WO2023092835A1

WO2023092835A1 - Obstacle avoidance during edgewise sweeping of unmanned sweeper

Info

Publication number: WO2023092835A1
Application number: PCT/CN2022/071307
Authority: WO
Inventors: 黄超; 叶玥
Original assignee: 上海仙途智能科技有限公司
Priority date: 2021-11-26
Filing date: 2022-01-11
Publication date: 2023-06-01
Also published as: CN115240160A

Abstract

A method and apparatus for obstacle avoidance during edgewise sweeping of an unmanned sweeper. The method comprises: before collisions with obstacles can occur, acquiring a point cloud and an image of each obstacle; segmenting the acquired point cloud; classifying points in the point cloud of each obstacle; then, matching the points in the point cloud of each obstacle with a preset ignoring rule; determining, according to a matching result, whether to execute ignoring processing; and then, planning a sweeping path of an unmanned sweeper according to a processing result. The ignoring processing is introduced into an obstacle processing method, such that the detour frequency of an unmanned sweeper can be reduced, thereby improving the edgewise sweeping quality and reducing potential safety hazards. In addition, a preset ignoring rule comprises a tree ignoring rule, and points in the point cloud of each obstacle that match the preset ignoring rule need to be located in a tree area that is marked in an offline map, and the tree area is updated periodically, such that the influence on the matching accuracy caused by matching errors and changes in the tree area due to seasons and manual pruning can be reduced.

Description

Obstacle avoidance when unmanned sweeper sweeps side by side

technical field

The present application relates to the field of intelligent driving, and in particular to a method, device and equipment for avoiding obstacles when an unmanned sweeper sweeps alongside.

Background technique

Urban garbage is easy to accumulate on the edge of the road, which is difficult to clean. The use of manual driving sanitation vehicles for edge cleaning requires the driver to master good driving skills, which brings high time and labor costs. The use of unmanned sweeping vehicles for edge cleaning can reduce cleaning costs, but in order to avoid collisions as much as possible, unmanned sweeping vehicles perform detour operations on all identified obstacles, which will reduce the cleaning effect and bring safety Hidden danger.

Contents of the invention

In order to overcome the problems existing in the prior art, the present application provides a method, device and equipment for avoiding obstacles when an unmanned sweeper sweeps alongside.

According to the first aspect of the present application, there is provided an obstacle avoidance method when an unmanned sweeper cleans side by side, the method includes: before colliding with the obstacle, obtaining the point cloud of each obstacle and each obstacle The image of each obstacle in the point cloud is an obstacle point; the point cloud of each obstacle is segmented; the image of each obstacle and the segmentation result of the point cloud are used to segment the Obstacle points are classified; according to the type of the obstacle point and the data in the point cloud of each obstacle, the preset ignoring rules are matched, the ignoring rules include tree ignoring rules, and the tree ignoring rules The matching conditions for the obstacle point include that the obstacle point is located in the tree area marked in the offline map, and the tree area marked in the offline map is updated according to a preset cycle; according to the matching result, it is determined whether the obstacle The object point is ignored, and the ignored process is not to perform a detour operation; the cleaning path of the unmanned sweeper is planned according to the processing result of the obstacle point.

According to the second aspect of the present application, there is provided an obstacle avoidance device for an unmanned sweeper during side-by-side cleaning, the device includes: an acquisition module, used to acquire the point cloud and For the image of each obstacle, the points in the point cloud of each obstacle are obstacle points; the segmentation module is used to segment the point cloud of each obstacle; the classification module is used to combine the points of each obstacle The image of the obstacle and the segmentation result of the point cloud are used to classify the obstacle points; the matching module is used to match the preset data with the type of the obstacle point and the data in the point cloud of each obstacle. Matching is performed by ignoring rules, the ignoring rules include tree ignoring rules, and the conditions met by the obstacle points matched with the tree ignoring rules include that the obstacle points are located in the tree area marked in the offline map, and the offline map The tree area marked in is updated according to a preset period; the determination module is used to determine whether to perform ignore processing on the obstacle point according to the matching result, and the ignore processing is not to perform a detour operation; the planning module is used to determine according to the matching result The cleaning path of the unmanned cleaning vehicle is planned based on the processing results of the obstacle points.

According to a third aspect of the present application, there is provided an electronic device, including: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it is used to implement the first aspect the method described.

According to a fourth aspect of the embodiments of the present application, there is provided a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in the above-mentioned first aspect are implemented.

The technical solution provided by the embodiment of the present application may include the following beneficial effects: In the embodiment of the present application, before the collision with the obstacle, when the unmanned sweeper cleans up the garbage on the edge of the road, the obstacle is handled in addition to Ignore processing is introduced outside the line, that is, the detour operation is not performed on the identified obstacles. Firstly, according to the acquired point cloud of each obstacle and the image information of each obstacle, the points in the point cloud of each obstacle (that is, obstacle points) are classified, and the data in the point cloud of each obstacle and the image information of each obstacle are combined. The point type is matched with the preset ignore rules, and whether to ignore each obstacle point is determined according to the matching effect, and then the cleaning path is planned according to whether to execute the ignore operation. Since the unmanned sweeper no longer detours all obstacles, it can solve the problem of cleaning quality degradation and safety hazards caused by frequent detours of the unmanned sweeper. In addition, the obstacle point matching the tree ignoring rule needs to be located in the tree area marked in the offline map to reduce the error that may occur during matching, and the tree area marked in the offline map is updated regularly to reduce seasonal changes, artificial pruning, etc. impact on the accuracy of matching results.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description serve to explain the principles of the application.

Fig. 1 is a flow chart of an obstacle avoidance method for an unmanned sweeper when sweeping alongside according to an exemplary embodiment of the present application.

Fig. 2 is a schematic diagram of a tree area marked in an offline map according to an exemplary embodiment of the present application.

Fig. 3 is a schematic diagram of an unmanned sweeper handling shrub obstacles according to an exemplary embodiment of the present application.

Fig. 4 is a flow chart of an obstacle avoidance method for an unmanned sweeper when sweeping alongside according to another exemplary embodiment of the present application.

Fig. 5 is a schematic structural diagram of an obstacle avoidance device for an unmanned sweeper when sweeping alongside according to an exemplary embodiment of the present application.

Fig. 6 is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present application.

Detailed ways

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present application as recited in the appended claims.

The terminology used in this application is for the purpose of describing particular embodiments only, and is not intended to limit the application. As used in this application and the appended claims, the singular forms "a", "the", and "the" are intended to include the plural forms as well, unless the context clearly dictates otherwise. It should also be understood that the term "and/or" as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of the present application, first information may also be called second information, and similarly, second information may also be called first information. Depending on the context, the word "if" as used herein may be interpreted as "at" or "when" or "in response to a determination."

Urban garbage is easy to accumulate on the edge of the road that is difficult to clean. Using a manual driving sanitation vehicle to clean the edge requires the driver to be proficient in driving and operating techniques, which brings high time and labor costs. The use of unmanned sweeping vehicles for edge cleaning can reduce cleaning costs and complete edge cleaning tasks without manual intervention. However, in order to avoid collisions as much as possible to ensure driving safety, the unmanned sweeper performs detour operations on all identified obstacles, and will not further judge whether the obstacles will actually collide with them, which will lead to cleaning The effect drops. In addition, unmanned sweeping vehicles frequently occupy the outer lanes due to frequent detours, which poses safety hazards.

In view of the above problems, the present application proposes an obstacle avoidance method when an unmanned sweeper sweeps side-by-side, and introduces the process of ignoring obstacles. Classify the identified obstacles before colliding with them, judge whether the identified obstacles will actually collide with the unmanned cleaning vehicle according to the type, height, location and other information of the obstacles and judge after the collision Whether it will affect the driving safety of unmanned cleaning vehicles, ignore the obstacles that will not actually cause collisions or will not affect driving safety after collisions, that is, do not perform detour operations. The judging method is to match the above items of information of the obstacle with preset rules, and determine whether to perform detour processing according to the matching result. As shown in Figure 1, the above obstacle avoidance method will be described in detail next.

S101. Before colliding with an obstacle, obtain the point cloud of each obstacle and the image of each obstacle, and the points in the point cloud of each obstacle are obstacle points; During cleaning, identifying existing obstacles and dealing with them all take place before collisions with obstacles. The point cloud of each obstacle includes a plurality of points, and the points in the point cloud of each obstacle are referred to as obstacle points in this application. The obtained point cloud of each obstacle contains the three-dimensional space information of each obstacle point, and each obstacle point corresponds to a three-dimensional space coordinate, which can reflect the height information of each obstacle point. The image of each obstacle is acquired to classify each obstacle point. There are many ways to acquire point clouds, such as using lidar, binocular camera, and depth camera. According to different acquisition methods, each point in the point cloud can also contain color information (RGB value) and reflection intensity information.

S102. Segment the point cloud of each obstacle; the acquired point cloud of each obstacle is actually a collection of points, and it is unknown which obstacle a point (ie, the above-mentioned obstacle point) belongs to Therefore, segmenting the point cloud of each obstacle is to divide the obstacle points belonging to the same obstacle into one piece, so as to distinguish different obstacles, and the obstacle points in the divided obstacle point cloud have similar characteristics.

S103, classify the obstacle points in combination with the images of the obstacles and the segmentation results of the point cloud;

The segmented point cloud of each obstacle is one-to-one matched with each obstacle in the acquired image, so as to know the type of each divided obstacle point cloud. As an example, the divided point cloud of each obstacle can be projected onto the image of each obstacle above, and the type of obstacle point in the point cloud of each obstacle can be known according to the position where the point cloud projection of each obstacle is located. .

S104. According to the type of the obstacle point and the data in the point cloud of each obstacle, match with a preset ignore rule, the ignore rule includes a tree ignore rule, and the tree ignore rule matches the The conditions that the obstacle point meets include that the obstacle point is located in the tree area marked in the offline map, and the tree area marked in the offline map is updated according to the preset cycle; the data in the point cloud of each obstacle is the above-mentioned The information of each obstacle point contained in the obstacle point cloud includes the three-dimensional space coordinates of each obstacle point, and the information of each obstacle point according to the different acquisition methods of each obstacle point cloud can also include color information (RGB value ) and reflection intensity information. The preset ignoring rules are formulated for obstacles that will not actually collide with the unmanned sweeper or will not affect the driving safety of the unmanned sweeper after the collision, including the tree ignoring rule, and the obstacle points that match the tree ignoring rule need It is located in the tree area 110 marked in the offline map, that is, the closed polygonal area 110 shown in FIG. 2 . In addition, since the tree area may change due to seasons, manual pruning, etc., which will affect the accuracy of the matching results, the tree area 110 marked in the offline map needs to be updated regularly, and the update cycle can be set according to seasonal changes or historical operating data , can also be set by referring to other factors, which is not limited in this application.

The tree area 110 marked in the offline map can be generated by manual marking, but this method requires a huge amount of work when implemented in a large area and multiple regions, and is almost infeasible. Therefore, in an embodiment of the present application, an automatic generation method is adopted, and in order to ensure the reliability of the data, the tree area 110 marked in the offline map is based on the history of tree obstacles collected by at least one unmanned sweeper in at least one day Point cloud generation. The process of automatically generating the tree area 110 marked in the above-mentioned offline map may include the following steps. First, obtain the above-mentioned historical point cloud of tree obstacles collected by at least one unmanned sweeper within at least one day, and then filter out the historical points misdetected points in the cloud, and then cluster the filtered historical point cloud 130 (as shown in Figure 2), and finally carry out convex processing on the point clouds whose number of points in the clustered various historical point clouds is greater than the preset value. Packetization can generate the tree area 110 marked in the offline map, and the default value is set according to the actual operating conditions of the unmanned sweeper and its own needs. Since the obtained historical point cloud family density does not change much, the number of families is not fixed, and there is no strict requirement on the processing speed when performing offline processing, the clustering algorithm used in the process of generating the tree area 110 marked in the offline map has Various, such as KNN algorithm, K-means algorithm, Mean-shift algorithm. In addition, there are many algorithms used for convex hulling the clustered historical point clouds whose number of points is greater than the preset value, such as Graham Scan algorithm, Jarvis algorithm, and Melkman algorithm.

In yet another embodiment of the present application, the part of filtering false detection points in the process of generating the tree region 110 marked in the above-mentioned offline map includes first dividing the offline map into blocks, and then performing three-dimensional analysis on each divided block. Rasterize, and project the obtained historical point cloud into each grid, then filter the points in the historical point cloud of each grid that do not meet the preset voting mechanism, and finally merge the filtered points Historical point cloud in raster. The preset voting mechanism can be set according to the historical operation data of the unmanned sweeper. For example, if there are n days of historical point clouds collected in a divided block, there are more than 2n/3 days of historical point clouds in the If there are projected points in a certain grid in this block, then the point cloud in this grid is used, otherwise, the historical point cloud projected into this grid is considered to be false detection points.

The tree ignoring rule is aimed at tree-type obstacles located on the roadside, including tree-like obstacles with some branches and leaves sticking out of the roadside, and the falling leaves will collide with unmanned sweepers but will not affect driving safety. In an embodiment of the present application, in addition to satisfying the condition of being located in the tree area 110 marked in the offline map, the obstacle point matching the tree ignoring rule also needs to satisfy that the type of the obstacle point is a tree, and the obstacle point is in the Within the preset range of the working range of the unmanned sweeper, and the height of the obstacle point is greater than the preset height. The value of the preset range can be set as required, and it can be a circle with the unmanned sweeper as the center, or a rectangular area where the unmanned sweeper is located, which is not limited in this application. The preset height can be set according to the size, body height and historical operation data of the unmanned sweeper under the condition of ensuring that even if a collision occurs, it will not affect driving safety.

In addition to the tree-type obstacles targeted by the tree-ignoring rules, there are other obstacles that meet the above two situations that the preset neglecting rules target. In an embodiment of the present application, the preset ignoring rules also include a first ignoring rule, and the first ignoring rule is aimed at obstacles located above the sweeping brush of the unmanned sweeper and will not collide with the unmanned sweeper, These obstacles are located on the curb but partially protrude from the curb. In consideration of computational efficiency, the obtained point cloud is projected onto a two-dimensional plane during collision detection. There are some obstacles that are located above the brush but do not actually collide with the brush in the prior art It is also considered that a collision may occur and a detour operation is performed, thus affecting the effect of welt cleaning, such as ornamental shrubs 330 on the curb surface, as shown in FIG. 3 . In order to facilitate the understanding of the various conditions in the first ignoring rule, it will be explained in conjunction with FIG. 3 , the conditions met by the obstacle point matching the first ignoring rule include:

(1) The height 210 of the obstacle point is greater than the height 220 of the brush 320 in the unmanned sweeper;

(2) And the type of obstacle point is non-human and non-vehicle;

(3) And the obstacle point is within the preset range of the working range of the unmanned sweeper;

(4) And the obstacle point is inside the roadside section; the roadside section is generated according to the detection results of the roadside, and the roadside section is located at the intersection of the vertical surface of the roadside and the driving road surface of the unmanned sweeper. The inner side of the road along the segment. The line segment 400 in Figure 3 is in the same plane as the vertical plane along the road and is perpendicular to the line segment along the road. From a top view, it is a point on the line segment along the road, which can be understood as a line segment along the road in Figure 3 The inner side and the outer side of the roadside section, and the side where the unmanned sweeper is located is the inner side 401 of the roadside section.

(5) and there is at least one point in the obstacle point cloud where the obstacle point is located is outside 402 of the roadside segment;

(6) And the horizontal distance 230 between the obstacle point and the obstacle point cloud where the obstacle point is located is the closest to the unmanned sweeper and the point on the inside of the road along the line segment is less than the body 310 in the unmanned sweeper. The distance 240 between the edges of the brushes 320 in the unmanned cleaning vehicle.

In the situation shown in FIG. 3 , the parts of the bushes 330 on the inner side 401 of the roadside segment are ignored by the unmanned sweeper. It should be noted that Figure 3 is only for illustration and does not represent all situations. The height 210 of the obstacle point and the obstacle point cloud where the obstacle point is located are closest to the unmanned sweeper and along the road. The lateral distance 230 between the points on the segment inner side 401 is determined according to the three-dimensional space positions of different obstacle points, and is not limited to the situation shown in FIG. 3 .

To determine whether the obstacle point is located on the inside or outside of the roadside segment, the vector cross multiplication method can be used to calculate the distance from the obstacle point to the roadside segment, and the distance from the obstacle point on the inside of the roadside segment to the roadside segment is set as positive. The distance from the obstacle point on the outside of the road along the line to the road along the line is negative, or the positive and negative relations corresponding to the inside and outside of the road along the line are exchanged, according to the positive and negative conditions of the calculated distance from the obstacle point to the road along the line Then it can be known that the obstacle point is inside or outside of the roadside segment.

In another embodiment of the present application, the preset ignoring rules also include a second ignoring rule. The second ignoring rule is aimed at obstacles located on the curb that will not intersect with the unmanned sweeper at all, and ignore them. The operation can improve the calculation efficiency; and the false detection point that is misjudged as an obstacle due to the unevenness of the roadside, the height of the above false detection point is very low, it will collide with the sweeping brush but will not affect the edge of the unmanned sweeper Influence, ignoring false detection points can improve the quality of edge cleaning. The conditions met by the obstacle point matching the second ignore rule include: the obstacle point is within the preset range of the working range of the unmanned sweeper, and the type of the obstacle point is non-human and non-vehicle, and the obstacle point The type is the roadside or on the outside of the roadside section, wherein the roadside section is generated according to the detection results of the roadside and is located at the intersection of the vertical surface of the roadside and the driving road surface of the unmanned sweeper. The inner side of the road along the line segment. The method of judging whether the obstacle point is located on the inside or outside of the roadside segment is similar to the above, and will not be repeated here.

S105. Determine whether to perform ignore processing on the obstacle point according to the matching result, and the ignore processing is not to perform a detour operation; in an embodiment of the present application, if the obstacle point matches any preset ignore rule The obstacle point is then ignored.

S106. Planning the cleaning path of the unmanned cleaning vehicle according to the processing result of the obstacle point.

Below, in combination with a preferred embodiment, an obstacle avoidance method for an unmanned sweeping vehicle proposed in this application is described, and the specific steps are shown in Figure 4:

S201. Before colliding with the obstacle, use the laser radar to obtain the point cloud of each obstacle within the working range of the unmanned sweeper, use the camera to obtain the image of each obstacle, and the points in the point cloud of each obstacle is an obstacle point, and each obstacle point corresponds to a three-dimensional space coordinate;

S202. Segment the point cloud of each obstacle;

S203. Project the segmented point cloud onto the images of the obstacles, and classify the obstacle points according to the projected positions of the point clouds on the obstacle images;

S204. According to the type of the obstacle point and the three-dimensional space coordinates of the obstacle point, match with a preset tree ignoring rule, and the condition that the obstacle point matching the tree ignoring rule satisfies includes the obstacle The type of the object point is a tree, and it is within the radius of 1 meter of the unmanned sweeper, and the height of the obstacle point is higher than 1 meter, and the obstacle point is located in the tree area marked in the offline map, The tree area marked in the offline map is updated according to a preset cycle;

S205, if the obstacle point matches the tree ignoring rule, perform ignore processing on the obstacle point;

S206. Planning the cleaning path of the unmanned cleaning vehicle according to whether to ignore the obstacle point.

Based on the above-mentioned method embodiment, the present application provides an obstacle avoidance device for unmanned sweeping vehicles during side-by-side cleaning, as shown in Figure 5, the device includes:

The obtaining module 510 is used to obtain the point cloud of each obstacle and the image of each said obstacle before colliding with the obstacle, and the points in the point cloud of each said obstacle are obstacle points;

A segmentation module 520, configured to segment the point cloud of each obstacle;

A classification module 530, configured to classify the obstacle points in combination with the images of the obstacles and the segmentation results of the point cloud;

The matching module 540 is configured to match preset ignore rules according to the type of the obstacle point and the data in the point cloud of each obstacle, the ignore rules include tree ignore rules, and the tree ignore rules The conditions met by the matched obstacle point include that the obstacle point is located in the tree area marked in the offline map, and the tree area marked in the offline map is updated according to a preset cycle;

A determining module 550, configured to determine whether to perform ignore processing on the obstacle point according to the matching result, the ignore processing being not to perform a detour operation;

The planning module 560 is configured to plan the cleaning path of the unmanned cleaning vehicle according to the processing results of the obstacle points.

For the implementation process of the functions and effects of each module in the above-mentioned device, please refer to the implementation process of the corresponding steps in the above-mentioned method for details, and details will not be repeated here.

The embodiment of the obstacle avoidance device when the unmanned sweeper sweeps side by side in this application can be applied to electronic equipment, and the device embodiment can be realized by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory for operation through the processor of the file processing where it is located. From the perspective of hardware, as shown in Figure 6, it is a hardware structure diagram of the computer equipment where the file processing device of the embodiment of this specification is located, except for the processor 610, memory 630, network interface 620, and non-volatile memory shown in Figure 6 In addition to the volatile memory 640, the electronic device in which the device 631 is located in the embodiment may generally include other hardware according to the actual function of the electronic device, and details will not be repeated here.

In addition, corresponding to the above-mentioned method embodiment, the present application also provides a computer-readable storage medium, the storage medium stores a computer program, and the computer program is used to perform the side-cleaning of the unmanned sweeping vehicle described in the present application. when the obstacle avoidance method.

The foregoing describes specific embodiments of this specification. Other implementations are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. Multitasking and parallel processing are also possible or may be advantageous in certain embodiments. The various technical features in the above embodiments can be combined arbitrarily, as long as there is no conflict or contradiction between the combinations of features, but due to space limitations, they are not described one by one, so the various technical features in the above embodiments can be combined arbitrarily It also belongs to the scope of this specification application.

Other embodiments of the description will readily occur to those skilled in the art from consideration of the specification and practice of the invention claimed herein. This description is intended to cover any modification, use or adaptation of this description. These modifications, uses or adaptations follow the general principles of this description and include common knowledge or conventional technical means in this technical field for which this description does not apply . The specification and examples are to be considered exemplary only, with a true scope and spirit of the specification indicated by the appended claims.

The above descriptions are only preferred embodiments of this specification, and are not intended to limit this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this specification shall be included in this specification. within the scope of protection.

Claims

A method for avoiding obstacles when an unmanned sweeper sweeps alongside, comprising:

Before colliding with the obstacle, obtain the point cloud of each obstacle and the image of each said obstacle, and the points in the point cloud of each said obstacle are obstacle points;

Segmenting the point cloud of each obstacle;

classifying the obstacle points in combination with the images of the obstacles and the segmentation results of the point cloud;

According to the type of the obstacle point and the data in the point cloud of each obstacle, match with the preset ignore rule, the ignore rule includes the tree ignore rule, and the obstacle matched with the tree ignore rule The condition that the point satisfies includes that the obstacle point is located in the tree area marked in the offline map, and the tree area marked in the offline map is updated according to a preset cycle;

Determine whether to perform ignore processing on the obstacle point according to the matching result, and the ignore processing is not to perform a detour operation;

The cleaning path of the unmanned cleaning vehicle is planned according to the processing results of the obstacle points.
The method according to claim 1, wherein the tree area marked in the offline map is generated based on historical point clouds of tree obstacles collected by at least one of the unmanned sweeping vehicles in at least one day.
The method according to claim 2, wherein the process of generating the marked tree area in the offline map comprises:

dividing the offline map into blocks;

performing three-dimensional rasterization processing on the divided blocks, and projecting the historical point cloud into each of the rasters;

filtering points in the historical point cloud in the grid that do not satisfy a preset voting mechanism;

Merging the historical point clouds in each of the filtered grids.
The method according to claim 1, wherein the condition that the obstacle point matched with the tree ignoring rule satisfies also includes:

The type of the obstacle point is a tree, and the obstacle point is within the preset range of the working range of the unmanned sweeper, and the height of the obstacle point is greater than the preset height.
The method according to claim 1, wherein the ignoring rule further comprises a first ignoring rule, and the obstacle point matching the first ignoring rule satisfies a condition comprising:

The height of the obstacle point is greater than the height of the brush in the unmanned sweeper;

And the type of the obstacle point is non-human and non-vehicle;

And the obstacle point is within the preset range of the working range of the unmanned sweeper;

And the obstacle point is inside the roadside segment, which is generated according to the curb detection result and is located at the intersection of the vertical surface of the roadside and the driving road surface of the unmanned sweeper, and the unmanned sweeper the vehicle is on the inside of said kerb segment;

And there is at least one point in the obstacle point cloud where the obstacle point is located is outside the roadside segment;

And the lateral distance between the obstacle point and the point in the obstacle point cloud where the obstacle point is located is closest to the unmanned sweeper and inside the road along the line segment is smaller than that of the unmanned sweeper The distance from the middle body to the edge of the brush in the unmanned sweeping vehicle.
The method according to claim 1, wherein the ignoring rule further includes a second ignoring rule, and the conditions met by the obstacle point matching the second ignoring rule include:

The obstacle point is within the preset range of the working range of the unmanned sweeper;

And the type of the obstacle point is non-human and non-vehicle;

And the type of the obstacle point is the roadside or the outside of the roadside line segment, the roadside line segment is generated according to the roadside detection result and is located at the intersection of the vertical surface of the roadside and the driving road surface of the unmanned sweeper , the unmanned sweeper is on the inner side of the road along the line segment.
The method according to claim 1, wherein the determining whether to ignore the obstacle point according to the matching result comprises:

If the obstacle point matches any one of the ignore rules, ignore processing is performed on the obstacle point.
An obstacle avoidance device for an unmanned sweeper when sweeping alongside, comprising:

The acquisition module is used to acquire the point cloud of each obstacle and the image of each obstacle before the collision with the obstacle, and the points in the point cloud of each obstacle are obstacle points;

A segmentation module, configured to segment the point cloud of each obstacle;

A classification module, configured to classify the obstacle points in combination with the images of the obstacles and the segmentation results of the point cloud;

A matching module, configured to match a preset ignoring rule according to the type of the obstacle point and the data in the point cloud of each obstacle, the ignoring rule includes a tree ignoring rule, and matches the tree ignoring rule The condition that the obstacle point satisfies includes that the obstacle point is located in the tree area marked in the offline map, and the tree area marked in the offline map is updated according to a preset cycle;

A determining module, configured to determine whether to perform ignore processing on the obstacle point according to the matching result, and the ignore processing is not to perform a detour operation;

A planning module, configured to plan the cleaning path of the unmanned cleaning vehicle according to the processing results of the obstacle points.
An electronic device comprising:

processor; and

memory for storing said processor-executable instructions;

Wherein, when the processor executes the executable instruction, it is used to realize the method described in any one of claims 1-7.
A computer-readable storage medium, on which a computer program is stored, characterized in that, when the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.