CN116990832A - Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium - Google Patents
Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium Download PDFInfo
- Publication number
- CN116990832A CN116990832A CN202310977709.6A CN202310977709A CN116990832A CN 116990832 A CN116990832 A CN 116990832A CN 202310977709 A CN202310977709 A CN 202310977709A CN 116990832 A CN116990832 A CN 116990832A
- Authority
- CN
- China
- Prior art keywords
- grid
- point cloud
- road edge
- height
- dangerous road
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 230000008447 perception Effects 0.000 title claims abstract description 23
- 238000001514 detection method Methods 0.000 claims abstract description 30
- 238000001914 filtration Methods 0.000 claims abstract description 9
- 230000008569 process Effects 0.000 claims description 22
- 238000003708 edge detection Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000012163 sequencing technique Methods 0.000 claims description 5
- 230000000295 complement effect Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 4
- 101100452236 Caenorhabditis elegans inf-1 gene Proteins 0.000 description 3
- 101100422770 Caenorhabditis elegans sup-1 gene Proteins 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 101150002826 inf2 gene Proteins 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000000280 densification Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
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
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
-
- 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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
-
- 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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/495—Counter-measures or counter-counter-measures using electronic or electro-optical means
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Electromagnetism (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The invention discloses a dangerous road edge perception method, a dangerous road edge perception device, terminal equipment and a storage medium, wherein the dangerous road edge perception method comprises the following steps: carrying out rasterization treatment on the two side areas or any side area of the robot task route to obtain a rasterization space on the corresponding side; acquiring a laser radar point cloud of a current frame of a sensor on a robot, and projecting the laser radar point cloud to the rasterization space according to a scanning line sequence of a laser radar to obtain a rasterization point cloud at a corresponding side; in the rasterized point cloud, each grid is status-tagged based on neighboring grid height gradient constraints and grid distance constraints; and window filtering is carried out on the marking result to obtain a dangerous grid detection result, namely a dangerous road edge perception result of the current frame. The invention is applied to the unmanned field, can effectively realize dangerous road edges on the basis of not changing the existing structure of the robot, and has higher detection precision under the condition that interference factors such as jolt exist on the ground.
Description
Technical Field
The invention relates to the technical field of unmanned aerial vehicles, in particular to a dangerous road edge perception method, a dangerous road edge perception device, terminal equipment and a storage medium.
Background
In the process of executing tasks, the robot needs to sense the physical world of the surrounding environment at all times to perform navigation walking or collision avoidance and other behaviors, so that safe running is realized. Among them, the method and research related to positive obstacle detection are perfected, while the research and application of negative obstacle, which is an undetectable object in the laser radar, are less. The damage consequences of the negative obstacle to the robot are generally more serious than those of the positive obstacle, and serious accidents such as overall rollover or falling down of the ditch are often caused. In order to make up for the technical defect, a common method is to specially add radar equipment or vision equipment for negative obstacle perception, and realize stable detection of the negative obstacle through a redundant sensor group, so that not only can the hardware cost be too high, but also the responsibility of a robot system can be increased. Therefore, it is a high technical challenge for commercial robots how to achieve a negative obstacle detection effect that is inexpensive and stable.
In the process of movement, the robot needs to actively avoid pedestrians and social vehicle targets, so that in the case that an obstacle deviates to the middle position of a road, the robot usually chooses to move along the edge of the road, and the edge of the road is often a cliff, and the main core of avoiding the risk that the whole robot falls off the cliff is to effectively identify a cliff area or a dangerous road edge. How the commercialized robot utilizes the vehicle-mounted laser radar equipment to effectively detect the dangerous road edge is an effective guarantee for realizing normal operation. The scheme for realizing effective detection of dangerous road edges by using the vehicle-mounted laser radar equipment disclosed at present comprises the following steps:
the invention patent with publication number of CN114089376A and the name of negative obstacle detection method based on single laser radar discloses a scheme for detecting negative obstacle through curvature after overlapping multi-frame laser radar point clouds. According to the scheme, a densification effect is achieved by superposing radar point clouds of a single laser through multiple frames, the laser point clouds participating in ground fitting and the point clouds participating in negative obstacle calculation are distinguished after curvature calculation, ground normal vectors are extracted by the ground fitting result, and then a negative obstacle grid is obtained by using two steps of height and distance jump, so that the area and the position where a negative obstacle is located are located. The key point of the scheme is that curvature is used for distinguishing diagonal points and plane points, the calculation of curvature requires enough point cloud density, and in order to obtain enough point cloud density, the scheme uses IMU, positioning and other devices for aligning data in time and space. However, there is a probability of drift in IMU and positioning; in addition, the curvature-based scheme is only suitable for leveling the ground, and when the ground is not ideal and has interference factors such as jolts and the like, the scheme has high possibility of failure.
The invention discloses a negative obstacle detection scheme for directly counting the rasterized state, which is disclosed in the patent with publication number CN111429520A, and is named as a negative obstacle detection method, a device, a terminal device and a storage medium. According to the scheme, after laser radar point clouds are projected to a plane area where a robot is located, the number of the point clouds with different heights of each basic grid is counted, and then whether the laser radar point clouds are negative obstacle grids is directly judged according to the sequence change of the plane grids and the negative grids. This solution is simple to design, but its practicality is not good enough due to very limited consideration of external influencing factors. For example, when there is a hole area due to occlusion of a positive obstacle, the scheme is likely to detect the slice of hole area as a negative obstacle area, resulting in more detection false alarms.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a dangerous road edge perception method, a dangerous road edge perception device, terminal equipment and a storage medium, which can effectively solve the problem that the traditional detection is only suitable for leveling the ground or detecting false alarms.
In order to achieve the above object, the present invention provides a dangerous road edge sensing method, comprising the following steps:
step 1, rasterizing the two side areas or any one side area of a robot task route to obtain a rasterizing space at the corresponding side;
step 2, acquiring a laser radar point cloud of a current frame of a sensor on a robot, and projecting the laser radar point cloud to the rasterization space according to a scanning line sequence of a laser radar to obtain a rasterization point cloud at a corresponding side;
step 3, in the rasterization point cloud, carrying out state marking on each grid based on the adjacent grid height gradient constraint and the grid distance constraint;
and step 4, window filtering is carried out on the marking result in the step 3, so that a dangerous grid detection result is obtained, and a dangerous road edge perception result of the current frame is obtained.
In one embodiment, in step 3, status marking each grid based on neighboring grid height gradient constraints and grid distance constraints includes:
the method comprises the steps of taking the height of the highest laser radar point in each grid in a gridding point cloud as the grid height of a corresponding grid, marking the grid with the grid height being larger than the first height as positive occupation, marking the grid with the grid height being smaller than the second height as negative occupation, and marking the grid with the grid height being smaller than or equal to the first height and larger than or equal to the second height as unknown occupation;
and on the part of the gridding point cloud of each scanning line of the laser radar, updating the marking state of the grids marked as unknown occupation based on the adjacent grid height gradient constraint and the grid distance constraint, and outputting the marking result of each grid in the gridding point cloud.
In one embodiment, the updating the marking status of the grid marked as unknown occupancy based on the adjacent grid height gradient constraint and the grid distance constraint on the part of the gridding point cloud to which each scanning line of the laser radar belongs includes:
step 301, extracting all grids covered by the current scanning line from the rasterization point cloud, and sequencing the extracted grids in a clockwise direction or a counterclockwise direction according to an angle direction to obtain a grid sequence, wherein i=1;
step 302, acquiring the ith grid marked as unknown occupation in the grid sequence, and defining the ith grid as a grid i;
step 303, acquiring the (i+1) th grid marked as unknown occupation in the grid sequence, and defining the grid as the (i+1) th grid;
step 304, obtaining the height difference between the grids i and i+1Distance d, and judge->And d satisfies both the neighboring grid height gradient constraint and the grid distance constraint:
if yes, go to step 305;
otherwise, go to step 306;
step 305, updating the state of the grid with the lower height in the grid i and the grid i+1 from unknown occupancy to negative occupancy, and then making i=i+2, and performing step 307;
otherwise, let i=i+1 followed by step 307;
step 307, determining whether i > M is true, where M is the number of grids marked as unknown occupancy in the grid sequence:
if yes, go to step 308;
otherwise, returning to step 302;
step 308, performing steps 301 to 307 on the part of the gridding point cloud to which each scanning line of the laser radar belongs, so as to obtain a marking result of each grid in the gridding point cloud.
In one embodiment, step 304, whenWhen, i.e. the neighboring grid height gradient constraint is satisfied, wherein +.>Is the ground level threshold for grid i.
In one embodiment, the ground level thresholdThe calculation process of (1) is as follows:
wherein d i For the distance of grid i from the center of the robot, L is the distance threshold,is the ground height threshold of the grid within L meters from the center of the robot.
In one embodiment, in step 304, when d inf ≤d≤d sup When, i.e. satisfying the grid distance constraint, where d inf D is the shortest effective distance sup Is the most effective distance.
In one embodiment, the dangerous road edge sensing method further comprises:
and 6, performing space-time complementation on the dangerous road edge sensing result of the current frame based on the dangerous road edge sensing result of the previous N frames to obtain a final dangerous road edge detection result of the current frame, and outputting the final dangerous road edge detection result.
In order to achieve the above object, the present invention further provides a dangerous road edge sensing device, which adopts the above method, and includes:
the grid processing unit is used for carrying out rasterization processing on the two side areas or any one side area of the robot task route to obtain a rasterization space on the corresponding side;
the point cloud projection unit is used for acquiring the laser radar point cloud of the current frame of the sensor on the robot, and projecting the laser radar point cloud to the rasterization space according to the scanning line sequence of the laser radar to obtain the rasterization point cloud at the corresponding side;
a status marking unit for status marking each grid based on the neighboring grid height gradient constraint and the grid distance constraint in the rasterized point cloud;
and the dangerous road edge detection unit is used for carrying out window filtering on the marking result of the state marking unit to obtain a dangerous grid detection result, namely obtaining a dangerous road edge sensing result of the current frame.
In order to achieve the above object, the present invention further provides a terminal device, where the terminal device is provided with:
a memory for storing a program;
and a processor for executing the program stored in the memory, the processor being configured to perform some or all of the steps of the dangerous road edge perception method as described above when the program is executed.
To achieve the above object, the present invention also provides a computer-readable storage medium having stored therein computer-executable instructions; the computer-executable instructions, when executed by the processor, are for performing part or all of the steps of a dangerous road edge awareness method as described above.
Compared with the prior art, the embodiment of the invention has the following beneficial technical effects:
1. according to the embodiment of the invention, the two side areas or any side area of the robot task route are subjected to rasterization, and the dangerous road edge detection is performed by taking the scanning line of the laser radar as a detection unit, so that the dangerous road edge detection can be effectively realized on the basis of not changing the existing structure of the robot;
2. the embodiment of the invention considers the grid height gradient and the grid distance simultaneously in the dangerous road edge detection process, so that the method is suitable for detecting the dangerous road edge of a flat ground, and has higher detection precision even if interference factors such as jolt exist on the ground;
3. in the embodiment of the invention, the dangerous road edge sensing result of the current frame is subjected to space-time complementation by the dangerous road edge sensing result of the history frame in the preferred scheme, so that the detection result of the current frame can comprehensively reflect the dangerous road edge condition of the real-time environment of the robot, and the problem that the traditional detection only has real-time radar scanning points or has more detection false alarms is effectively solved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to the structures shown in these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a dangerous road edge sensing method according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating a real-time detection result of a current frame according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of the final dangerous road edge detection result after the current frame is time-space completed in the embodiment of the present invention;
FIG. 4 is a block diagram of a dangerous road edge sensor according to an embodiment of the present invention;
fig. 5 is a block diagram of a terminal device according to an embodiment of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that all directional indicators (such as up, down, left, right, front, and rear … …) in the embodiments of the present invention are merely used to explain the relative positional relationship, movement, etc. between the components in a particular posture (as shown in the drawings), and if the particular posture is changed, the directional indicator is changed accordingly.
Furthermore, descriptions such as those referred to as "first," "second," and the like, are provided for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implying an order of magnitude of the indicated technical features in the present disclosure. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the present invention, unless specifically stated and limited otherwise, the terms "connected," "affixed," and the like are to be construed broadly, and for example, "affixed" may be a fixed connection, a removable connection, or an integral body; the device can be mechanically connected, electrically connected, physically connected or wirelessly connected; either directly or indirectly, through intermediaries, or both, may be in communication with each other or in interaction with each other, unless expressly defined otherwise. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In addition, the technical solutions of the embodiments of the present invention may be combined with each other, but it is necessary to be based on the fact that those skilled in the art can implement the technical solutions, and when the technical solutions are contradictory or cannot be implemented, the combination of the technical solutions should be considered as not existing, and not falling within the scope of protection claimed by the present invention.
Example 1
The embodiment discloses a dangerous road edge perception method, which is suitable for leveling dangerous road edges on the ground by carrying out rasterization processing on left side areas and/or right side areas of a robot task route, carrying out dangerous road edge detection by taking scanning lines of a laser radar as detection units, and simultaneously considering grid height gradients and grid distances in the detection process. Meanwhile, the device has higher detection precision when interference factors such as jolt exist on the ground. Wherein the robot may be an unmanned device, a self-service robot, or the like.
Referring to fig. 1, the dangerous road edge sensing method in the present embodiment specifically includes the following steps:
step 1, according to the current position coordinates and the traveling direction of a robot, rasterizing two side areas or any side area (namely a left side area and/or a right side area) of a robot task route to obtain a rasterizing space at a corresponding side, namely a left side rasterizing space side and/or a right side rasterizing space, wherein the left side area of the robot task route mainly refers to a left side and a left front area of the robot, and the right side area refers to a right side and a right front area;
step 2, acquiring laser radar point clouds of a current frame of a sensor on a robot, and projecting the laser radar point clouds to a rasterization space according to a scanning line sequence of the laser radar to obtain rasterization point clouds on the corresponding side, namely left rasterization point clouds and/or right rasterization point clouds, wherein the left rasterization point clouds are point cloud data corresponding to a left area, and the right rasterization point clouds are point cloud data corresponding to a right area;
step 3, in each group of rasterizing point clouds, each grid is subjected to state marking based on the adjacent grid height gradient constraint and the grid distance constraint, namely, one of the left side rasterizing point cloud and the right side rasterizing point cloud can be selected to carry out state marking on the grids according to the actual road condition of the robot, and the grids in the left side rasterizing point cloud and the right side rasterizing point cloud can be simultaneously subjected to state marking;
and 4, performing window filtering with the scanning line of the laser radar as a window on the marking result in the step 3 to obtain a dangerous grid detection result, namely obtaining a dangerous road edge sensing result of the current frame, and likewise, according to the actual road condition of the robot, not only sensing the dangerous road edge on the left side or the right side of the robot, but also sensing the dangerous road edges on the left side and the right side of the robot.
In step 1, before rasterizing the left area and/or the right area of the robot task route, a coordinate system of an area near the robot is further constructed, then the left area and the right area of the robot task route are divided according to the constructed coordinate system, and finally the left area and/or the right area of the robot task route are rasterized. The specific implementation process is as follows:
firstly, acquiring task route points of a robot, and simulating a forward main direction according to the position distribution of the task route points, wherein the coordinate center of the simulated main direction takes the robot as a center, and the main direction is a forward straight line, namely the advancing direction of the robot; in this embodiment, in order to calculate the main direction of the simulation, the straight line direction may be extracted by directly fitting the task route points by using RANSAC, or may be obtained by processing the task route points by using a least squares fitting method;
secondly, according to the two information of the center point and the advancing direction of the current robot point, a regional coordinate system can be constructed by taking the normal direction of the advancing direction;
and dividing the road area where the robot is positioned into a left sub area and a right sub area according to the current coordinate system, namely selecting one or respectively carrying out rasterization processing on the left area and the right area of the robot task route to obtain rasterization configuration and representation of the area around the robot, namely obtaining the left rasterization point cloud and/or the right rasterization point cloud. The left area and the right area are selected to be rasterized or respectively rasterized according to the actual road condition of the robot, for example, when one side of the road on which the robot is currently traveling is a wall surface or a guardrail, the rasterization can be performed only on the other side of the road.
In step 3, the process of status marking each grid based on the neighboring grid height gradient constraints and the grid distance constraints includes:
the method comprises the steps of taking the height of the highest laser radar point in each grid in a gridding point cloud as the grid height of a corresponding grid, marking the grid with the grid height being larger than the first height as positive occupation, marking the grid with the grid height being smaller than the second height as negative occupation, and marking the grid with the grid height being smaller than or equal to the first height and larger than or equal to the second height as unknown occupation;
and on the part of the gridding point cloud of each scanning line of the laser radar, updating the marking state of the grids marked as unknown occupation based on the adjacent grid height gradient constraint and the grid distance constraint, and outputting the marking result of each grid in the gridding point cloud.
For example, if the status of the grid in the left-side rasterization point cloud needs to be marked, the process includes:
firstly, defining a grid state in a left side grid point cloud, wherein the grid of the left side grid point cloud in the embodiment has four states, namely a positive occupied state, a negative occupied state, an unknown occupied state and an idle state, specifically:
the positive occupied state indicates that positive barriers such as guardrails, signal lamps and the like exist in the area corresponding to the grille;
the negative occupied state indicates that the area corresponding to the grille has negative barriers such as cliffs, hollows and the like;
the unknown occupancy state is the obstacle information position of the area corresponding to the grid, and the state of each grid is marked based on the adjacent grid height gradient constraint and the grid distance constraint, namely the grid with the unknown occupancy state is mainly aimed at;
and the idle state indicates that no laser radar scanning point exists in the area corresponding to the grating.
Secondly, pre-updating state information of each grid in the left-side rasterization point cloud based on the grid height, and specifically: taking the height of the highest laser radar point in each grid in the left side rasterization point cloud as the grid height of the corresponding grid, marking the grid with the grid height being larger than the first height as positive occupation, marking the grid with the grid height being smaller than the second height as negative occupation, marking the grids with the grid height being smaller than or equal to the first height and larger than or equal to the second height as unknown occupation, and marking the rest grids as idle;
finally, on the part of left side rasterization point cloud to which each scanning line of the laser radar belongs, after updating the marking state of the grids marked as unknown occupation based on the adjacent grid height gradient constraint and the grid distance constraint, outputting the marking result of each grid in the left side rasterization point cloud, wherein the marking process comprises the following steps: traversing grids with unknown occupation states in each scanning line, updating the state of the grids with lower heights from unknown occupation to negative occupation states if the height difference between the grids of the current grid and the next grid meets the height gradient constraint of the adjacent grids and the distance between the two grids meets the grid distance constraint, and continuing the next wheel-like updating iteration until all the grids in the left area are traversed.
Similarly, if the status of the grid in the right-side rasterization point cloud needs to be marked, the process includes:
firstly, defining grid states in a right side grid point cloud, wherein the grid states are the same as a left side grid point cloud, the grid of the right side grid point cloud has four states which are a positive occupied state, a negative occupied state, an unknown occupied state and an idle state respectively, and the meanings of the grid states are the same as those of the grid states in the left side grid point cloud, so that the embodiment does not describe the grid states in detail;
secondly, pre-updating state information of each grid in the right-side rasterization point cloud based on the grid height, and specifically: the method comprises the steps of taking the height of the highest laser radar point in each grid in a right side rasterization point cloud as the grid height of a corresponding grid, marking the grid with the grid height being larger than the first height as positive occupation, marking the grid with the grid height being smaller than the second height as negative occupation, marking the grids with the grid height being smaller than or equal to the first height and larger than or equal to the second height as unknown occupation, and marking the rest grids as idle;
finally, on the part of the right side rasterization point cloud to which each scanning line of the laser radar belongs, after updating the marking state of the grids marked as unknown occupation based on the adjacent grid height gradient constraint and the grid distance constraint, outputting the marking result of each grid in the right side rasterization point cloud, wherein the marking process comprises the following steps: traversing grids with unknown occupation states in each scanning line, updating the state of the grids with lower heights from unknown occupation to negative occupation states if the height difference between the grids of the current grid and the next grid meets the height gradient constraint of the adjacent grids and the distance between the two grids meets the grid distance constraint, and continuing the next wheel-like updating iteration until all the grids in the right area are traversed.
In this embodiment, on a part of the gridding point cloud to which each scan line of the laser radar belongs, the marking status of the grid marked as unknown occupancy is updated based on the adjacent grid height gradient constraint and the grid distance constraint, and the process specifically includes the following steps:
step 301, extracting all grids covered by a current scanning line in a rasterization point cloud, selecting any grid which is intersected with the current scanning line by a robot task route as a starting point, adopting a Breshenham algorithm, and sequencing the extracted grids in a clockwise direction or a counterclockwise direction according to an angle direction to obtain a grid sequence, and enabling an iteration parameter i=1;
step 302, acquiring the i-th grid marked as unknown occupied in the grid sequence and defining the i-th grid as the i-th grid, and notably, if the i-th grid does not exist in the grid sequence corresponding to the current scanning line, performing steps 301 to 307 on a part of the gridding point cloud to which the next scanning line belongs;
step 303, acquiring the (i+1) th grid marked as unknown occupation in the grid sequence, defining the grid as the (i+1) grid, and notably, if the (i+1) grid does not exist in the grid sequence corresponding to the current scanning line, performing steps 301 to 307 on a part of the gridding point cloud to which the next scanning line belongs;
step 304, obtaining the height difference between the grids i and i+1Distance d from Europe, and judge +.>And d satisfies both the neighboring grid height gradient constraint and the grid distance constraint:
if yes, go to step 305;
otherwise, go to step 306;
step 305, updating the state of the grid with the lower height in the grid i and the grid i+1 from unknown occupancy to negative occupancy, and then making i=i+2, and performing step 307;
otherwise, let i=i+1 followed by step 307;
step 307, determining whether i > M is true, where M is the number of grids marked as unknown occupancy in the grid sequence:
if yes, go to step 308;
otherwise, returning to step 302;
step 308, performing steps 301 to 307 on the part of the gridding point cloud to which each scanning line of the laser radar belongs, so as to obtain a marking result of each grid in the gridding point cloud.
In step 304, whenWhen, the neighboring grid height gradient constraint is satisfied. When d inf ≤d≤d sup When, the grid distance constraint is satisfied. Wherein (1)>Ground height threshold, d, for grid i inf D is the shortest effective distance sup Is the most effective distance.
In the specific implementation process, the height of the position of the robot is 0, and the ground height threshold value of the grid is defined to be a fixed value within L meters (for example, 2 meters) from the center of the robotThen the floor height threshold of grid i +.>The method comprises the following steps:
wherein d i Is the distance of grid i from the center of the robot.
It is noted that the Euclidean distance d between the grid i and the grid i+1 in the step 304 refers to the distance between the center of the grid i and the center of the grid i+1. In step 307, the number M of grids marked as unknown occupancy in the grid sequences corresponding to different lidar scan lines may be the same or different.
For example, if the grid marked as unknown occupancy in the left-side rasterization point cloud needs to be updated in a marking state, the process specifically includes the following steps:
step 311, selecting any grid where the robot task route intersects with the current scanning line as a starting point in the left rasterization point cloud, extracting all grids which can be covered by the current scanning line in the left rasterization point cloud by adopting a Breshenham algorithm, sequencing the extracted grids in a counterclockwise direction according to an angle direction to obtain a left grid sequence, and enabling an iteration parameter k=1;
step 312, the k-th grid marked as unknown occupation in the left grid sequence is acquired and defined as grid k, and it is noted that if grid k does not exist in the left grid sequence corresponding to the current scanning line, steps 311 to 317 are performed on the part of left grid point cloud to which the next scanning line belongs;
step 313, obtaining the grid marked as unknown occupation by the (k+1) th grid in the left grid sequence and defining the grid as the grid k+1, and notably, if the grid k+1 does not exist in the left grid sequence corresponding to the current scanning line, performing steps 311 to 317 on a part of left grid point cloud to which the next scanning line belongs;
step 314, obtaining a floor height threshold for grid kPreset left shortest effective distance d inf1 Distance d furthest from left sup1 Simultaneous acquisition of the height difference between grid k and grid k+1>Distance from Europe d 1 And judge And d inf1 ≤d 1 ≤d sup1 Whether or not it is:
if so, go to step 315;
otherwise, go to step 316;
step 315, updating the state of the grid with the lower height of the grid k and the grid k+1 from unknown occupancy to negative occupancy, and then letting k=k+2, and performing step 317;
otherwise, directly let k=k+1 followed by step 317;
step 317, determine k>M 1 Whether or not it is true, wherein M 1 For the number of grids in the left grid sequence marked as unknown occupancy:
if yes, go to step 318;
otherwise, return to step 312;
step 318, performing steps 311 to 317 on the left side rasterization point cloud of the part to which each scanning line of the laser radar belongs, so as to obtain a marking result of each grid in the left side rasterization point cloud.
Similarly, if the grid marked as unknown occupation in the right side rasterization point cloud needs to be updated in a marking state, the process specifically comprises the following steps:
step 321, selecting any grid where the robot task route meets the current scanning line from the right-side rasterization point cloud as a starting point, extracting all grids which can be covered by the current scanning line in the right-side rasterization point cloud by adopting a Breshenham algorithm, sequencing the extracted grids in a clockwise direction according to an angle direction to obtain a right-side grid sequence, and enabling an iteration parameter j=1;
step 322, the j-th grid marked as unknown occupation in the right grid sequence is acquired and defined as a grid j, and it is noted that if the grid j does not exist in the right grid sequence corresponding to the current scanning line, steps 321 to 327 are performed on a part of right grid point cloud to which the next scanning line belongs;
step 323, the j+1th grid marked as unknown occupation in the right grid sequence is acquired and defined as the grid j+1, and it is noted that if the grid j+1 does not exist in the right grid sequence corresponding to the current scanning line, step 321 to step 327 are performed on a part of the right grid point cloud to which the next scanning line belongs;
step 324, obtain ground level threshold for grid jPreset right shortest effective distance d inf2 Distance d furthest from right sup2 Simultaneous acquisition of the height difference between grid j and grid j+1>Distance from Europe d 2 And judge And d inf2 ≤d 2 ≤d sup2 Whether or not it is:
if so, go to step 325;
otherwise, go to step 326;
step 325, after updating the state of the lower grid of the grids j and j+1 from unknown occupancy to negative occupancy, let j=j+2 and then go to step 327;
otherwise, directly let j=j+1 and then go to step 327;
step 327, judge j>M 2 Whether or not it is true, wherein M 2 For the number of grids in the right grid sequence marked as unknown occupancy:
if yes, go to step 328;
otherwise, return to step 322;
step 328, performing steps 321 to 327 on the part of the right-side rasterization point cloud to which each scanning line of the laser radar belongs, so as to obtain a marking result of each grid in the right-side rasterization point cloud.
Notably, a preset d inf2 Can be combined with d inf1 The same or different; preset d sup2 Can be combined with d sup1 The same or different.
After the marking result of each grid in the left side rasterization point cloud and/or the right side rasterization point cloud is obtained, window filtering with the scanning line as a window is executed on all grids marked as negative occupation states, namely a scanning window with a certain size is set, and then all grids are scanned by the scanning window. Because negative obstacles in the real scenes such as the structured environment generally exist continuously in the same direction, when the scanning window has only one grid with a negative occupation state in the window in the moving process, the grid with the negative occupation state is judged to be a noise point, and the state of the grid is updated to be an unknown occupation state again. After the window filtering is finished, outputting all grids marked as negative occupied states, namely, the dangerous road edge sensing result of the current frame, for example, as shown in fig. 2.
As a preferred implementation manner, the dangerous road edge sensing method in this embodiment may optionally perform space-time complement on the detected negative obstacle. The space-time complement aims to output continuous negative obstacle results with consistent space, avoid incomplete results caused by incomplete measurement of a single frame, perform space-time complement on dangerous road edge perception results of a current frame based on dangerous road edge perception results of a previous N frames, obtain final dangerous road edge detection results of the current frame, and output the final dangerous road edge detection results. In this embodiment, an occupation raster pattern scheme is adopted to construct a complete dangerous road edge, and an occupation update rule is as follows:
S + ＝S - +log free
S + ＝S - +log occup
wherein S is - And S is + The state changes before and after updating, respectively, log free represents the state where the current measurement of the grid is not negative occupied, and log occup represents the state where the current grid is negative occupied. In the specific implementation process, the idea of occupying the raster image can be adopted to carry out space-time complement on the latest 5-frame detection result, so as to obtain a more complete detection result. For example, as shown in FIG. 3, the hazard is achieved by occupying a gridThe road edge result is completed, the detection result can comprehensively reflect the dangerous road edge condition of the real-time environment of the robot, and the problem that the traditional detection only has real-time radar scanning points or has more detection false alarms is effectively solved.
Example 2
Based on the dangerous road edge sensing method in embodiment 1, this embodiment discloses a dangerous road edge sensing device. Referring to fig. 4, the dangerous road edge sensing device includes a grid processing unit, a point cloud projection unit, a state marking unit, a dangerous road edge detection unit, and a space-time complement unit. The dangerous road edge sensing device is used for executing part or all of the steps of the dangerous road edge sensing method in the embodiment 1, so as to realize dangerous road edge sensing in the running process of the robot. Specifically:
the grid processing unit is used for carrying out grid processing on the two side areas or any one side area of the robot task route to obtain a grid space on the corresponding side;
the point cloud projection unit is used for acquiring laser radar point clouds of a current frame of the sensor on the robot, and projecting the laser radar point clouds to the rasterization space according to the scanning line sequence of the laser radar to obtain rasterization point clouds on the corresponding side;
the state marking unit is used for marking the state of each grid in the rasterization point cloud based on the adjacent grid height gradient constraint and the grid distance constraint;
the dangerous road edge detection unit is used for carrying out window filtering on the marking result of the state marking unit to obtain a dangerous grid detection result, namely a dangerous road edge sensing result of the current frame;
the space-time complement unit is used for performing space-time complement on the dangerous road edge perception result of the current frame based on the dangerous road edge perception result of the previous N frames to obtain the final dangerous road edge detection result of the current frame and outputting the final dangerous road edge detection result.
In this embodiment, the specific working processes and working principles of the grid processing unit, the point cloud projection unit, the state marking unit, the dangerous road edge detection unit and the space-time complement unit are the same as those of the method in embodiment 1, so that the detailed description thereof is omitted in this embodiment.
Example 3
Fig. 5 shows a terminal device disclosed in this embodiment, which includes a transmitter, a receiver, a memory, and a processor. The transmitter is used for transmitting instructions and data, the receiver is used for receiving instructions and data, the memory is used for storing computer-executable instructions, and the processor is used for executing the computer-executable instructions stored in the memory so as to realize part or all of the steps executed by the dangerous road edge sensing method in the embodiment 1. The specific implementation process is the same as that of the dangerous road edge sensing method in the embodiment 1.
It should be noted that the memory may be separate or integrated with the processor. When the memory is provided separately, the terminal device further comprises a bus for connecting the memory and the processor.
In a specific application process, the terminal equipment is a computer, an unmanned vehicle, an unmanned plane, an unmanned device or a mobile robot and the like.
Example 4
The present embodiment discloses a computer readable storage medium, in which computer executable instructions are stored, and when the processor executes the computer executable instructions, part or all of the steps executed by the dangerous road edge sensing method in the above embodiment 1 are implemented.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the invention, and all equivalent structural changes made by the description of the present invention and the accompanying drawings or direct/indirect application in other related technical fields are included in the scope of the invention.
Claims (10)
1. The dangerous road edge sensing method is characterized by comprising the following steps of:
step 1, rasterizing the two side areas or any one side area of a robot task route to obtain a rasterizing space at the corresponding side;
step 2, acquiring a laser radar point cloud of a current frame of a sensor on a robot, and projecting the laser radar point cloud to the rasterization space according to a scanning line sequence of a laser radar to obtain a rasterization point cloud at a corresponding side;
step 3, in the rasterization point cloud, carrying out state marking on each grid based on the adjacent grid height gradient constraint and the grid distance constraint;
and step 4, window filtering is carried out on the marking result in the step 3, so that a dangerous grid detection result is obtained, and a dangerous road edge perception result of the current frame is obtained.
2. The method of claim 1, wherein in step 3, status marking each grid based on neighboring grid height gradient constraints and grid distance constraints comprises:
the method comprises the steps of taking the height of the highest laser radar point in each grid in a gridding point cloud as the grid height of a corresponding grid, marking the grid with the grid height being larger than the first height as positive occupation, marking the grid with the grid height being smaller than the second height as negative occupation, and marking the grid with the grid height being smaller than or equal to the first height and larger than or equal to the second height as unknown occupation;
and on the part of the gridding point cloud of each scanning line of the laser radar, updating the marking state of the grids marked as unknown occupation based on the adjacent grid height gradient constraint and the grid distance constraint, and outputting the marking result of each grid in the gridding point cloud.
3. The dangerous road edge sensing method according to claim 2, wherein the updating the marking status of the grid marked as the unknown occupancy based on the adjacent grid height gradient constraint and the grid distance constraint on the part of the gridding point cloud to which each scanning line of the laser radar belongs comprises:
step 301, extracting all grids covered by the current scanning line from the rasterization point cloud, and sequencing the extracted grids in a clockwise direction or a counterclockwise direction according to an angle direction to obtain a grid sequence, wherein i=1;
step 302, acquiring the ith grid marked as unknown occupation in the grid sequence, and defining the ith grid as a grid i;
step 303, acquiring the (i+1) th grid marked as unknown occupation in the grid sequence, and defining the grid as the (i+1) th grid;
step 304, obtaining the height difference between the grids i and i+1Distance d, and judge->And d satisfies both the neighboring grid height gradient constraint and the grid distance constraint:
if yes, go to step 305;
otherwise, go to step 306;
step 305, updating the state of the grid with the lower height in the grid i and the grid i+1 from unknown occupancy to negative occupancy, and then making i=i+2, and performing step 307;
otherwise, let i=i+1 followed by step 307;
step 307, determining whether i > M is true, where M is the number of grids marked as unknown occupancy in the grid sequence:
if yes, go to step 308;
otherwise, returning to step 302;
step 308, performing steps 301 to 307 on the part of the gridding point cloud to which each scanning line of the laser radar belongs, so as to obtain a marking result of each grid in the gridding point cloud.
4. A dangerous road edge sensing method according to claim 3, wherein in step 304, when When, i.e. the neighboring grid height gradient constraint is satisfied, wherein +.>Is the ground level threshold for grid i.
5. The method of claim 4, wherein the ground level height threshold valueThe calculation process of (1) is as follows:
wherein d i For the distance of grid i from the center of the robot, L is the distance threshold,is the ground height threshold of the grid within L meters from the center of the robot.
6. The method of claim 3 to 5, wherein in step 304, when d inf ≤d≤d sup When, i.e. satisfying the grid distance constraint, where d inf D is the shortest effective distance sup Is the most effective distance.
7. The dangerous road edge perception method according to any one of claims 1 to 5, further comprising:
and 6, performing space-time complementation on the dangerous road edge sensing result of the current frame based on the dangerous road edge sensing result of the previous N frames to obtain a final dangerous road edge detection result of the current frame, and outputting the final dangerous road edge detection result.
8. A dangerous road edge sensing device, characterized in that it employs the method of any one of claims 1 to 7, and comprises:
the grid processing unit is used for carrying out rasterization processing on the two side areas or any one side area of the robot task route to obtain a rasterization space on the corresponding side;
the point cloud projection unit is used for acquiring the laser radar point cloud of the current frame of the sensor on the robot, and projecting the laser radar point cloud to the rasterization space according to the scanning line sequence of the laser radar to obtain the rasterization point cloud at the corresponding side;
a status marking unit for status marking each grid based on the neighboring grid height gradient constraint and the grid distance constraint in the rasterized point cloud;
and the dangerous road edge detection unit is used for carrying out window filtering on the marking result of the state marking unit to obtain a dangerous grid detection result, namely obtaining a dangerous road edge sensing result of the current frame.
9. A terminal device, characterized in that the terminal device is provided with:
a memory for storing a program;
a processor for executing the program stored in the memory, the processor being for performing part or all of the steps of the dangerous road edge perception method according to any one of claims 1 to 7 when the program is executed.
10. A computer-readable storage medium having stored therein computer-executable instructions; the computer-executable instructions, when executed by a processor, for performing part or all of the steps of the hazard road edge awareness method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310977709.6A CN116990832A (en) | 2023-08-04 | 2023-08-04 | Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310977709.6A CN116990832A (en) | 2023-08-04 | 2023-08-04 | Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116990832A true CN116990832A (en) | 2023-11-03 |
Family
ID=88531774
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310977709.6A Pending CN116990832A (en) | 2023-08-04 | 2023-08-04 | Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116990832A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110320504A (en) * | 2019-07-29 | 2019-10-11 | 浙江大学 | A kind of unstructured road detection method based on laser radar point cloud statistics geometrical model |
| WO2021219353A1 (en) * | 2020-04-30 | 2021-11-04 | Zf Friedrichshafen Ag | Autonomous and/or assisted coupling of a trailer, taking into consideration the height profile of the ground |
| CN114359866A (en) * | 2021-12-31 | 2022-04-15 | 中国第一汽车股份有限公司 | Road boundary detection method and device based on laser radar |
| CN114581753A (en) * | 2022-01-14 | 2022-06-03 | 长沙行深智能科技有限公司 | Method, system and equipment for completing negative obstacles in blind area based on occupancy grid |
| CN114779206A (en) * | 2022-03-09 | 2022-07-22 | 三一智矿科技有限公司 | Method and device for identifying road boundary, storage medium and equipment |
| 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 |
| US20230003893A1 (en) * | 2022-02-23 | 2023-01-05 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and apparatus for detecting operating terrain, and engineering equipment for detecting operating terrain |
| WO2023040437A1 (en) * | 2021-09-18 | 2023-03-23 | 北京京东乾石科技有限公司 | Curbstone determination method and apparatus, and device and storage medium |
-
2023
- 2023-08-04 CN CN202310977709.6A patent/CN116990832A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110320504A (en) * | 2019-07-29 | 2019-10-11 | 浙江大学 | A kind of unstructured road detection method based on laser radar point cloud statistics geometrical model |
| WO2021219353A1 (en) * | 2020-04-30 | 2021-11-04 | Zf Friedrichshafen Ag | Autonomous and/or assisted coupling of a trailer, taking into consideration the height profile of the ground |
| WO2023040437A1 (en) * | 2021-09-18 | 2023-03-23 | 北京京东乾石科技有限公司 | Curbstone determination method and apparatus, and device and storage medium |
| CN114359866A (en) * | 2021-12-31 | 2022-04-15 | 中国第一汽车股份有限公司 | Road boundary detection method and device based on laser radar |
| CN114581753A (en) * | 2022-01-14 | 2022-06-03 | 长沙行深智能科技有限公司 | Method, system and equipment for completing negative obstacles in blind area based on occupancy grid |
| US20230003893A1 (en) * | 2022-02-23 | 2023-01-05 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and apparatus for detecting operating terrain, and engineering equipment for detecting operating terrain |
| 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 |
| CN114779206A (en) * | 2022-03-09 | 2022-07-22 | 三一智矿科技有限公司 | Method and device for identifying road boundary, storage medium and equipment |
Non-Patent Citations (3)
| Title |
|---|
| RYAN D. MORTON: "Positive and negative obstacle detection using the HLD classifier", 《2011 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS》, 5 December 2011 (2011-12-05), pages 1579 - 1584 * |
| 林辉: "基于车载多激光雷达的地图构建与障碍物检测", 《中国优秀硕士学位论文全文数据库信息科技》, 15 January 2018 (2018-01-15) * |
| 郭景明: "基于多线激光雷达的可行驶区域提取和目标分割技术研究", 《中国优秀硕士学位论文全文数据库信息科技》, 15 January 2021 (2021-01-15) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110687539B (en) | Parking space detection method, device, medium and equipment | |
| US10260892B2 (en) | Data structure of environment map, environment map preparing system and method, and environment map updating system and method | |
| CN108873013B (en) | Method for acquiring passable road area by adopting multi-line laser radar | |
| WO2018181974A1 (en) | Determination device, determination method, and program | |
| CN110632617B (en) | Laser radar point cloud data processing method and device | |
| CN112455430B (en) | Method for detecting inclined parking places without parking place lines, parking method and parking system | |
| CN111175788B (en) | Transverse positioning method and positioning system for automatic driving vehicle | |
| US11709260B2 (en) | Data driven resolution function derivation | |
| CN112744217B (en) | Collision detection method, travel path recommendation device, and storage medium | |
| CN114926809A (en) | Passable area detection method and device, moving tool and storage medium | |
| CN110203192A (en) | Location technology selection | |
| CN115235500A (en) | Lane line constraint-based pose correction method and device and all-condition static environment modeling method and device | |
| CN110850859A (en) | Robot and obstacle avoidance method and obstacle avoidance system thereof | |
| CN112805766A (en) | Apparatus and method for updating detailed map | |
| CN112629548A (en) | Three-dimensional grid map creating and updating method based on roadside laser radar | |
| CN113933859A (en) | Pavement and two-side retaining wall detection method for unmanned mine card driving scene | |
| CN113859228A (en) | Vehicle control method and device, electronic equipment and storage medium | |
| CN115164919A (en) | Method and device for constructing spatial travelable area map based on binocular camera | |
| KR102367138B1 (en) | Method of detection crosswalk using lidar sensor and crosswalk detection device performing method | |
| CN116990832A (en) | Dangerous road edge perception method, dangerous road edge perception device, terminal equipment and storage medium | |
| CN112183157A (en) | Road geometry identification method and device | |
| US20170148326A1 (en) | Device and method for detecting surrounding vehicles | |
| CN112020722B (en) | Three-dimensional sensor data-based road shoulder identification | |
| CN110909569B (en) | Road condition information identification method and terminal equipment | |
| KR102423781B1 (en) | Method of detecting guard-rail using lidar sensor and guard-rail detection device performing method |
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 |