Real-time path correction method for general operation power platform of hill and mountain machinery

Abstract

The invention provides a real-time path correction method for a general operation power platform of hill and mountain machinery, and relates to the technical field of path planning, a camera and an LIDAR sensor are installed in front of the top of the platform, images and distance data are collected in real time, and denoising and edge detection are carried out on the images to identify obstacle boundaries. Then, feature matching is carried out in camera images at different positions by using an SIFT algorithm, dense point cloud is reconstructed, and a three-dimensional terrain model is generated; on the basis of the secondary recognition image and the three-dimensional terrain model, path planning is conducted through an A * algorithm, path deviation is determined, a control instruction is generated to adjust the steering angle, and real-time path correction is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a real-time path correction method for a universal operating power platform of hilly and mountainous machinery. Background Art

[0002] In complex terrain environments such as hilly and mountainous areas, traditional mechanical power platforms have difficulty achieving efficient and accurate path planning and navigation due to the complexity and irregularity of the terrain. Conventional path planning technologies mostly rely on pre-mapped maps or simple sensor data. These methods are powerless under dynamic and ever-changing terrain conditions, and can easily lead to problems such as path misjudgment, detours or collisions. In addition, existing path correction methods can usually only handle path deviations on a two-dimensional plane and cannot meet the navigation requirements of three-dimensional complex terrain, especially when the path needs to be adjusted in real time. It is difficult to respond quickly and accurately. Therefore, it is particularly important to develop a mechanical general-purpose operating power platform that can dynamically correct the path based on real-time data.

[0003] In the prior art, publication number CN114399748A discloses a real-time path correction method for agricultural machinery based on visual lane detection. On the basis of the existing feature-based lane line detection method, a fitted lane line, i.e., the vehicle's scheduled driving route, is obtained through a short-distance lateral strip window sliding method, and the lateral offset distance and yaw angle of the agricultural machinery during driving are calculated through binocular vision, thereby correcting the driving path of the agricultural machinery.

[0004] The main problems with the above method are: it relies solely on cameras for data collection. In actual environments, lane lines may be blocked by obstacles such as soil and vegetation, resulting in unstable detection results; and it mainly focuses on detecting lane lines and calculating path offsets, and has limited perception and adaptability to three-dimensional terrain. It may not be able to accurately calculate and process terrain features such as slope changes and potholes, thus affecting detection accuracy.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a real-time path correction method for a universal working power platform of hilly and mountainous machinery to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A real-time path correction method for a universal operating power platform for hilly mountainous machinery, the specific steps comprising:

[0009] Step 1: Install cameras in front of the top of the power platform and other locations that can detect the same area, covering a path at least 50 meters in front of the power platform, and install a LIDAR sensor in front of the top of the power platform;

[0010] Step 2: Collect real-time images and distance data through the camera and LIDAR sensor, scale the image to 224×224 size, denoise the scaled image through a Gaussian filter, convert the denoised image into a grayscale image, generate a first recognition image, and use the Canny edge detection algorithm to extract edge points from the first recognition image to generate a second recognition image that can reflect the boundary position of the obstacle;

[0011] Step 3: Based on the images of the same area collected by the cameras installed at all positions on the power platform, the images are sorted in the order of the camera positions, the feature descriptors are detected in each image by the SIFT algorithm, the features of adjacent images are matched, the matched feature points are projected into the three-dimensional space based on triangulation, and a dense point cloud is reconstructed by the multi-view stereo matching algorithm, and the dense point cloud data is converted into an ordered three-dimensional terrain model;

[0012] Step 4: According to actual requirements, determine the starting point and target point of the power platform, determine the path nodes that the power platform can pass through based on the second recognition image, perform path planning on the three-dimensional terrain model using the A* algorithm, and generate path deviations according to the positions of the planned path nodes and target points;

[0013] Step 5: Generate control instructions based on the path deviation and adjust the steering angle for direction control.

[0014] Furthermore, the principle for converting the denoised image into a grayscale image is:

[0015] H=0.299·R+0.587·G+0.114·B

[0016] Among them, H represents the grayscale value of the pixel, R represents the red channel value, G represents the green channel value, and B represents the blue channel value.

[0017] Furthermore, the principle on which the edge points are extracted from the first recognition image using the canny edge detection algorithm to generate the second recognition image is:

[0018] For each pixel in the first recognition image, the matrix consisting of the pixel and its neighboring pixels is convolved with the horizontal template and vertical template of the Prewitt operator to generate the grayscale difference of the pixel in the horizontal and vertical directions. The formula is:

[0019]

[0020] Among them, P X represents the horizontal template of the Prewitt operator, P Y represents the vertical template of the Prewitt operator, G x Represents the horizontal difference of the pixel, G y Represents the vertical difference of the pixel point, (x, y) represents the plane coordinates of the pixel point;

[0021] The gradient amplitude of each pixel is generated based on the grayscale difference in the horizontal and vertical directions. The formula is:

[0022]

[0023] Where G(x,y) represents the gradient amplitude of the pixel with coordinates (x,y);

[0024] The edge threshold is preset. When the gradient amplitude of a pixel point is higher than the edge threshold, the pixel point is retained and marked as an edge pixel point, otherwise the pixel point is discarded.

[0025] Furthermore, the principle of reconstructing and generating dense point clouds through multi-view stereo matching algorithm is based on:

[0026] The depth value of the pixel is generated according to the feature matching points of the image, based on the formula:

[0027]

[0028] Among them, d i represents the depth value of the i-th pixel, i is the index of the pixel, E represents the baseline distance, f represents the focal length, x L and x R Respectively represent the horizontal coordinates of the matching points in the left and right images adjacent to the image where the pixel point is located;

[0029] The coordinates of the pixel points in the image are converted into homogeneous coordinates, and the homogeneous coordinates of the pixel points are converted into three-dimensional coordinates according to the intrinsic parameter matrix of the camera and the depth value of the pixel points. The formula is:

[0030]

[0031] Among them, K represents the internal parameter matrix, f x represents the focal length in the x direction, f y represents the focal length in the y direction, c x Indicates the central horizontal coordinate of the image, c y Indicates the central ordinate of the image, (X i ,Y i ,Z i ) represents the three-dimensional coordinates of the i-th pixel, xi ,y i Respectively represent the horizontal and vertical coordinates of the i-th pixel in the image;

[0032] The three-dimensional coordinates of the pixel points and the color information of the pixel points are stitched together to generate point cloud data. The formula is as follows:

[0033] D i =(X i ,Y i ,Z i ,R i ,G i ,B i )

[0034] Among them, D i Represents the point cloud data of the i-th pixel, R i , G i , B i Respectively represent the red channel value, green channel value, and blue channel value of the i-th pixel.

[0035] Furthermore, the principle of path planning by A* algorithm is:

[0036] The comprehensive evaluation function is generated based on the actual cost and estimated cost of the path, and the formula is as follows:

[0037]

[0038] m(n)=g(n)+h(n)

[0039] Among them, m(n) represents the comprehensive evaluation function of the nth node of the path, g(n) represents the actual cost of the nth node, h(n) represents the estimated cost of the nth node, and x n ,y n 、z n They represent the x, y, and z coordinates of the nth node in the three-dimensional space. x0, y0, and z0 represent the x, y, and z coordinates of the starting point in the three-dimensional space. goal ,y goal 、z goal Respectively represent the coordinates of the target in the x, y, and z directions of the three-dimensional space;

[0040] The comprehensive evaluation function value is calculated for each node, and the path where the node with the minimum comprehensive evaluation function value is located is selected as the planned path.

[0041] Furthermore, the principle for generating path deviation is:

[0042]

[0043] Among them, e represents the path deviation, x min ,y min 、z min They respectively represent the coordinates of the node with the smallest comprehensive evaluation function value in the x, y, and z directions of the three-dimensional space.

[0044] Furthermore, the principle for adjusting the steering angle is:

[0045]

[0046] Among them, θ(t) represents the steering angle adjusted at time t, k p represents the proportional gain, e(t) represents the path deviation at time t, k d Represents the differential gain.

[0047] The DAR sensor combination collects data, obtains image information and distance information at the same time, accurately identifies the boundaries of obstacles, provides clear path environment information, reduces misjudgment and enhances environmental perception capabilities, laying the foundation for subsequent three-dimensional model reconstruction; using multiple cameras to collect images from different positions can obtain more comprehensive stereo information, and through multi-view fusion and high-precision feature matching, generate detailed and accurate three-dimensional models, improving the accuracy and reliability of path planning.

[0048] The present invention also performs path planning on the basis of the three-dimensional model through the A* algorithm. By determining path nodes and evaluating path deviations, the path planning can be continuously optimized, and the navigation accuracy and execution efficiency of the power platform in the actual environment can be improved. Accurate path planning and real-time path deviation evaluation are provided, so that the power platform can perform tasks more reliably, reduce navigation errors and potential risks caused by improper path planning, and generate control instructions according to the path deviation to control the direction of travel of the power platform, thereby realizing efficient and automated navigation and path tracking, and improving the autonomy and intelligence level of the power platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The figure is a schematic diagram of the method flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0051] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0052] Example:

[0053] See also Figure 1 , the present invention provides a technical solution:

[0054] A real-time path correction method for a universal operating power platform for hilly mountainous machinery, the specific steps comprising:

[0055] Step 1: Install cameras in front of the top of the power platform and other locations that can detect the same area, covering a path at least 50 meters in front of the power platform, and install a LIDAR sensor in front of the top of the power platform;

[0056] In this embodiment, a camera is installed at the front position on the top of the power platform in order to provide sufficient front environment information for path planning. At other positions where the same area can be detected, it means that the same area can be covered from different angles to establish a three-dimensional model that can reflect sufficient terrain features.

[0057] Step 2: Collect real-time images and distance data through the camera and LIDAR sensor, scale the image to 224×224 size, denoise the scaled image through a Gaussian filter, convert the denoised image into a grayscale image, generate a first recognition image, and use the Canny edge detection algorithm to extract edge points from the first recognition image to generate a second recognition image that can reflect the boundary position of the obstacle;

[0058] In this embodiment, the principle on which the first recognition image uses the canny edge detection algorithm to extract edge points to generate the second recognition image is:

[0059] For each pixel in the first recognition image, the matrix consisting of the pixel and its neighboring pixels is convolved with the horizontal template and vertical template of the Prewitt operator to generate the grayscale difference of the pixel in the horizontal and vertical directions. The formula is:

[0060]

[0061]

[0062] Among them, P X represents the horizontal template of the Prewitt operator, P Y represents the vertical template of the Prewitt operator, G x Represents the horizontal difference of the pixel, G y Represents the vertical difference of the pixel point, (x, y) represents the plane coordinates of the pixel point;

[0063] The gradient amplitude of each pixel is generated based on the grayscale difference in the horizontal and vertical directions. The formula is:

[0064]

[0065] Where G(x,y) represents the gradient amplitude of the pixel with coordinates (x,y);

[0066] The edge threshold is preset. When the gradient amplitude of a pixel point is higher than the edge threshold, the pixel point is retained and marked as an edge pixel point, otherwise the pixel point is discarded.

[0067] The principle of presetting the edge threshold is: combining the gradient amplitudes of all pixels to generate a histogram of the gradient amplitude, and selecting the initial edge threshold within the range of 70% to 90% of the histogram. A higher percentile helps to suppress noise and reduce false alarms, but may miss subtle edges. A lower percentile can capture more details, but may cause too much noise to be identified as edges. Therefore, after selecting the initial edge threshold, the edge pixels of the first recognition image are extracted, and experts in the field judge the extraction effect of the edge pixels, and adjust the edge threshold according to the extraction effect of the edge pixels until the edge extracted according to the edge threshold meets the requirement, which is to be able to identify the boundaries of environmental obstacles.

[0068] Step 3: Based on the images of the same area collected by the cameras installed at all positions on the power platform, the images are sorted in the order of the camera positions, the feature descriptors are detected in each image by the SIFT algorithm, the features of adjacent images are matched, the matched feature points are projected into the three-dimensional space based on triangulation, and a dense point cloud is reconstructed by the multi-view stereo matching algorithm, and the dense point cloud data is converted into an ordered three-dimensional terrain model;

[0069] In this embodiment, the principle for performing feature matching on images is:

[0070] The collected images of the same area are converted into grayscale images, and Gaussian blur processing is performed on the grayscale images to generate Gaussian blurred images. The difference of the Gaussian blurred images is calculated to generate Gaussian difference pyramids. Feature points are detected through the DoG pyramid. Local extreme points are found from the DoG pyramid as key points, that is, local maximum and local minimum feature points in the scale space and spatial domain. A direction is assigned to each key point so that it remains unchanged under rotation changes. A local image block is constructed around each key point, and the direction histogram is calculated. A 128-dimensional vector is generated according to the direction histogram as the feature descriptor of the image. The Euclidean distance between the feature descriptors of the two images is calculated by the nearest neighbor algorithm. The formula based on it is:

[0071]

[0072] in, Represent the feature descriptor vectors of the two images respectively, represents the Euclidean distance between feature descriptor vectors, J represents the dimension of the feature descriptor vector, j represents the index of the dimension of the feature descriptor vector, x j Represented in the feature descriptor vector The value of the jth dimension in , y j Indicates that the descriptor vector The value of the jth dimension in ;

[0073] For two images, the Euclidean distance of each set of feature descriptors is calculated, and the feature descriptor with the smallest distance is recorded as the nearest neighbor, and the feature descriptor with the second smallest distance is recorded as the second nearest neighbor. When the ratio of the nearest neighbor and the second nearest neighbor of a feature descriptor is less than the descriptor feature threshold, the matching of the feature descriptor is valid. The formula is:

[0074]

[0075] Among them, L1 represents the nearest neighbor, L2 represents the next nearest neighbor, d0 represents the descriptor feature threshold, usually L0 = 0.8. A lower threshold can reduce false matches, but will lead to prejudgment of the correct matching points. L0 = 0.8 is a selection result based on experiments and experience.

[0076] In this embodiment, the principle of reconstructing and generating a dense point cloud through a multi-view stereo matching algorithm is based on:

[0077] The depth value of the pixel is generated according to the feature matching points of the image, based on the formula:

[0078]

[0079] Among them, d i represents the depth value of the i-th pixel, i is the index of the pixel, E represents the baseline distance, f represents the focal length, x L and x R Respectively represent the horizontal coordinates of the matching points in the left and right images adjacent to the image where the pixel point is located;

[0080] The baseline distance represents the fixed distance between the acquisition cameras of the two images to be matched, x L -x R It reflects the difference in the x-coordinates of the same feature point in the left and right images, that is, the disparity. The larger the disparity, the closer the feature point is to the camera, and the smaller the disparity, the farther the feature point is from the camera. The depth value of the pixel point is proportional to the baseline distance and focal length, and inversely proportional to the disparity of the feature point.

[0081] The coordinates of the pixel points in the image are converted into homogeneous coordinates, and the homogeneous coordinates of the pixel points are converted into three-dimensional coordinates according to the intrinsic parameter matrix of the camera and the depth value of the pixel points. The formula is:

[0082]

[0083] Among them, K represents the internal parameter matrix, f x represents the focal length in the x direction, f y represents the focal length in the y direction, c x Indicates the central horizontal coordinate of the image, c y Indicates the central ordinate of the image, (X i ,Y i ,Z i ) represents the three-dimensional coordinates of the i-th pixel, x i ,y i Respectively represent the horizontal and vertical coordinates of the i-th pixel in the image;

[0084] The intrinsic parameter matrix reflects the internal parameters of the camera that collects the image, including the focal length and the image center coordinates. The coordinates of the pixel points in the image are converted into homogeneous coordinates so that coordinate transformation operations can be performed with the intrinsic parameter matrix.

[0085] The three-dimensional coordinates of the pixel points and the color information of the pixel points are stitched together to generate point cloud data. The formula is as follows:

[0086] D i =(X i ,Y i ,Z i ,R i ,G i ,B i )

[0087] Among them, D i Represents the point cloud data of the i-th pixel, R i , G i , B i Respectively represent the red channel value, green channel value, and blue channel value of the i-th pixel.

[0088] The purpose of generating a dense point cloud is to create a three-dimensional model that contains the three-dimensional coordinates and color information of each point in the scene. The point cloud data should be able to describe the shape and position of the object on the one hand, and reflect the color information of the object's surface on the other hand. The three-dimensional coordinates and color information of the pixel points are combined to generate dense point cloud data of the corresponding pixel points.

[0089] Step 4: According to actual requirements, determine the starting point and target point of the power platform, determine the path nodes that the power platform can pass through based on the second recognition image, perform path planning on the three-dimensional terrain model using the A* algorithm, and generate path deviations according to the positions of the planned path nodes and the target points;

[0090] In this embodiment, the principle of path planning by the A* algorithm is:

[0091] The comprehensive evaluation function is generated based on the actual cost and estimated cost of the path, and the formula is as follows:

[0092]

[0093] m(n)=g(n)+h(n)

[0094] Among them, m(n) represents the comprehensive evaluation function of the nth node of the path, g(n) represents the actual cost of the nth node, h(n) represents the estimated cost of the nth node, and x n ,y n 、z n They represent the x, y, and z coordinates of the nth node in the three-dimensional space. x0, y0, and z0 represent the x, y, and z coordinates of the starting point in the three-dimensional space. goal ,y goal 、z goal Respectively represent the coordinates of the target in the x, y, and z directions of the three-dimensional space;

[0095] The A* algorithm finds the optimal path from the starting point to the target point by combining the actual cost and the estimated cost. The actual cost represents the cumulative cost from the starting point to the current node, reflects the completed part of the path, and is a known and certain value. The estimated cost represents the predicted cost from the current node to the target point, reflects the unfinished part of the path. The comprehensive evaluation function combines the actual cost and the estimated cost to evaluate the overall cost of the path, balances the known part and the unfinished part to find the optimal path. The comprehensive evaluation function is proportional to the actual cost and the estimated cost of the path.

[0096] The comprehensive evaluation function value is calculated for each node, and the path where the node with the minimum comprehensive evaluation function value is located is selected as the planned path.

[0097] The principle on which the path deviation is generated is:

[0098]

[0099] Among them, e represents the path deviation, x min ,y min 、z min They respectively represent the coordinates of the node with the smallest comprehensive evaluation function value in the x, y, and z directions of the three-dimensional space.

[0100] The path deviation indicates the distance between the actual path of the power platform and the target path, reflecting the degree to which the trajectory of the power platform deviates from the predetermined trajectory under the current driving state. The path deviation of the power platform is determined by the Euclidean distance between the node with the smallest path cost and the target point. The path deviation is proportional to the Euclidean distance between the node with the smallest path cost and the target point.

[0101] Step 5: Generate control instructions based on the path deviation and adjust the steering angle for direction control.

[0102] In this embodiment, the principle for adjusting the steering angle is:

[0103]

[0104] Among them, θ(t) represents the steering angle adjusted at time t, k p represents the proportional gain, e(t) represents the path deviation at time t, k d Represents the differential gain.

[0105] Direction control is performed based on the PD controller. Steering control instructions are generated according to the path deviation and its rate of change. The steering angle is the instruction for controlling the direction of the power platform. The path deviation indicates the vertical distance between the actual driving path and the expected path. The larger the path deviation, the larger the steering angle that needs to be adjusted. It reflects the rate of change of path deviation and indicates the speed at which path deviation changes over time.p Represents the proportional gain, which controls the effect of the current path deviation on the steering angle. A higher proportional gain makes the system more sensitive to path deviation and responds faster, but may cause oscillation; k d Represents the differential gain, which controls the effect of the path deviation change rate on the steering angle. A higher differential gain can suppress the rapid change of path deviation and reduce system oscillation.

[0106] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0107] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0108] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0109] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A real-time path correction method for a universal operating power platform of hilly mountainous machinery, characterized in that: The specific steps include: Step 1: Install cameras in front of the top of the power platform and other locations that can detect the same area, covering a path at least 50 meters in front of the power platform, and install a LIDAR sensor in front of the top of the power platform; Step 2: Collect real-time images and distance data through the camera and LIDAR sensor, scale the image to 224×224 size, denoise the scaled image through a Gaussian filter, convert the denoised image into a grayscale image, generate a first recognition image, extract edge points from the first recognition image using the Canny edge detection algorithm, and generate a second recognition image that can reflect the boundary position of the obstacle; Step 3: Based on the images of the same area collected by the cameras installed at all positions on the power platform, the images are sorted in the order of the camera positions, the feature descriptors are detected in each image by the SIFT algorithm, the features of adjacent images are matched, the matched feature points are projected into the three-dimensional space based on triangulation, and a dense point cloud is reconstructed by the multi-view stereo matching algorithm, and the dense point cloud data is converted into an ordered three-dimensional terrain model; Step 4: According to actual requirements, determine the starting point and target point of the power platform, determine the path nodes that the power platform can pass through based on the second recognition image, perform path planning on the three-dimensional terrain model using the A* algorithm, and generate path deviations according to the positions of the planned path nodes and target points; Step 5: Generate control instructions based on the path deviation and adjust the steering angle for direction control.

2. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 1 is characterized by: The principle for converting the denoised image into a grayscale image in step 2 is: H=0.299·R+0.587·G+0.114·B Among them, H represents the grayscale value of the pixel, R represents the red channel value, G represents the green channel value, and B represents the blue channel value.

3. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 1 is characterized by: The principle of using the canny edge detection algorithm to extract edge points from the first recognition image to generate the second recognition image in step 2 is: For each pixel in the first recognition image, the matrix consisting of the pixel and its neighboring pixels is convolved with the horizontal template and vertical template of the Prewitt operator to generate the grayscale difference of the pixel in the horizontal and vertical directions. The formula is: Among them, P X represents the horizontal template of the Prewitt operator, P Y represents the vertical template of the Prewitt operator, G x represents the horizontal difference of the pixel, Gy represents the vertical difference of the pixel, and (x, y) represents the plane coordinates of the pixel; The gradient amplitude of each pixel is generated based on the grayscale difference in the horizontal and vertical directions. The formula is: Where G(x, y) represents the gradient amplitude of the pixel with coordinates (x, y); The edge threshold is preset. When the gradient amplitude of a pixel point is higher than the edge threshold, the pixel point is retained and marked as an edge pixel point, otherwise the pixel point is discarded.

4. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 1 is characterized in that: The principle of reconstructing and generating a dense point cloud by the multi-view stereo matching algorithm in step 3 is: The depth value of the pixel is generated according to the feature matching points of the image, based on the formula: Among them, d i represents the depth value of the i-th pixel, i is the index of the pixel, E represents the baseline distance, f represents the focal length, x L and x R Respectively represent the horizontal coordinates of the matching points in the left and right images adjacent to the image where the pixel point is located; The coordinates of the pixel points in the image are converted into homogeneous coordinates, and the homogeneous coordinates of the pixel points are converted into three-dimensional coordinates according to the intrinsic parameter matrix of the camera and the depth value of the pixel points. The formula is: Among them, K represents the internal parameter matrix, f x represents the focal length in the x direction, f y represents the focal length in the y direction, c x Indicates the central horizontal coordinate of the image, c y Indicates the central ordinate of the image, (X i , Y i , Z i ) represents the three-dimensional coordinates of the i-th pixel, x i ,y i Respectively represent the horizontal and vertical coordinates of the i-th pixel in the image; The three-dimensional coordinates of the pixel points and the color information of the pixel points are spliced ​​to generate point cloud data. The formula is as follows: D i =(X i ,Y i ,Z i ,R i ,G i ,B i ) Among them, D i Represents the point cloud data of the i-th pixel, R i , G i , B i Respectively represent the red channel value, green channel value, and blue channel value of the i-th pixel.

5. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 1 is characterized by: The principle of path planning by the A* algorithm in step 4 is: The comprehensive evaluation function is generated based on the actual cost and estimated cost of the path, and the formula is as follows: m(n)=g(n)+h(n) Among them, m(n) represents the comprehensive evaluation function of the nth node of the path, g(n) represents the actual cost of the nth node, h(n) represents the estimated cost of the nth node, and x n ,y n 、z n They represent the x, y, and z coordinates of the nth node in the three-dimensional space. x0, y0, and z0 represent the x, y, and z coordinates of the starting point in the three-dimensional space. goal ,y goal 、z goal Respectively represent the coordinates of the target in the x, v, and z directions of the three-dimensional space; The comprehensive evaluation function value is calculated for each node, and the path where the node with the minimum comprehensive evaluation function value is located is selected as the planned path.

6. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 5 is characterized by: The principle on which the path deviation is generated is: Among them, e represents the path deviation, x min ,y min 、z min They respectively represent the coordinates of the node with the smallest comprehensive evaluation function value in the x, y, and z directions of the three-dimensional space.

7. The real-time path correction method for a universal working power platform of hilly mountainous machinery according to claim 1 is characterized by: The principle for adjusting the steering angle in step 5 is: Among them, θ(t) represents the steering angle adjusted at time t, k p represents the proportional gain, e(t) represents the path deviation at time t, k d Represents the differential gain.

Images