Lidar-based automatic pruning method and system for three-dimensional modeling of plant shapes

Abstract

The present invention proposes a method and system for automatic pruning of plant shapes based on laser radar, comprising: obtaining plant shape data of the plant to be pruned by running the laser radar on a trolley and generating three-dimensional point cloud data Cloud p ; and plant standard shape point cloud data Cloud s After high-precision adaptation processing, point cloud data of the area to be trimmed is generated cut After clustering, the area to be pruned is generated. cut ; In the area to be trimmed Areas cut The pruning path is planned to generate an operation trajectory, and the automatic pruning of the plants to be pruned is performed. The present invention automatically prunes fixed plants, solving the problem of repeated pruning of garden plants, which is labor-intensive and inefficient.

Description

TECHNICAL FIELD

[0001] The present application relates to the application of automation technology in the field of horticulture, in particular a laser radar-based automatic pruning method and system for three-dimensional modeling of plant shapes. BACKGROUND

[0002] Garden shaping is a kind of garden art that uses plants in the environment as materials to form various profile three-dimensional geometric shapes according to specific requirements of artistic concepts of the environment, and is a perfect combination of plant cultivation technology and horticultural art, and is also a means to transform and improve the beauty of the environment.

[0003] At present, there are many garden shaping plants everywhere, and there are a large number of plants that need to be pruned in communities, parks, and roadsides. The traditional pruning method is for gardeners to use pruning shears to prune one by one according to their own considerations, which is a lot of work and low in efficiency, and as the plants grow, repeated pruning is required. SUMMARY

[0004] To solve the above technical problems, the present application provides a laser radar-based automatic pruning method and system for three-dimensional modeling of plant shapes, which can automatically design a model of the shape of a plant in a computer and automatically prune according to the designed model, and can repeatedly and automatically prune fixed plants, thereby solving the problems of repeated pruning of garden shaping plants, a lot of work, and low efficiency. In particular, for long-distance, single, large-amount, and long-term repeated pruning on both sides of the road, the device and method have obvious advantages.

[0005] In a first aspect, the present application provides a laser radar-based automatic pruning method for three-dimensional modeling of plant shapes, which specifically includes the following steps:

[0006] Obtain plant shaping data of a plant to be pruned by running a laser radar on a trolley and generate three-dimensional point cloud data Cloudp;

[0007] Adapt the three-dimensional point cloud data Cloudp p to plant standard shaping point cloud data Cloud s to generate point cloud data Cloud cut of a region to be pruned after high-precision adaptation processing, and perform clustering processing on the point cloud data Cloud cut of the region to be pruned to generate a region Areas cut to be pruned.

[0008] Generate a work track by planning a pruning path in the region Areas cut to be pruned, and perform automatic pruning on the plant to be pruned.

[0009] As a further improved technical solution, the plant modeling data of the plant to be pruned is acquired by the laser radar on the running trolley, and three-dimensional point cloud data Cloud p is generated The step specifically includes:

[0010] The point cloud of the plant to be pruned is collected in real time by the laser radar on the running trolley, and the point cloud of the plant region is generated according to the prior information of satellite positioning and plant position after filtering out invalid point clouds.

[0011] Based on the field of view of the laser radar and the size of the plant, n collection points are selected, and the pose transformation relationship of each collection point relative to the initial collection point is calculated.

[0012] The point cloud of the plant region at different angles is superimposed according to the pose transformation relationship, and three-dimensional point cloud data Cloud p is generated.

[0013] As a further improved technical solution, the point cloud of the plant to be pruned is collected in real time by the laser radar on the running trolley, and the point cloud of the plant region is generated after filtering out invalid point clouds according to the prior information of satellite positioning and plant position.

[0014] The prior spatial position of the plant is set as c k =[X k , Y k , Z k ] T , assuming that the current collection is the kth plant object; the laser radar point cloud at the current frame t is Cloud p_t , the three-dimensional point coordinates in Cloud p_ t are converted into satellite latitude and longitude coordinates through coordinate conversion to obtain Cloud pGPS_t , it is judged whether the three-dimensional space points in Cloud pGPS_t belong to the space c k , if yes, keep, otherwise filter out, and generate the filtered point cloud of the plant region Cloud pGPS_t_F .

[0015] Among them, the judgment formula is as follows:

[0016]

[0017] In the formula, k represents the kth working plant or plant association, and Cloud pGPS_t_F (i) represents the ith three-dimensional space point in the filtered point cloud.

[0018] As a further improved technical solution, the step of selecting n collection points based on the field of view of the laser radar and the size of the plant, and calculating the pose transformation relationship of each collection point relative to the initial collection point includes:

[0019] The latitude and longitude coordinates of the initial acquisition point are L(lat, lon), the latitude and longitude coordinates of the ωth acquisition point are L ω (lat ω ,lon ω ), and the vehicle attitude of the running trolley at the initial acquisition point is The vehicle attitude of the ωth acquisition point is

[0020] According to the formula:

[0021]

[0022] The displacement dis of the ωth acquisition point L ω (lat ω ,lon ω ) relative to the initial acquisition point L(lat, lon) is calculated, so that the translation matrix T ω is obtained based on the relative angle and displacement of the ωth acquisition point in the initial acquisition point position.

[0023] According to the vehicle attitude of the ωth acquisition point, the rotation matrix of the vehicle attitude of the ωth acquisition point relative to the transformation of the vehicle attitude of the initial acquisition point is obtained R ω ;

[0024] Among them, the rotation matrix based on the x-axis is:

[0025]

[0026] The rotation matrix around the y-axis is:

[0027]

[0028] The rotation matrix around the z-axis is:

[0029]

[0030] R ω =R ωz ·R ωy ·R ωz

[0031] In the above formula, indicates the heading angle, θ indicates the pitch angle, and y indicates the roll angle.

[0032] As a further improved technical solution, the step of generating three-dimensional point cloud data Cloud p after the point cloud superposition of the plant regions at different angles according to the pose transformation relationship includes calculating the three-dimensional point cloud data Cloudp :

[0033] Cloud p = Cloud pGPS_t_F_P0 + Cloud pGPS_t_F_p1 · [R1 T1] + … + Cloud pGPS_t_F_pn · [R n T n ]

[0034] In the formula, Cloud pGPS_t_F_Pn represents the filtered point cloud Cloud pGPS_t_F obtained at the nth acquisition angle.

[0035] As a further improved technical solution, the three-dimensional point cloud data Cloud p is subjected to high-precision adaptive processing with the plant standard modeling point cloud data Cloud s to generate the point cloud data Cloud cut of the area to be pruned, and the point cloud data Cloud cut of the area to be pruned is subjected to clustering processing to generate the area Areas cut to be pruned.

[0036] The three-dimensional point cloud data Cloud p is encoded according to a pre-set encoding format, and is subjected to high-precision adaptation with the plant standard modeling point cloud data Cloud s through a nearest neighbor iteration algorithm, and is subjected to coordinate transformation to enter a spatial coordinate system with the ground as the X-Y plane and the Z axis as the vertical upward direction.

[0037] The space of the plant area is subjected to columnar grid division based on a spatial resolution of 3cm x 3cm, three-dimensional points of Cloud p and Cloud s falling into the columnar grid are calculated respectively as point sets Points p_G_Xi_Yi and Points s_G_Xi_Yi , and the average heights d p_G_Xi_Yi and d p_G_Xi_Yi of the points in Points s_G_Xi_Yi are calculated respectively.

[0038] Columnar grids with d p_G_Xi_Yi -d s_G_Xi_Yi ≥τ are marked as pruning grids, after all columnar grids in the space of the plant area are traversed, all grids that need to be pruned are obtained, and the plant standard modeling point cloud data falling into the pruning grids is obtained as the point cloud Cloud cut of the area to be pruned, wherein τ is a threshold greater than 0 set.

[0039] The point cloud in Cloud cut is clustered by using the Euclidean distance clustering method to generate Areas cut to be pruned.

[0040] As a further improved technical solution, the step of generating an operation trajectory by pruning path planning in the Areas cut to be pruned and performing automatic pruning of the plants to be pruned comprises:

[0041] According to the pruning space of the pruning manipulator on the running trolley, the plant pruning operation is divided into m operation points, and a corresponding relationship is established with the Areas cut to be pruned. For multiple Areas to be pruned belonging to the same operation point, they are sorted in order from left to right and from top to bottom. For the same Areas to be pruned, they are divided into several sub-areas according to the operation width. The operation key points are selected, that is, the operation start point and end point of the sub-area. The key points are connected by a hybrid A* algorithm to plan a path, generate an operation sub-trajectory of a sub-area, and realize traversal pruning. An operation trajectory is generated based on the following encoding method:

[0042] [Operation point 1: trajectory 1, …, trajectory n1;

[0043] Operation point 2: trajectory 1, …, trajectory n2;

[0044] …

[0045] Operation point n: trajectory 1, …, trajectory nn]

[0046] The running trolley performs automatic pruning of the plants to be pruned according to the operation trajectory.

[0047] In a second aspect, the present application further provides a plant modeling three-dimensional modeling automatic pruning system based on a laser radar, which comprises a running trolley, a pruning device and a back-end data processing platform. The pruning device is located on the running trolley and comprises a laser radar and two pruning manipulators. The running trolley further comprises a control module for controlling the running trolley and the pruning device and for data interaction with the back-end data processing platform.

[0048] The running trolley walks around the plants to be pruned for one round, acquires plant modeling data of the plants to be pruned by the laser radar, and generates three-dimensional point cloud data Cloudp.

[0049] The three-dimensional point cloud data Cloud p is sent to the back-end data processing platform through a TCP / IP protocol and compared with standard plant modeling point cloud data Cloud sGenerate the point cloud data Cloud of the to-be-trimmed area after high-precision adaptive processing cut And generate the to-be-trimmed area Areas after clustering processing on the point cloud data Cloud of the to-be-trimmed area cut cut In the to-be-trimmed area Areas cut Generate the operation trajectory of the trimming manipulator after trimming path planning is performed on the to-be-trimmed area Areas

[0050] Send the operation trajectory to the control module, and control the trimming manipulator to perform automatic trimming on the to-be-trimmed plant under the control of the control module.

[0051] Compared with the prior art, the present application has the following beneficial technical effects:

[0052] The present application is mainly aimed at garden modeling plant trimming, and proposes a laser radar-based automatic plant modeling three-dimensional modeling and trimming device and method, which can realize full-automatic, repeated, and less-manual plant trimming, and can efficiently complete batch trimming of garden modeling plants. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 The present application proposes a laser radar-based automatic plant modeling three-dimensional modeling and trimming method flowchart.

[0054] Figure 2 The present application proposes a laser radar-based automatic plant modeling three-dimensional modeling and trimming system framework diagram. DETAILED DESCRIPTION

[0055] For the sake of understanding, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0056] The terms used herein are only used for the purpose of describing specific example embodiments, and are not limiting. For example, unless the context clearly indicates otherwise, as used herein, the singular forms "a", "an", and "the" can also include the plural forms. When used in the present specification, the terms "comprise", "include" and / or "contain" mean that the associated integers, steps, operations, elements and / or components exist, but do not exclude the presence of one or more other features, integers, steps, operations, elements, components and / or groups, or that other features, integers, steps, operations, elements, components and / or groups can be added to the system / method.

[0057] ​These and other features, and characteristics of the present specification, as well as the methods of operation and functions of the related elements of structure and the combination of parts and economies of manufacture, will become more apparent upon consideration of the following description and the appended claims with reference to the accompanying drawings. The current specification, therefore, is conceived in a manner that others skilled in the art will be able to purchase and practice without undue effort.

[0058] The flowcharts used in this specification show the operations of system implementations according to some embodiments in this specification. It should be clearly understood that the operations of the flowcharts can not be implemented in sequence. Instead, the operations can be implemented in reverse order or simultaneously. In addition, one or more other operations can be added to the flowchart. One or more operations can be removed from the flowchart.

[0059] Figure 1 The flowchart of the plant modeling automatic pruning method based on laser radar proposed by the present application is shown in Figure 1. The plant modeling automatic pruning method based on laser radar proposed by the present application specifically includes the following steps:

[0060] Step S100. Obtain plant modeling data of the plant to be pruned and generate three-dimensional point cloud data Cloud by running the laser radar on the trolley p .

[0061] In this embodiment, before the running trolley moves the pruning device to the front of the pruning plant, the running trolley walks around the plant for one round, the laser radar collects 360° plant shape data, and through coupling with inertial navigation data and satellite positioning data, a complete plant shape three-dimensional point cloud Cloud p is generated. Specifically, the following steps are included:

[0062] Step S101. Real-time collection of point cloud of the plant to be pruned by laser radar on the running trolley, filtering out invalid point cloud, and generating point cloud of plant area according to satellite positioning and priori information of plant position.

[0063] In this embodiment, the priori space position of the plant is set as c k =[X k , Y k , Z k ] T , assuming that the current collection is the kth plant object; the laser radar point cloud at the current frame t time is Cloud p_t , through coordinate conversion, the three-dimensional point coordinates in Cloud p_t are converted into satellite latitude and longitude coordinates to obtain Cloud pGPS_t , and it is judged whether the space three-dimensional points in Cloud pGPS_t belong to the space c k, if belongs to, then keep, otherwise filter out, generating the filtered plant region point cloud Cloud pGPS_t_F ;

[0064] wherein the judgment formula is as follows:

[0065]

[0066] wherein k represents the kth working plant or plant association, Cloud pGPS_t_F (i) represents the ith three-dimensional space point in the filtered point cloud.

[0067] Step S102. Select n collection point positions based on the laser radar field of view and the plant size, and calculate the pose transformation relationship of each collection point position relative to the initial collection point position.

[0068] In this embodiment, the latitude and longitude coordinates of the initial collection point position are set as L(lat, lon), the latitude and longitude coordinates of the ωth collection point position are L ω (lat ω ,lon ω ), and the vehicle pose of the initial collection point position is The vehicle pose of the ωth collection point position is

[0069] According to the formula:

[0070]

[0071] The displacement dis of the ωth collection point position L ω (lat ω ,lon ω ) relative to the initial collection point position L(lat, lon) is calculated, so that the translation matrix T ω is obtained based on the relative angle and displacement of the ωth collection point position in the initial collection point position.

[0072] According to the vehicle pose of the ωth collection point position, the rotation matrix of the vehicle pose of the ωth collection point position relative to the vehicle pose transformation of the initial collection point position is obtained based on the x-axis, y-axis and z-axis. ω ;

[0073] wherein the rotation matrix based on the x-axis is:

[0074]

[0075] The rotation matrix around the y-axis is:

[0076]

[0077] The rotation matrix around the z-axis is:

[0078]

[0079] R ω = R ωx · R ωy · R ωz

[0080] In the above formula, denotes a heading angle, θ denotes a pitch angle, and γ denotes a roll angle.

[0081] The filtered point cloud Cloud pGPS_ t _F is generated after superimposition of the point clouds of different angles of the plant region according to the pose transformation relationship. p The step of generating the three-dimensional point cloud data Cloud

[0082] Cloud p = Cloud pGPS_t_F_P0 + Cloud pGPS_t_F_ p1·[R1 T1]+…+ Cloud pGPS_t_F_pn ·[R n T n ]

[0083] In the formula, Cloud pGPs_t_F_Pn denotes the filtered point cloud Cloud pGPS_t_F obtained at the nth acquisition angle.

[0084] The three-dimensional point cloud data Cloud p is calculated.

[0085] Step S103. The three-dimensional point cloud data Cloud p is generated after superimposition of the point clouds of different angles of the plant region according to the pose transformation relationship.

[0086] In this embodiment, the three-dimensional point cloud data Cloud

[0087] Cloud p = Cloud P0 + Cloud P1 ·[R1 T1]+…+ Cloud Pn ·[R n T n ]

[0088] The three-dimensional point cloud data Cloud p is calculated.

[0089] Step S200. The three-dimensional point cloud data Cloud p is compared with the plant standard model point cloud data Clouds After high-precision adaptation processing, point cloud data of the area to be trimmed is generated cut , and the point cloud data Cloud of the area to be pruned cut After clustering, the area to be pruned is generated. cut .

[0090] In this embodiment, step S200 specifically includes the following contents:

[0091] Step S201. The three-dimensional point cloud data Cloud p After encoding according to the preset encoding format, the standard plant shape point cloud data Cloud s High-precision adaptation is performed through the nearest neighbor iterative algorithm, and after coordinate transformation, it enters the spatial coordinate system with the ground as the XY plane and the z-axis vertically upward.

[0092] The specific encoding method is shown in the following table:

[0093] Serial number Protocol Content Description 1 $ Frame header 2 PGA Protocol name 3 PLANTNO Plant serial number Range: 0~10000 4 Number Point number Range: 0~10000000 5 ddd.mmmmmmm Longitude 6 ddd.mmmmmmm Latitude 8 dd.mmm Height above sea level Range: 0~3; 9 M Unit: meter 10 * Frame trailer 11 HH Check sum

[0094] Step S202: Divide the space of the plant area into cylindrical grids based on a spatial resolution of 3cm×3cm, and calculate the Cloud p and Cloud s The three-dimensional points that fall into the cylindrical grid are recorded as point sets Points p_G_Xi_Yi and Points s_G_Xi_Yi , calculate Points respectively p_G_Xi_Yi The average height of the midpoint d p_G_Xi_Yi and d s_G_Xi_Yi ;

[0095] Step S203. p_G_Xi_Yi -d s_G_Xi_Yi The cylindrical grids with a value greater than or equal to τ are marked as the grids to be pruned. After traversing all the cylindrical grids in the space of the plant area, all the grids to be pruned are obtained, and the standard shape point cloud data of the plant falls into the grids to be pruned, and the point cloud of the area to be pruned is obtained. cut , where τ is a threshold value set to be greater than 0;

[0096] Using the Euclidean distance clustering method, Cloud cut Cluster the point cloud in to generate the area to be pruned Areas cut .

[0097] Step S300: In the area to be trimmed Areas cut Perform pruning path planning to generate an operation trajectory, and execute automatic pruning of the plants to be pruned.

[0098] In the embodiment, the pruning operation of the plants is divided into m operation points according to the pruning space of the pruning manipulator on the running trolley, and the operation points are matched with the to-be-pruned areas Areas cut After the corresponding relationship is established, the to-be-pruned areas belonging to the same operation point are sorted in the order from left to right and from top to bottom; for the same to-be-pruned area, the to-be-pruned area is divided into a plurality of sub-areas according to the operation width; the operation path key points, i.e., the operation starting points and ending points of the sub-areas, are selected, the paths between the key points are planned through the hybrid A* algorithm, the operation sub-trajectories of the sub-areas are generated, and the pruning is realized; and the operation trajectory is generated based on the following coding mode:

[0099] [operation point 1: trajectory 1, …, trajectory n1;

[0100] operation point 2: trajectory 1, …, trajectory n2;

[0101] …

[0102] operation point n: trajectory 1, …, trajectory nn]

[0103] The running trolley performs automatic pruning on the to-be-pruned plants according to the operation trajectory.

[0104] As Figure 2 shown, the application further provides an automatic pruning system based on the automatic pruning method of the first aspect, which comprises a running trolley, a pruning device and a back-end data processing platform, the pruning device is located on the running trolley and comprises a laser radar and two pruning manipulators, and the running trolley further comprises a control module for controlling the running of the running trolley and the pruning device and for data interaction with the back-end data processing platform.

[0105] The running trolley walks around the to-be-pruned plants for one round, acquires the plant modeling data of the to-be-pruned plants through the laser radar and generates three-dimensional point cloud data Cloud p ;

[0106] The three-dimensional point cloud data Cloud p is sent to the back-end data processing platform through the TCP / IP protocol and is subjected to high-precision adaptive processing with the plant standard modeling point cloud data Cloud s to generate to-be-pruned area point cloud data Cloud cut , and the to-be-pruned area point cloud data Cloud cut is subjected to clustering processing to generate to-be-pruned areas Areas cut , the to-be-pruned areas Areas cut are subjected to pruning path planning to generate the operation trajectory of the pruning manipulator.

[0107] The operation track is sent to the control module, and the control module controls the pruning manipulator to perform automatic pruning on the plants to be pruned.

[0108] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software plus a necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, and the computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the method described in each embodiment or some parts of the embodiment.

[0109] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A three-dimensional plant modeling and automatic pruning method based on laser radar, characterized in that: The method specifically comprises the following steps: The laser radar on the running car obtains the plant shape data of the plant to be pruned and generates 3D point cloud data ; The three-dimensional point cloud data Point cloud data of standard plant shapes Generate point cloud data of the area to be trimmed after high-precision adaptation processing , and the point cloud data of the area to be trimmed After clustering, the area to be pruned is generated ; In the area to be trimmed Performing pruning path planning to generate an operation trajectory, and executing automatic pruning of the plants to be pruned; The three-dimensional point cloud data Point cloud data of standard plant shapes Generate point cloud data of the area to be trimmed after high-precision adaptation processing , and the point cloud data of the area to be trimmed After clustering, the area to be pruned is generated The steps include: The three-dimensional point cloud data After encoding according to the preset encoding format, the standard plant shape point cloud data High-precision adaptation is performed through the nearest neighbor iterative algorithm, and the coordinate transformation is performed to enter the ground-based flat, In a spatial coordinate system where the axis is vertically upward; The space of the plant area is based on The spatial resolution is used to divide the cylindrical grid and calculate and The three-dimensional points that fall into the cylindrical grid are recorded as point sets and , respectively calculated Average height of midpoint and ; Will The cylindrical grid is marked as the grid to be pruned. After traversing all the cylindrical grids in the space of the plant area, all the grids that need to be pruned are obtained, and the standard shape point cloud data of the plant falls into the grid to be pruned to obtain the point cloud of the area to be pruned. ,in is the threshold value set to be greater than 0; Using the Euclidean distance clustering method, Cluster the point cloud in to generate the area to be pruned .

2. The automatic pruning method for plant shape three-dimensional modeling based on laser radar according to claim 1 is characterized in that: The laser radar on the running car obtains the plant shape data of the plant to be pruned and generates 3D point cloud data The steps specifically include: The laser radar on the running vehicle collects point clouds of the plants to be pruned in real time, and after filtering out invalid point clouds, generates a point cloud of the plant area based on satellite positioning and prior information of the plant position; Selection based on LiDAR field of view and plant size collection points, and calculate the posture transformation relationship of each collection point relative to the initial collection point; According to the posture transformation relationship, the point clouds of the plant areas at different angles are superimposed to generate three-dimensional point cloud data .

3. The automatic pruning method for plant shape three-dimensional modeling based on laser radar according to claim 2 is characterized in that: The steps of collecting point clouds of the plants to be pruned in real time by operating a laser radar on a trolley, filtering out invalid point clouds, and generating point clouds of the plant area according to satellite positioning and prior information of plant positions include: Assume the prior space position of the plant is , assuming that the current collection is the Plant objects; current frame The laser radar point cloud at this moment is , through coordinate transformation, The three-dimensional point coordinates in the image are converted into satellite longitude and latitude coordinates. ,judge Does the three-dimensional point in space belong to space If it belongs to the plant area, it is retained, otherwise it is filtered out to generate the point cloud of the filtered plant area ; The judgment formula is as follows: ； Where, Indicates the a working plant or a combination of plants, Represents the first point in the filtered point cloud A three-dimensional space point.

4. The method for automatic pruning of plant shape three-dimensional modeling based on laser radar according to claim 3, characterized in that: Selection based on LiDAR field of view and plant size The steps of obtaining the acquisition points and calculating the posture transformation relationship of each acquisition point relative to the initial acquisition point include: Set the latitude and longitude coordinates of the initial collection point to , No. The longitude and latitude coordinates of the collection points are , the vehicle posture of the running car at the initial collection point , No. The vehicle posture at each collection point is ; According to the formula: ； Calculate the Collection points Relative to the initial acquisition point Displacement , thus based on the The relative angle and displacement of each acquisition point in the initial acquisition point position are used to obtain the translation matrix ; According to The vehicle posture of each collection point is obtained based on the rotation matrix of the x-axis, y-axis and z-axis. The rotation matrix of the vehicle posture at each collection point relative to the vehicle posture at the initial collection point ; Among them, the rotation matrix based on the x-axis is: ； The rotation matrix around the y-axis is: ； The rotation matrix around the z-axis is: ； ； In the above formula, Indicates the heading angle, represents the pitch angle, Indicates the roll angle.

5. The method for automatic pruning of plant shape three-dimensional modeling based on laser radar according to claim 4 is characterized in that: According to the posture transformation relationship, the point clouds of the plant areas at different angles are superimposed to generate three-dimensional point cloud data The steps include obtaining three-dimensional point cloud data through formula calculation : = ； Where, Represents the filtered point cloud obtained from the nth acquisition angle .

6. The method for automatic pruning of plant shape three-dimensional modeling based on laser radar according to claim 5, characterized in that: In the area to be trimmed The steps of planning a pruning path to generate an operation trajectory and automatically pruning the plants to be pruned include: According to the pruning space of the pruning manipulator on the running trolley, the plant pruning operation is divided into operating points and the area to be trimmed After establishing the corresponding relationship, multiple to-be-pruned areas belonging to the same operation point are sorted from left to right and from top to bottom. For the same to-be-pruned area, it is divided into several sub-areas according to the width of the operation. The key points of the operation path are selected as the operation, and the key points are the operation starting and ending points of the sub-area. Among them, the path between the key points is planned by the hybrid A* algorithm to generate an operation sub-trajectory of a sub-area to achieve traversal pruning. The operation trajectory is generated based on the following encoding method: ； The running vehicle automatically prunes the plants to be pruned according to the operation trajectory.

7. An automatic pruning system based on the automatic pruning method for plant modeling 3D modeling based on laser radar according to claims 1-6, characterized in that: The system includes: a running trolley, a pruning device, and a back-end data processing platform. The pruning device is located on the running trolley and includes a laser radar and two pruning manipulators. The running trolley also includes a control module that controls the operation of the running trolley and the pruning device and exchanges data with the back-end data processing platform. The running car walks around the plant to be pruned, obtains the plant shape data of the plant to be pruned through the laser radar and generates three-dimensional point cloud data ; The three-dimensional point cloud data It is sent to the back-end data processing platform through TCP / IP protocol and compared with the plant standard modeling point cloud data Generate point cloud data of the area to be trimmed after high-precision adaptation processing , and the point cloud data of the area to be trimmed After clustering, the area to be pruned is generated , in the area to be trimmed Perform pruning path planning to generate the operation trajectory of the pruning manipulator; The operation trajectory is sent to the control module, and the control module controls the pruning robot to perform automatic pruning of the plants to be pruned.

Images