstruction identification method of claim 1 wherein processing the set of longitudinal obstruction data points and the set of transverse obstruction data points to obtain a set of obstruction data points comprises,
removing repeated obstacle data point sets in the transverse obstacle data point set and the longitudinal obstacle data point set to obtain an initial obstacle data point set;
splicing the data point sets belonging to the same obstacle in the initial obstacle data point set; obtaining a plurality of sub-obstacle data point sets;
the plurality of sub-sets of obstacle data points constitute the set of obstacle data points.
6. The multi-line laser radar obstacle identification device is characterized by comprising a preprocessing module, a longitudinal obstacle data point set acquisition module, a transverse obstacle data point set acquisition module and an obstacle data point set acquisition module;
the preprocessing module is used for preprocessing laser radar data to obtain a longitudinal interval and a transverse interval;
the longitudinal obstacle data point set acquisition module is used for performing longitudinal gradient obstacle extraction on the longitudinal interval to obtain a longitudinal obstacle data point set;
the transverse obstacle data point set acquisition module is used for carrying out transverse gradient obstacle extraction on the transverse interval to obtain a transverse obstacle data point set;
the obstacle data point set acquisition module is used for processing the longitudinal obstacle data point set and the transverse obstacle data point set to obtain an obstacle data point set.
7. Multiline lidar obstacle identification apparatus of claim 6, wherein the preprocessing module is further configured to,
dividing laser radar data into a plurality of subintervals along a radius by taking the laser radar as a circle center;
dividing each subinterval into a plurality of subinterval blocks based on the laser beams of different lines; wherein the plurality of subinterval blocks of each subinterval constitute the longitudinal interval; sub-interval blocks containing laser beams of the same line among the plurality of sub-intervals constitute the transverse interval;
a center of gravity of a set of data points within each subinterval block is calculated, by which center of gravity the position of the set of data points is represented.
8. The multiline lidar obstruction identification apparatus of claim 6 wherein the longitudinal obstruction data point set acquisition module is further configured to,
calculating the longitudinal gradient of adjacent longitudinal intervals;
when the longitudinal gradient is larger than the maximum value of a longitudinal preset threshold, marking a raised obstacle interval on the two corresponding subinterval blocks;
when the longitudinal gradient is smaller than the minimum value of a longitudinal preset threshold value, marking a sunken barrier interval on the corresponding two subinterval blocks;
extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of longitudinal obstacle data points.
9. The multiline lidar obstruction identification apparatus of claim 6 wherein the transverse obstruction data point set acquisition module is further configured to,
calculating the transverse gradient of the adjacent transverse interval;
when the transverse gradient is larger than the maximum value of a transverse preset threshold value, marking a raised obstacle interval on the corresponding two sub-interval blocks;
when the transverse gradient is smaller than the minimum value of a transverse preset threshold value, marking a concave obstacle interval on the corresponding two subinterval blocks;
extracting a set of data points for obstacles within the raised obstacle interval and the recessed obstacle interval as the set of lateral obstacle data points.
10. The multiline lidar obstruction identification apparatus of claim 6 wherein the obstruction data point set acquisition module is further configured to,
removing repeated obstacle data point sets in the transverse obstacle data point set and the longitudinal obstacle data point set to obtain an initial obstacle data point set;
splicing the data point sets belonging to the same obstacle in the initial obstacle data point set; obtaining a plurality of sub-obstacle data point sets;
the plurality of sub-sets of obstacle data points constitute the set of obstacle data points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211279561.0A CN115356747B (en) | 2022-10-19 | 2022-10-19 | Multi-line laser radar obstacle identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211279561.0A CN115356747B (en) | 2022-10-19 | 2022-10-19 | Multi-line laser radar obstacle identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115356747A true CN115356747A (en) | 2022-11-18 |
| CN115356747B CN115356747B (en) | 2023-01-24 |
Family
ID=84008431
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211279561.0A Active CN115356747B (en) | 2022-10-19 | 2022-10-19 | Multi-line laser radar obstacle identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115356747B (en) |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011242277A (en) * | 2010-05-19 | 2011-12-01 | Ihi Aerospace Co Ltd | Travel area determination device and travel area determination method of mobile robot |
| US20110309967A1 (en) * | 2009-03-03 | 2011-12-22 | Tok Son Choe | Apparatus and the method for distinguishing ground and obstacles for autonomous mobile vehicle |
| CN104597453A (en) * | 2015-01-27 | 2015-05-06 | 长春理工大学 | Detection method and device for safety driving area of vehicle corrected by inertial measuring unit |
| WO2015129842A1 (en) * | 2014-02-27 | 2015-09-03 | 株式会社次世代技術研究所 | Radar device |
| CN105866790A (en) * | 2016-04-07 | 2016-08-17 | 重庆大学 | Laser radar barrier identification method and system taking laser emission intensity into consideration |
| DE102018112151A1 (en) * | 2017-05-25 | 2018-11-29 | General Motors Llc | METHOD AND DEVICE FOR CLASSIFYING LIDARDATES FOR OBJECT DETECTION |
| EP3517997A1 (en) * | 2018-01-30 | 2019-07-31 | Wipro Limited | Method and system for detecting obstacles by autonomous vehicles in real-time |
| CN110674705A (en) * | 2019-09-05 | 2020-01-10 | 北京智行者科技有限公司 | Small-sized obstacle detection method and device based on multi-line laser radar |
| CN111033314A (en) * | 2017-04-10 | 2020-04-17 | Bea股份公司 | Human body recognition method and human body recognition sensor |
| CN111239757A (en) * | 2020-03-12 | 2020-06-05 | 湖南大学 | Automatic extraction method and system for road surface characteristic parameters |
| US20200219264A1 (en) * | 2019-01-08 | 2020-07-09 | Qualcomm Incorporated | Using light detection and ranging (lidar) to train camera and imaging radar deep learning networks |
| WO2020187103A1 (en) * | 2019-03-19 | 2020-09-24 | 深圳市镭神智能系统有限公司 | Prism and multi-beam lidar system |
| US20210110176A1 (en) * | 2019-10-14 | 2021-04-15 | Denso Corporation | Obstacle identification apparatus and obstacle identification program |
| CN114280626A (en) * | 2021-12-17 | 2022-04-05 | 成都朴为科技有限公司 | Laser radar SLAM method and system based on local structure information expansion |
| CN114488194A (en) * | 2022-01-21 | 2022-05-13 | 常州大学 | Method for detecting and identifying targets under structured road of intelligent driving vehicle |
| CN114842166A (en) * | 2022-03-03 | 2022-08-02 | 长沙行深智能科技有限公司 | Negative obstacle detection method, system, medium, and apparatus applied to structured road |
| CN114910881A (en) * | 2021-02-08 | 2022-08-16 | 重庆兰德适普信息科技有限公司 | Negative obstacle detection method and device and vehicle |
-
2022
- 2022-10-19 CN CN202211279561.0A patent/CN115356747B/en active Active
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110309967A1 (en) * | 2009-03-03 | 2011-12-22 | Tok Son Choe | Apparatus and the method for distinguishing ground and obstacles for autonomous mobile vehicle |
| JP2011242277A (en) * | 2010-05-19 | 2011-12-01 | Ihi Aerospace Co Ltd | Travel area determination device and travel area determination method of mobile robot |
| WO2015129842A1 (en) * | 2014-02-27 | 2015-09-03 | 株式会社次世代技術研究所 | Radar device |
| CN104597453A (en) * | 2015-01-27 | 2015-05-06 | 长春理工大学 | Detection method and device for safety driving area of vehicle corrected by inertial measuring unit |
| CN105866790A (en) * | 2016-04-07 | 2016-08-17 | 重庆大学 | Laser radar barrier identification method and system taking laser emission intensity into consideration |
| CN111033314A (en) * | 2017-04-10 | 2020-04-17 | Bea股份公司 | Human body recognition method and human body recognition sensor |
| DE102018112151A1 (en) * | 2017-05-25 | 2018-11-29 | General Motors Llc | METHOD AND DEVICE FOR CLASSIFYING LIDARDATES FOR OBJECT DETECTION |
| EP3517997A1 (en) * | 2018-01-30 | 2019-07-31 | Wipro Limited | Method and system for detecting obstacles by autonomous vehicles in real-time |
| US20200219264A1 (en) * | 2019-01-08 | 2020-07-09 | Qualcomm Incorporated | Using light detection and ranging (lidar) to train camera and imaging radar deep learning networks |
| WO2020187103A1 (en) * | 2019-03-19 | 2020-09-24 | 深圳市镭神智能系统有限公司 | Prism and multi-beam lidar system |
| CN110674705A (en) * | 2019-09-05 | 2020-01-10 | 北京智行者科技有限公司 | Small-sized obstacle detection method and device based on multi-line laser radar |
| US20210110176A1 (en) * | 2019-10-14 | 2021-04-15 | Denso Corporation | Obstacle identification apparatus and obstacle identification program |
| CN111239757A (en) * | 2020-03-12 | 2020-06-05 | 湖南大学 | Automatic extraction method and system for road surface characteristic parameters |
| CN114910881A (en) * | 2021-02-08 | 2022-08-16 | 重庆兰德适普信息科技有限公司 | Negative obstacle detection method and device and vehicle |
| CN114280626A (en) * | 2021-12-17 | 2022-04-05 | 成都朴为科技有限公司 | Laser radar SLAM method and system based on local structure information expansion |
| CN114488194A (en) * | 2022-01-21 | 2022-05-13 | 常州大学 | Method for detecting and identifying targets under structured road of intelligent driving vehicle |
| CN114842166A (en) * | 2022-03-03 | 2022-08-02 | 长沙行深智能科技有限公司 | Negative obstacle detection method, system, medium, and apparatus applied to structured road |
Non-Patent Citations (3)
| Title |
|---|
| O"CONNELL, MJ等: "Measurement and Monitoring of Barrier Island Forest Sensitivity to Ecohydrological Change Using LIDAR Remote Sensing", 《JOURNAL OF COASTAL RESEARCH》 * |
| 张穗华等: "基于三维激光雷达的障碍物检测方法研究", 《机电产品开发与创新》 * |
| 彭梦: "基于多传感器融合的移动机器人障碍物检测研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115356747B (en) | 2023-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109752701B (en) | Road edge detection method based on laser point cloud | |
| CN110163871B (en) | Ground segmentation method and device for multi-line laser radar | |
| JP5822255B2 (en) | Object identification device and program | |
| CN107274417A (en) | A kind of single wooden dividing method based on airborne laser point cloud aggregation | |
| CN108180912B (en) | Mobile robot positioning system and method based on hybrid navigation band | |
| KR101495333B1 (en) | Apparatus and method for detecting obstacles | |
| CN110349260B (en) | Automatic pavement marking extraction method and device | |
| CN110794424B (en) | Full-waveform airborne laser radar ground feature classification method and system based on feature selection | |
| JP6381137B2 (en) | Label detection apparatus, method, and program | |
| CN112698301B (en) | Laser radar target identification method, device and equipment based on distance strength association | |
| CN116148808A (en) | Automatic driving laser repositioning method and system based on point cloud descriptor | |
| CN101408937A (en) | Method and apparatus for locating character row | |
| CN112800938B (en) | Method and device for detecting occurrence of side rockfall of unmanned vehicle | |
| CN115598637B (en) | Tunnel surrounding rock deformation monitoring method and system | |
| CN118502453B (en) | Unmanned ship autonomous berthing method and system | |
| CN115077414A (en) | Device and method for measuring bottom contour of sea surface target by underwater vehicle | |
| CN113570005A (en) | Long-distance ship type identification method based on airborne photon radar | |
| CN111736167B (en) | Method and device for obtaining laser point cloud density | |
| CN115356747B (en) | Multi-line laser radar obstacle identification method and device | |
| CN115267827B (en) | Laser radar harbor area obstacle sensing method based on high density screening | |
| CN114862761B (en) | Power transformer liquid level detection method, device, equipment and storage medium | |
| CN115908541A (en) | Curvature information-based vehicle-mounted laser radar point cloud clustering algorithm for indoor environment | |
| CN115167435A (en) | Forward-looking line laser obstacle identification method and device for mobile robot | |
| CN112989946B (en) | Lane line determining method, device, equipment and vehicle | |
| KR102568111B1 (en) | Apparatus and method for detecting road edge |
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 |