Outdoor plant sample information acquisition method based on ground-air cooperation

Abstract

The invention discloses an outdoor plant sample information acquisition method based on ground-air cooperation, and relates to the field plant sample automatic acquisition technology field, a vegetation community map is acquired based on a satellite multispectral image and is sent to an unmanned aerial vehicle base station, and an unmanned aerial vehicle carries a laser radar and an unmanned vehicle to go to a nearest vegetation area; putting the unmanned vehicle after the unmanned aerial vehicle arrives, obtaining a vegetation community elevation map and planning a global route, carrying out air-ground cooperation with the unmanned aerial vehicle to plan a local route while the unmanned vehicle advances along the global route, carrying out species identification by using a visible light camera, planning branch routes according to a matched vegetation location, and carrying out leaf and soil sampling; after arriving at the end point of the global route, the unmanned vehicle sends a return flight request, and the unmanned aerial vehicle hoists the unmanned vehicle to return to the base station and marks a completed area; according to the method, efficient and automatic vegetation sample collection is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of automated collection of wild plant samples, and in particular is a method for acquiring outdoor plant sample information based on ground-air collaboration. Background Art

[0002] Traditional outdoor plant community surveys and sample collection rely primarily on manual fieldwork. This approach has significant drawbacks, including low efficiency, high workload, limited coverage, and difficulty operating in complex or hazardous terrain. Recent developments in drone and autonomous vehicle technology have made automated field surveys possible, but using either drone or autonomous vehicle alone often struggles to balance efficiency and accuracy. Drone surveys can capture macroscopic vegetation distribution, but they cannot precisely identify species or collect physical samples. While autonomous vehicles can perform species identification and sampling, they rely on pre-set map information and lack real-time path planning and obstacle avoidance capabilities in unknown or changing environments.

[0003] How to integrate the drone's macroscopic vision and efficient scanning capabilities with the unmanned vehicle's precise ground operation capabilities to achieve complementary advantages between ground and air and collaborative operations for automated plant sample information acquisition is a question we need to consider. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for acquiring outdoor plant sample information based on ground-air collaboration.

[0005] The purpose of the present invention can be achieved by the following technical solution: a method for obtaining outdoor plant sample information based on ground-air collaboration, comprising the following steps:

[0006] Step S1: using satellite multispectral images to obtain the distribution area of ​​vegetation communities, mapping the vegetation community areas to a scaled map to obtain a vegetation community map, and sending the vegetation community map to the UAV base station;

[0007] Step S2: The UAV base station obtains the vegetation community area closest to the UAV base station based on the obtained vegetation community map. The UAV performs global route planning for the obtained vegetation community area and performs ground-air collaborative planning of local routes with the unmanned vehicle.

[0008] Step S3: The unmanned vehicle travels along the planned global route and local route and captures visible light images, performs species identification based on the obtained visible light images, and samples the matching species;

[0009] Step S4: After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone lifts the unmanned vehicle and returns, and marks the vegetation community area that has completed the survey.

[0010] Furthermore, the process of obtaining the distribution area of ​​vegetation communities using satellite multispectral images includes:

[0011] Acquire a multispectral image of the target area through a satellite, and obtain the reflectance of the near-infrared band and the reflectance of the red band of each pixel in the multispectral image;

[0012] The obtained near-infrared band reflectance NIR and red band reflectance RED are normalized to obtain a normalized vegetation index grayscale image;

[0013] Binarize the normalized vegetation index grayscale image to obtain a binary image. Construct a corresponding proportional map based on the coverage of the target area obtained by the satellite, and mark the location of the drone base station on the proportional map.

[0014] The vegetation community area in the obtained binary image is mapped to a scaled map to obtain a vegetation community map, and the vegetation community map is sent to the UAV base station.

[0015] Furthermore, after obtaining the vegetation community map, the drone sets the location of the drone base station as the take-off point, selects the vegetation community area with the shortest distance from the take-off point, and after the drone arrives at the vegetation community area from the take-off point, it deploys the unmanned vehicle to the ground, using the unmanned vehicle deployment location as the starting point, and plans a global route for the vegetation community area based on the starting point;

[0016] The drone is equipped with a laser radar, which reads the scanning radius of the laser radar carried by the drone. The unmanned vehicle is equipped with a visible light camera, a laser radar and a robotic arm, which reads the length of the unmanned vehicle body, the safe slope range, the scanning radius of the laser radar carried by the unmanned vehicle and the shooting range of the visible light camera.

[0017] Furthermore, the process of planning the global route of the drone includes:

[0018] Construct a plane coordinate system based on the scale map, with the starting point as the origin;

[0019] A route interval is set in the vegetation community area according to the scanning radius of the UAV laser radar, wherein the distance of the route interval is less than or equal to the scanning radius of the UAV laser radar, and a corresponding UAV route is generated in the vegetation community map according to the set route interval, and the UAV elevation data and flight altitude of each position of the UAV on the UAV route are obtained;

[0020] Based on the real-time elevation data and flight altitude of the UAV, the elevation data of each location in the vegetation community area is obtained;

[0021] Mapping the obtained elevation data of each position of the vegetation community to the corresponding position in the constructed plane coordinate system to obtain the vegetation community elevation map;

[0022] According to the shooting range of the visible light camera carried by the unmanned vehicle, the route interval of the unmanned vehicle is set, wherein the distance of the route interval is less than or equal to the shooting range of the visible light camera, the corresponding initial route is generated in the vegetation community map according to the set route interval, the initial route is divided into a plurality of route segments, wherein the length of each route segment is the same as the length of the vehicle body of the unmanned vehicle, the end points of the route segments are marked as reference points and reference points respectively, wherein the reference points are the end points close to the unmanned vehicle, and the reference points are the points away from the unmanned vehicle, and the slope of the route segment is obtained according to the height difference between the reference points and the reference points.

[0023] When the slope of the route segment exceeds the safe slope range:

[0024] If the route segment exceeding the safe slope range is continuous on the initial route, the continuous route segment exceeding the safe slope range is spliced to obtain a dangerous route segment;

[0025] If the route segment exceeding the safe slope range is not continuous, the route segment exceeding the safe slope range is taken as a dangerous route segment alone;

[0026] Further, the process of obtaining the bypass route includes:

[0027] Set the probability P, the angle range ω and the step length L, construct a random tree T with the start point of the dangerous route segment as the root node, obtain a horizontal vector in the vegetation community elevation map pointing from the root node to the end point of the dangerous route segment, generate a random node Q1 within the angle range ω centered on the obtained horizontal vector and with a step length less than or equal to L from the root node, when generating the random node Q1, if the distance between the end point of the dangerous route segment and the root node is less than or equal to the step length L and within the angle range ω, then the end point of the dangerous route segment is directly taken as the random node Q1 with a probability P;

[0028] According to the height difference between the root node and the random node Q1, the slope from the root node to the obtained node Q1 is obtained, when the obtained slope is within the safe slope range, the node Q1 is taken as the parent node, a horizontal vector in the vegetation community elevation map pointing from the parent node to the end point of the dangerous route segment is obtained, a random node Q2 within the angle range ω centered on the obtained horizontal vector and with a step length less than or equal to L from the root node is generated, when generating the random node Q2, if the distance between the end point of the dangerous route segment and the parent node is less than or equal to the step length L and within the angle range ω, then the end point of the dangerous route segment is directly taken as the random node Q2 with a probability P;

[0029] By analogy, the random tree T is expanded towards the direction of the end point of the dangerous route segment, the bypass route is obtained, the initial route corresponding to the dangerous route segment is replaced by the bypass route, and the global route is obtained.

[0030] Further, the process of unmanned vehicle and unmanned aerial vehicle for air-ground collaborative planning of local route includes:

[0031] After completing global route planning, the drone returns to the starting point to conduct ground-air collaborative planning of a local route. While the unmanned vehicle is moving, the drone always flies above it, obtaining the horizontal vector of the unmanned vehicle's forward direction in real time and setting scanning ranges for both the drone and the unmanned vehicle.

[0032] The drone maintains a constant flight altitude and uses the lidar to scan the area within the drone's scanning range to obtain the drone's point cloud;

[0033] The unmanned vehicle uses its own laser radar to scan the area within the unmanned vehicle's scanning range to obtain the unmanned vehicle point cloud;

[0034] Integrate the UAV point cloud and the unmanned vehicle point cloud with the unmanned vehicle as the reference point to obtain a real-time point cloud;

[0035] Set the degree J to divide the real-time point cloud into several sectors. The center angle of each sector is J. Get the vector from the center of the sector to the center point of the arc. Get the standard deviation of the elevation data of the real-time point cloud within the sector. Sort the obtained standard deviations in ascending order. The smaller the standard deviation, the flatter the sector.

[0036] Set thresholds K and M. When the angle between the vector from the center of a sector to the center of the arc and the global route of the unmanned vehicle's current position is less than threshold K and the standard deviation is less than threshold M, obtain the sorting number of the sector that meets the conditions. The unmanned vehicle will prioritize setting the local route to the sector with the highest sorting number that meets the conditions.

[0037] When there is no sector area whose horizontal vector deviation from the center point of the arc to the global route is less than the threshold K and the standard deviation is less than the threshold M, the unmanned vehicle sets the local route to the sector area with the smallest current standard deviation.

[0038] Furthermore, the process of the unmanned vehicle performing real-time species identification and sample collection includes:

[0039] The unmanned vehicle uses a visible light camera to capture visible light images within a 180-degree range in the current direction of travel in real time. The obtained visible light image is histogram-equalized to obtain a preprocessed image. The obtained preprocessed image is binarized to obtain a binary image. Feature extraction is performed based on the obtained binary image. The extracted features are compared with the target vegetation leaf contour and texture feature library, and a similarity threshold is set. When the comparison similarity exceeds the similarity threshold, the corresponding part of the binary image is marked as the vegetation to be sampled.

[0040] According to the direction of the marked vegetation to be sampled, the unmanned vehicle suspends the currently planned route, adds a branch route to the vegetation to be sampled and returns to the current location on the global route, and collaborates with the drone to plan the local route. After arriving next to the vegetation to be sampled, the robotic arm is used to clamp the leaves of the vegetation and the soil at the roots of the vegetation and store them.

[0041] Furthermore, after the unmanned vehicle completes the vegetation community sampling, the process of the unmanned vehicle initiating a return request and returning includes:

[0042] After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone descends, lifts the unmanned vehicle, and automatically returns to the drone base station. After returning, the drone marks the vegetation community area where the survey has been completed on the vegetation community map.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The point clouds obtained by aerial drones and ground unmanned vehicles from scanning the surface in front of the unmanned vehicle at different perspectives are integrated to obtain real-time point clouds, providing more detailed and complete data support for the unmanned vehicle's obstacle avoidance. The unmanned vehicle divides the real-time point cloud into several areas, obtains the standard deviation of the point cloud elevation data in each area, and obtains the optimal route by comparing the standard deviations of different areas, thereby enhancing adaptability to complex terrain while maintaining the overall route direction as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0046] like Figure 1 As shown, the method for obtaining outdoor plant sample information based on ground-air collaboration includes the following steps:

[0047] Step S1: using satellite multispectral images to obtain the distribution area of ​​vegetation communities, mapping the vegetation community areas to a scaled map to obtain a vegetation community map, and sending the vegetation community map to the UAV base station;

[0048] Step S2: The UAV base station obtains the vegetation community area closest to the UAV base station based on the obtained vegetation community map. The UAV performs global route planning for the obtained vegetation community area and performs ground-air collaborative planning of local routes with the unmanned vehicle.

[0049] Step S3: The unmanned vehicle travels along the planned global route and local route and captures visible light images, performs species identification based on the obtained visible light images, and samples the matching species;

[0050] Step S4: After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone lifts the unmanned vehicle and returns, and marks the vegetation community area that has completed the survey.

[0051] It should be further explained that, in the specific implementation process, the process of obtaining the distribution area of ​​vegetation communities using satellite multispectral images includes:

[0052] Calling a satellite equipped with a multispectral sensor to perform multispectral imaging to obtain a multispectral image, and obtaining the reflectance NIR of the near infrared band and the reflectance RED of the red band of each pixel point in the image through the obtained multispectral image;

[0053] The normalized vegetation index (NDVI) of each pixel is obtained based on the reflectance NIR of the near infrared band and the reflectance RED of the red band, where , obtain the normalized vegetation index grayscale image;

[0054] Binarize the normalized vegetation index grayscale image to obtain a binary image. Construct a corresponding proportional map based on the coverage of the target area obtained by the satellite, and mark the location of the drone base station on the proportional map.

[0055] The vegetation community area in the obtained binary image is mapped to a scaled map to obtain a vegetation community map, and the vegetation community map is sent to the UAV base station.

[0056] It should be further explained that in the specific implementation process, after the drone obtains the vegetation community map, it sets the location of the drone base station as the take-off point, selects the vegetation community area with the shortest distance from the take-off point, and after the drone arrives at the vegetation community area from the take-off point, it deploys the unmanned vehicle to the ground, using the unmanned vehicle deployment location as the starting point, and plans a global route for the vegetation community area based on the starting point;

[0057] The drone is equipped with a laser radar, which reads the scanning radius of the laser radar carried by the drone. The unmanned vehicle is equipped with a visible light camera, a laser radar and a robotic arm, which reads the length of the unmanned vehicle body, the safe slope range, the scanning radius of the laser radar carried by the unmanned vehicle and the shooting range of the visible light camera.

[0058] It should be further explained that, in the specific implementation process, the process of the drone planning the global route includes:

[0059] Construct a plane coordinate system based on the scale map, with the starting point as the origin;

[0060] According to the scanning radius of the unmanned aerial vehicle laser radar, a route interval is set in the vegetation community area, wherein the distance of the route interval is less than or equal to the scanning radius of the unmanned aerial vehicle laser radar, a corresponding unmanned aerial vehicle route is generated in the vegetation community map according to the set route interval, and unmanned aerial vehicle elevation data and flight height of each position of the unmanned aerial vehicle in the process of flight on the unmanned aerial vehicle route are obtained.

[0061] According to the real-time elevation data and flight height of the unmanned aerial vehicle, the elevation data of each position of the vegetation community area is obtained.

[0062] The obtained elevation data of each position of the vegetation community is mapped to the corresponding position in the constructed plane coordinate system to obtain a vegetation community elevation map.

[0063] According to the shooting range of the visible light camera carried by the unmanned vehicle, a route interval of the unmanned vehicle is set, wherein the distance of the route interval is less than or equal to the shooting range of the visible light camera, a corresponding initial route is generated in the vegetation community map according to the set route interval, the initial route is divided into a plurality of road segments, wherein the length of each road segment is the same as the length of the vehicle body of the unmanned vehicle, the endpoints of the road segments are marked as reference points and reference points respectively, wherein the reference points are the endpoints close to the unmanned vehicle, and the reference points are the points away from the unmanned vehicle, and the slope of the road segment is obtained according to the elevation difference between the reference points and the reference points.

[0064] It needs to be further explained that in the specific implementation process, when the slope of a road segment exceeds the safe slope range:

[0065] If the road segment exceeding the safe slope range is continuous on the initial route, the continuous road segment exceeding the safe slope range is spliced to obtain a dangerous road segment;

[0066] If the road segment exceeding the safe slope range is not continuous, the road segment exceeding the safe slope range is taken as a dangerous road segment alone;

[0067] It needs to be further explained that in the specific implementation process, the process of obtaining the bypass route includes:

[0068] Set the probability P, the angle range ω and the step length L, construct a random tree T with the starting point of the dangerous road segment as the root node, obtain a horizontal vector in the vegetation community elevation map pointing from the root node to the end point of the dangerous road segment, generate a random node Q1 within the angle range ω centered on the obtained horizontal vector and having a step length less than or equal to L from the root node, and when generating the random node Q1, if the distance between the end point of the dangerous road segment and the root node is less than or equal to the step length L and is within the angle range w, the end point of the dangerous road segment is directly taken as the random node Q1 with a probability P.

[0069] The slope from the root node to the obtained node Q1 is obtained based on the elevation difference between the root node and the random node Q1. When the obtained slope is within the safe slope range, the Q1 node is used as the parent node. The horizontal vector pointing from the parent node to the end point of the dangerous section in the vegetation community elevation map is obtained. A random node Q2 is generated within the angle range ω centered on the obtained horizontal vector direction, with a step length less than or equal to L from the root node. When generating random point Q2, if the distance between the end point of the dangerous section and the root node is less than or equal to the step length L and is within the angle range w, there is a probability P that the end point of the dangerous section is directly used as the random node Q2.

[0070] In this way, the random tree T is expanded toward the end point of the dangerous section to obtain a detour route, and the corresponding dangerous section in the initial route is replaced by the detour route to obtain a global route.

[0071] It should be further explained that in the specific implementation process, after the drone completes the global route planning, it returns to the starting point to carry out ground-air collaborative planning of the local route. While the unmanned vehicle is moving, the drone always flies above the unmanned vehicle, obtains the horizontal vector of the unmanned vehicle's forward direction in real time, and sets the scanning range for the drone and the unmanned vehicle respectively;

[0072] The specific scanning range of the drone is: a sector-shaped area with the unmanned vehicle as the center, the horizontal vector of the unmanned vehicle's forward direction as the midline, a radius of R and a central angle of 180 degrees;

[0073] The scanning range of the unmanned vehicle is specifically: a sector-shaped area with itself as the center, the horizontal vector of its own forward direction as the midline, a radius of R and a central angle of 180 degrees;

[0074] The drone maintains a constant flight altitude and uses the lidar to scan the area within the drone's scanning range to obtain the drone's point cloud;

[0075] The unmanned vehicle uses its own laser radar to scan the area within the unmanned vehicle's scanning range to obtain the unmanned vehicle point cloud;

[0076] Integrate the drone point cloud and the unmanned vehicle point cloud with the unmanned vehicle to obtain a real-time point cloud.

[0077] It should be further explained that in the specific real-time process, the process of the unmanned vehicle planning a local route based on the obtained real-time point cloud includes:

[0078] Set the degree J to divide the real-time point cloud into several sectors. The center angle of each sector is J. Get the vector from the center of the sector to the center point of the arc. Get the standard deviation of the elevation data of the real-time point cloud within the sector. Sort the obtained standard deviations in ascending order. The smaller the standard deviation, the flatter the sector.

[0079] Set thresholds K and M. When the angle between the vector from the center of a sector to the center of the arc and the global route of the unmanned vehicle's current position is less than threshold K and the standard deviation is less than threshold M, obtain the sorting number of the sector that meets the conditions. The unmanned vehicle will prioritize setting the local route to the sector with the highest sorting number that meets the conditions.

[0080] When there is no sector area whose horizontal vector deviation from the center point of the arc to the global route is less than the threshold K and the standard deviation is less than the threshold M, the unmanned vehicle sets the local route to the sector area with the smallest current standard deviation.

[0081] It should be further explained that, in the specific implementation process, the process of unmanned vehicles performing real-time species identification and sample collection includes:

[0082] The unmanned vehicle uses a visible light camera to capture a visible light image within a 180-degree range centered on the current direction of travel in real time. The obtained visible light image is histogram-equalized to obtain a preprocessed image. The obtained preprocessed image is binarized to obtain a binarized image. Feature extraction is performed based on the obtained binarized image. The extracted features are compared with the target vegetation leaf contour and texture feature library, and a similarity threshold is set. When the comparison similarity exceeds the similarity threshold, the corresponding part of the binarized image is marked as the vegetation to be sampled.

[0083] Based on the direction of the marked vegetation to be sampled, the unmanned vehicle pauses its currently planned route and adds a branch route to the vegetation to be sampled and returns to the current location. It then collaborates with the drone to plan a local route. After arriving at the vegetation to be sampled, the robot arm grabs the leaves and soil from the roots of the vegetation and stores them.

[0084] It should be further explained that, in the specific implementation process, after the unmanned vehicle completes the collection of vegetation community area, the process of the unmanned vehicle initiating a return request and returning includes:

[0085] After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone descends, lifts the unmanned vehicle, and automatically returns to the drone base station. After returning, the drone marks the vegetation community area where the survey has been completed on the vegetation community map.

[0086] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any modification or equivalent replacement of the above embodiments made according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. A method for acquiring outdoor plant sample information based on ground-air collaboration, characterized in that: The following steps are involved: Step S1: using satellite multispectral images to obtain the distribution area of ​​vegetation communities, mapping the vegetation community areas to a scaled map to obtain a vegetation community map, and sending the vegetation community map to the UAV base station; Step S2: The UAV base station obtains the vegetation community area closest to the UAV base station based on the obtained vegetation community map. The UAV performs global route planning for the obtained vegetation community area and performs ground-air collaborative planning of local routes with the unmanned vehicle. Step S3: The unmanned vehicle travels along the planned global route and local route and captures visible light images, performs species identification based on the obtained visible light images, and samples the matching species; Step S4: After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone lifts the unmanned vehicle and returns, and marks the vegetation community area that has completed the survey.

2. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 1 is characterized in that: The process of obtaining vegetation community areas using satellite multispectral images includes: Acquire a multispectral image of the target area through a satellite, and obtain the reflectance of the near-infrared band and the reflectance of the red band of each pixel in the multispectral image; The obtained near-infrared band reflectance NIR and red band reflectance RED are normalized to obtain a normalized vegetation index grayscale image; Binarize the normalized vegetation index grayscale image to obtain a binary image. Construct a corresponding proportional map based on the coverage of the target area obtained by the satellite, and mark the location of the drone base station on the proportional map. The vegetation community area in the obtained binary image is mapped to a scaled map to obtain a vegetation community map, and the vegetation community map is sent to the UAV base station.

3. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 2 is characterized in that: After obtaining the vegetation community map, the drone sets the location of the drone base station as the take-off point and selects the vegetation community area with the shortest distance from the take-off point. After the drone arrives at the vegetation community area from the take-off point, it deploys the unmanned vehicle to the ground, using the unmanned vehicle deployment location as the starting point. Based on the starting point, a global route is planned for the vegetation community area. The drone is equipped with a laser radar, which reads the scanning radius of the laser radar carried by the drone. The unmanned vehicle is equipped with a visible light camera, a laser radar and a robotic arm, which reads the length of the unmanned vehicle body, the safe slope range, the scanning radius of the laser radar carried by the unmanned vehicle and the shooting range of the visible light camera.

4. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 3 is characterized in that: The process of planning a global route for a drone includes: Construct a plane coordinate system based on the scale map, with the starting point as the origin; A route interval is set in the vegetation community area according to the scanning radius of the UAV laser radar, wherein the distance of the route interval is less than or equal to the scanning radius of the UAV laser radar, and a corresponding UAV route is generated in the vegetation community map according to the set route interval, and the UAV elevation data and flight altitude of each position of the UAV on the UAV route are obtained; Based on the real-time elevation data and flight altitude of the UAV, the elevation data of each location in the vegetation community area is obtained; Mapping the obtained elevation data of each position of the vegetation community to the corresponding position in the constructed plane coordinate system to obtain the vegetation community elevation map; The unmanned vehicle route interval is set according to the shooting range of the visible light camera carried by the unmanned vehicle, wherein the distance of the route interval is less than or equal to the shooting range of the visible light camera. The corresponding initial route is generated in the vegetation community map according to the set route interval, and the initial route is divided into several sections, wherein the length of each section is the same as the length of the unmanned vehicle body. The endpoints of the section are marked as benchmark points and reference points respectively, wherein the benchmark point is the endpoint close to the unmanned vehicle, and the reference point is the point far from the unmanned vehicle. The slope of the section is obtained according to the elevation difference between the benchmark point and the reference point; When the slope of a road section exceeds the safe slope range: If there are sections that exceed the safe slope range continuously on the initial route, the sections that exceed the safe slope range continuously are spliced ​​together to obtain the dangerous section; If the road section beyond the safe slope range is discontinuous, the road section beyond the safe slope range shall be treated as a dangerous road section alone; Set the probability P, angle range ω, and step size L, and construct a random tree T with the starting point of the dangerous section as the root node. Obtain a horizontal vector from the vegetation community elevation map with the root node pointing to the end point of the dangerous section. Generate a random node Q1 with a step size less than or equal to L from the root node within the angle range ω centered on the obtained horizontal vector direction. When generating random point Q1, if the distance between the end point of the dangerous section and the root node is less than or equal to the step size L and is within the angle range w, then there is a probability P that the end point of the dangerous section is directly used as the random node Q1. The slope from the root node to the obtained node Q1 is obtained based on the elevation difference between the root node and the random node Q1. When the obtained slope is within the safe slope range, the Q1 node is used as the parent node. The horizontal vector pointing from the parent node to the end point of the dangerous section in the vegetation community elevation map is obtained. A random node Q2 is generated within the angle range ω centered on the obtained horizontal vector direction and with a step length less than or equal to L from the root node. When generating random point Q2, if the distance between the end point of the dangerous section and the parent node is less than or equal to the step length L and is within the angle range w, there is a probability P that the end point of the dangerous section is directly used as the random node Q2. In this way, the random tree T is expanded toward the end point of the dangerous section to obtain a detour route, and the corresponding dangerous section in the initial route is replaced by the detour route to obtain a global route.

5. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 4 is characterized in that: After completing global route planning, the drone returns to the starting point to conduct ground-air collaborative planning of a local route. While the unmanned vehicle is moving, the drone always flies above it, obtaining the horizontal vector of the unmanned vehicle's forward direction in real time and setting scanning ranges for both the drone and the unmanned vehicle. The drone maintains a constant flight altitude and uses the lidar to scan the area within the drone's scanning range to obtain the drone's point cloud; The unmanned vehicle uses its own laser radar to scan the area within the unmanned vehicle's scanning range to obtain the unmanned vehicle point cloud; Integrate the drone point cloud and the unmanned vehicle point cloud with the unmanned vehicle to obtain a real-time point cloud.

6. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 5 is characterized in that: The process of planning a local route for an unmanned vehicle based on the real-time point cloud includes: Set the degree J to divide the real-time point cloud into several sectors. The center angle of each sector is J. Get the vector from the center of the sector to the center point of the arc. Get the standard deviation of the elevation data of the real-time point cloud in the sector. Sort the obtained standard deviations in ascending order. Set thresholds K and M. When the angle between the vector from the center of a sector to the center of the arc and the global route of the unmanned vehicle's current position is less than threshold K and the standard deviation is less than threshold M, obtain the sorting number of the sector that meets the conditions. The unmanned vehicle will prioritize setting the local route to the sector with the highest sorting number that meets the conditions. When there is no sector area whose horizontal vector deviation from the center point of the arc to the global route is less than the threshold K and the standard deviation is less than the threshold M, the unmanned vehicle sets the local route to the sector area with the smallest current standard deviation.

7. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 6 is characterized in that: The process of the unmanned vehicle identifying species and collecting samples in real time includes: The unmanned vehicle uses a visible light camera to capture a visible light image within a 180-degree range centered on the current direction of travel in real time. The obtained visible light image is histogram-equalized to obtain a preprocessed image. The obtained preprocessed image is binarized to obtain a binarized image. Feature extraction is performed based on the obtained binarized image. The extracted features are compared with the target vegetation leaf contour and texture feature library, and a similarity threshold is set. When the comparison similarity exceeds the similarity threshold, the corresponding part of the binarized image is marked as the vegetation to be sampled. According to the direction of the marked vegetation to be sampled, the unmanned vehicle suspends the currently planned route, adds a branch route to the vegetation to be sampled and returns to the current location on the global route, and collaborates with the drone to plan the local route. After arriving next to the vegetation to be sampled, the robotic arm is used to clamp the leaves of the vegetation and the soil at the roots of the vegetation and store them.

8. The method for obtaining outdoor plant sample information based on ground-air collaboration according to claim 7 is characterized in that: After the unmanned vehicle completes the vegetation community sampling, the process of initiating a return request and returning includes: After the unmanned vehicle reaches the end of the global route, it sends a return request to the drone. After receiving the return request, the drone descends, lifts the unmanned vehicle, and automatically returns to the drone base station. After returning, the drone marks the vegetation community area where the survey has been completed on the vegetation community map.

Images