Adaptive mowing control method for mowing robot

Abstract

An adaptive mowing control method for a mowing robot. By means of settings, a real-time operation environment image is divided into a near-environment image and a far-environment image, such that the computing power consumed when processing the real-time operation environment image can be reduced. The method comprises: on the basis of an operation traveling speed of a mowing robot, performing block planning on a real-time operation environment image by taking an effective photographing distance as a division parameter, so as to obtain a near-environment image and a far-environment image, wherein the near-environment image is an image which is centered on the mowing robot and extends outwards without exceeding a quarter of the area of the real-time operation environment image, the far-environment image is the real-time operation environment image other than the near-environment image, the highest operation traveling speed is directly proportional to an area proportion of the near-environment image in the real-time environment image, the area proportion of the near-environment image in the real-time environment image does not exceed 25%, and the area proportion of the near-environment image is correlated with an external illumination intensity.

Description

Adaptive mowing control method of mowing robot TECHNICAL FIELD

[0001] The present application belongs to the technical field of mowing robots, in particular to an adaptive mowing control method of a mowing robot. BACKGROUND

[0002] With the rapid development of industrial economy, it brings greater challenges to environmental governance. To a large extent, increasing the global green coverage is one of the effective means of current environmental governance. Grassland needs a lot of manpower and material resources in maintenance and maintenance. How to improve the efficiency of green grass maintenance and management and reduce the labor intensity of relevant personnel is a problem that needs to be solved urgently. Manual mowing is a monotonous and time-consuming work, and it also needs to face the threat of lawn slope, obstacles, gravel, and even harmful insects in grass. The emergence of mowing robots solves these problems.

[0003] The current mowing robot mainly relies on traditional means such as contact sensor and wire technology in the obstacle avoidance control system, which is not efficient and easy to damage obstacles. Due to the shortcomings of the sensor itself, the environmental information perceived is not comprehensive and sufficient in the actual working environment, and it is easy to miss and miss. With the rapid development of deep learning in recent years, target detection technology based on deep learning can make the mowing robot more intelligent and effective in identifying obstacles. By detecting the types of obstacles, relevant obstacle avoidance strategies can be developed.

[0004] The existing mowing robot has poor obstacle avoidance efficiency and path control accuracy during work, which leads to waste of a lot of time when performing large-scale mowing tasks, and the robot is easily disturbed by external light environment when avoiding obstacles, resulting in low obstacle recognition accuracy. Therefore, we propose an adaptive mowing control method of a mowing robot. SUMMARY

[0005] The purpose of the present application is to provide an adaptive mowing control method of a mowing robot, which solves the problem of poor obstacle avoidance efficiency and path control accuracy of the existing mowing robot during work, which leads to waste of a lot of time when performing large-scale mowing tasks, and the robot is easily disturbed by external light environment when avoiding obstacles, resulting in low obstacle recognition accuracy.

[0006] To solve the above technical problems, the present application is realized by the following technical scheme:

[0007] The present application is an adaptive mowing control method of a mowing robot, comprising the following steps:

[0008] S1, acquire the real-time working environment image and path motion information of the mowing robot;

[0009] S2, adaptively control the effective shooting distance of the real-time working environment image of the mowing robot according to the current working external light environment;

[0010] S3, divide the real-time working environment image into a near environment image and a far environment image according to the working driving speed of the mowing robot and the effective shooting distance as a division parameter, wherein the near environment image is an image with a radiation area not more than one fourth of the real-time working environment image area and centered on the mowing robot, the far environment image is all the real-time working environment image except the near environment image, the working driving speed is proportional to the area ratio of the near environment image in the real-time environment image, and the area ratio of the near environment image in the real-time environment image is not more than 25%;

[0011] S4, process and analyze the real-time working environment image by using image processing technology to identify and judge the obstacles;

[0012] S5, when the real-time working environment image is identified to include obstacles, the type of the obstacles is obtained, and the obstacle avoidance range corresponding to the type of the obstacles is obtained;

[0013] S6, control the mowing robot to actively avoid obstacles according to the obstacle avoidance range.

[0014] Preferably, the real-time working environment image and the motion information of the mowing robot are obtained by a camera and a position sensor, the camera is used to collect the front fan-shaped image area of the mowing robot during working, and the collected image information is sent to a trained SVM model to identify and classify the image, so as to finally identify the type of the front obstacles of the mowing robot during working and obtain the bounding box and confidence information; the position sensor is used to collect the real-time position data of the mowing robot, so as to plan the travel path.

[0015] Preferably, the effective shooting distance of the real-time working environment image in step S2 includes the following specific contents: a light-sensitive sensor arranged on the mowing robot is used to collect the front light intensity information during the current mowing working, and compare the current light intensity information with the standard light intensity range to find the standard light intensity information with the same light intensity in the standard light intensity range, and adaptively control the effective shooting distance of the real-time working environment image of the mowing robot according to the distance data corresponding to the standard light intensity information; if the current light intensity information is lower than the lowest critical value of the standard light intensity range, stop obtaining the real-time working environment image of the mowing robot and stop the mowing working thereof; if the current light intensity information is higher than the highest critical value of the standard light intensity range, control the effective shooting distance of the real-time working environment image of the mowing robot according to the distance data corresponding to the highest critical value of the standard light intensity range.

[0016] Preferably, the effective shooting distance is proportional to the front light intensity information data during operation, the higher the front light intensity information data, the farther the effective shooting distance, and vice versa, the lower the front light intensity information data, the closer the effective shooting distance, the working driving speed of the lawn mower is limited by the effective shooting distance, the lawn mower adaptively controls the working driving speed on the basis of completing the mowing standard, the working driving speed is limited by the shooting distance of the real-time working environment image, the farther the effective shooting distance, the faster the maximum working driving speed of the lawn mower, when the effective shooting distance reaches the critical distance, the maximum working driving speed of the lawn mower reaches the fastest.

[0017] Preferably, the lawn mower is internally provided with a control system for controlling the operation of the lawn mower, the adaptive control of the lawn mower is completed by the control system, and the real-time working environment image processing weight distribution is performed before the real-time working environment image is processed and analyzed in step S4, which includes the following contents: according to the proportion of the near environment image area and the far environment image area in the real-time working environment image, the weight distribution is performed, the higher the proportion, the higher the weight, the higher the weight, the more system resources are allocated to the processing of the image, and the sum of the near environment image processing weight and the far environment image processing weight is equal to 1.

[0018] The control system simultaneously processes the near environment image and the far environment image.

[0019] Preferably, the processing and analysis of the real-time working environment image by using the image processing technology includes the following contents:

[0020] S4.1, the real-time working environment image is split into a plurality of network blocks, and the texture features of each network block are extracted respectively;

[0021] S4.2, each network block is classified and distinguished by using the trained SVM model;

[0022] S4.3, according to the classification of the network block, the image is segmented to obtain a binary image, the real-time working environment image presents obvious black and white effect, and the identification of the obstacle is realized.

[0023] Preferably, in step S4.1, the image features are extracted based on the texture features, the problem feature extraction method of the gray level co-occurrence matrix is adopted, according to two gray levels i and j, the directions of the two gray levels in the image plane are θ and d respectively, and the gray level co-occurrence matrix of a real-time working environment image with a gray value of L is set as

[0024] ,​​​

[0025] wherein, represents the number of elements x in the set, .

[0026] Preferably, the greater the number of gray levels, the more texture features are obtained by using the gray level co-occurrence matrix to calculate the texture features, and therefore the image gray level is degraded, and L and d of the gray level co-occurrence matrix are determined by establishing an experimental model, including the following steps:

[0027] 1) Select several lawn images and images with obstacles, randomly select 10 sample image blocks from them, and obtain sample image libraries with gray levels L of 8, 16, and 32;

[0028] 2) Extract effective texture features from the sample image library, and provide 4x4 problem features from 0°, 45°, 90°, and 135°;

[0029] 3) Set the number of gray levels to 8, 16, and 32 in turn, and increase the step size d between [0, 15] to change, calculate the average value of the lawn and obstacle texture features under different combinations of L and d, and calculate the difference between the two; based on different gray levels and statistical step sizes, calculate the difference between the average values of the lawn and obstacle texture features;

[0030] 4) Statistically analyze the curve of the average difference between the lawn and obstacle texture features, and select the parameter combination with the largest difference between the two;

[0031] 5) Through statistical calculation, the number of gray levels L and the step size d are selected as 16 and 1, respectively.

[0032] Preferably, according to the parameter combination, a data set of lawn and obstacle texture features is made, and a support vector machine SVM model is trained, 16-dimensional data is obtained by using 0°, 45°, 90°, 135°, and 16 combinations of energy, entropy, contrast, and inverse difference matrix, and the data is normalized to the interval, and a plurality of training samples and a plurality of test samples are obtained; a plurality of training samples are used to classify and train the SVM model.

[0033] Preferably, the obstacle avoidance range of the mowing robot exceeds the obstacle range by at least 10%, and when the mowing robot follows an initial path and detects that there is an obstacle on the initial path, it is determined whether it can drive in the effective area corresponding to the initial path under the condition of bypassing the obstacle; the effective path corresponding to the initial path is an area formed by expanding outward within a range of 180° to the left and right of the initial path in front of the mowing robot before the mowing robot moves;

[0034] If it is judged that it can be bypassed, the lawn mowing robot is controlled to bypass the obstacle and travel in the effective area corresponding to the initial path;

[0035] If it is judged that it cannot be bypassed, the path is re-planned, and the lawn mowing robot is controlled to bypass the obstacle and travel on the new path.

[0036] The present application has the following beneficial effects:

[0037] 1、The present application controls the effective shooting distance of the real-time working environment image of the lawn mowing robot according to the light intensity of the real-time working environment of the lawn mowing robot, when the environment light is more intense, the effective shooting distance of the real-time working environment image is farther, and vice versa, when the environment light is dimmer, the effective shooting distance of the real-time working environment image is closer, the more intense the environment light represents the clearer the shooting of the real-time working environment image, the control system is more identifiable when identifying obstacles, so the effective shooting distance can be appropriately increased, the dimmer the environment light represents the more blurred the shooting of the real-time working environment image, the control system is more difficult to identify obstacles, and more system power is needed to identify, so the effective shooting distance needs to be appropriately reduced, the amount of effective shooting distance can be directly reflected in the shooting content of the real-time working environment image, the less the shooting content of the real-time working environment image, the easier it is to process, and the more the shooting content of the real-time working environment image, the more difficult it is to process, the present application alleviates the power consumption of the lawn mowing robot when it is adaptively controlled, and improves the working safety of the lawn mowing robot.

[0038] 2、The present application binds the working travel speed of the lawn mowing robot with the effective shooting distance, and adaptively controls the working travel speed on the basis of completing the lawn mowing standard when the lawn mowing robot performs the lawn mowing work, when the effective shooting distance is farther, the area of the near environment image accounts for a higher proportion, and the highest working travel speed is faster, when the effective shooting distance is closer, the area of the near environment image accounts for a lower proportion, and the highest working travel speed is lower, so that the working efficiency of the lawn mowing robot is improved while the safety of the work is ensured.

[0039] 3、The present application divides the real-time working environment image into a near environment image and a far environment image, which can reduce the power consumption when processing the real-time working environment image. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description, and obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating laborious work.

[0041] Fig. 1 is a flowchart of the adaptive lawn mowing control method of the lawn mowing robot provided by the present application. DETAILED DESCRIPTION

[0042] In order to make the technical means, creative features, purposes and effects of the present application easy to understand, the present application is further described below in conjunction with specific embodiments.

[0043] In the description of the present application, it should be noted that the terms "upper", "lower", "inner", "outer", "front end", "rear end", "two ends", "one end", "the other end" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first" and "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0044] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "provided with", "connection" and the like should be broadly understood, for example, "connection" can be fixed connection, can also be detachable connection, or integrally connected; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0045] Referring to FIG. 1, the adaptive mowing control method of the mowing robot comprises the following steps:

[0046] S1, acquiring real-time working environment images and movement information of the mowing robot;

[0047] S2, adaptively controlling the effective shooting distance of the real-time working environment images of the mowing robot according to the current working external light environment;

[0048] S3, dividing the real-time working environment images into block planning for near environment images and far environment images according to the working driving speed of the mowing robot and the effective shooting distance as the division parameter, wherein the near environment images are images radiating outward from the mowing robot as the center not more than one fourth of the area of the real-time working environment images, and the far environment images are all real-time working environment images except the near environment images, and the working driving speed is directly proportional to the area ratio of the near environment images in the real-time environment images, and the area ratio of the near environment images in the real-time environment images is not more than 25%;

[0049] S4, processing and analyzing the real-time working environment images by using image processing technology to identify and judge obstacles;

[0050] S5, when it is identified that the real-time working environment image includes an obstacle, the type of the obstacle is acquired, and a type corresponding obstacle avoidance range of the obstacle is acquired;

[0051] S6, the mowing robot is controlled to actively avoid obstacles according to the obstacle avoidance range.

[0052] The real-time working environment image of the mowing robot and the traveling motion information are acquired by a camera and a position sensor. The camera is used to collect a front fan-shaped image area of the mowing robot during working, and send the collected image information to a trained SVM model to identify and classify the image, so as to finally identify the type of the front obstacle of the mowing robot during working and obtain the bounding box and the confidence information. The position sensor is used to collect real-time position data of the mowing robot, so as to plan the traveling path.

[0053] The effective shooting distance of the real-time working environment image in step S2 includes the following specific contents: a photosensitive sensor arranged on the mowing robot is used to collect front light intensity information during current mowing working, and compare the current light intensity information with a standard light intensity range, find standard light intensity information with the same light intensity in the standard light intensity range, and adaptively control the effective shooting distance of the real-time working environment image of the mowing robot according to distance data corresponding to the standard light intensity information. If the current light intensity information is lower than the lowest critical value of the standard light intensity range, the acquisition of the real-time working environment image of the mowing robot and the mowing working of the mowing robot are stopped. If the current light intensity information is higher than the highest critical value of the standard light intensity range, the effective shooting distance of the real-time working environment image of the mowing robot is controlled according to distance data corresponding to the highest critical value of the standard light intensity range.

[0054] The effective shooting distance is proportional to the front light intensity information data during working. The higher the current light intensity information data is, the farther the effective shooting distance is. Conversely, the lower the current light intensity information data is, the closer the effective shooting distance is. The working traveling speed of the mowing robot is limited by the effective shooting distance of the mowing robot. The mowing robot adaptively controls the working traveling speed on the basis of completing the mowing standard during mowing working. The working traveling speed is limited by the shooting distance of the real-time working environment image. The farther the effective shooting distance is, the faster the highest working traveling speed of the mowing robot is. When the effective shooting distance reaches the critical distance, the highest working traveling speed of the mowing robot reaches the fastest.

[0055] The mowing robot is internally provided with a control system for controlling the operation of the mowing robot, and the adaptive control of the mowing robot is completed through the control system. The real-time working environment image is processed and weighted before being processed and analyzed in step S4, which includes the following contents: the weight is allocated according to the proportion of the near environment image area and the far environment image area in the real-time working environment image, the higher the proportion, the higher the weight, the higher the weight, the more system resources are allocated to the processing of the image, and the sum of the near environment image processing weight and the far environment image processing weight is equal to 1.

[0056] The control system simultaneously processes the near environment image and the far environment image.

[0057] The real-time working environment image is processed and analyzed by using image processing technology, including the following contents:

[0058] S4.1, the real-time working environment image is split into multiple network blocks, and the texture features of each network block are extracted respectively;

[0059] S4.2, each network block is classified and distinguished by using the trained SVM model;

[0060] S4.3, according to the classification of the network block, the image is segmented to obtain the binary image, the real-time working environment image presents obvious black and white effect, and the identification of the obstacle is realized.

[0061] In step S4.1, the image features are extracted based on the texture features, and the problem feature extraction method of the gray level co-occurrence matrix is adopted, according to two gray levels i and j, the directions of the two in the image plane are θ and d, and the gray level co-occurrence matrix is set as A real-time working environment image, whose gray value is L, satisfies its gray level co-occurrence matrix The elements of the gray level co-occurrence matrix are shown as formula (1):

[0062] ,

[0063] Wherein, Indicates the number of x elements in the set, .

[0064] ​The gray level co-occurrence matrix cannot directly reflect the texture features of the image to be classified and recognized, but can be expressed by its second order statistics. The gray level co-occurrence matrix includes more than ten kinds of statistics such as energy ASM, entropy ENT, contrast CON, inverse difference moment IDM, uniformity, correlation, variance, average, variance, entropy, difference variance, difference average, difference entropy, correlation information measure and maximum correlation coefficient. Among them, the energy ASM generally represents the uniformity of the gray distribution and the thickness of the texture; the entropy ENT can be used to represent the complexity of the image texture, and the value is larger when the randomness of the elements contained is higher. The contrast CON represents the depth of the image texture, and the inverse difference moment IDM represents the degree of local change of the texture. The four kinds of statistics are used to analyze and process the features of the image to be recognized, and the calculation formula is:

[0065] ,

[0066] ,

[0067] ,

[0068] .

[0069] When the gray level co-occurrence matrix is used to calculate the texture features, the more the gray levels are, the more the texture features obtained by statistics are, so the image gray is degraded, and the L and d of the gray level co-occurrence matrix are determined by establishing an experimental model, including the following steps:

[0070] 1) Select several lawn images and images with obstacles, randomly select 10 sample image blocks from them, and obtain sample image libraries with gray levels L of 8, 16 and 32;

[0071] 2) Extract effective texture features from the sample image library, and provide 4x4 problem features from 0°, 45°, 90° and 135°;

[0072] 3) Set the gray levels to be 8, 16 and 32 in turn, and the step d increases between [0, 15] in increments, statistics and calculation of the average values of the lawn and obstacle texture features under different combinations of L and d, and the difference between the two are calculated; based on different gray levels and statistical steps, the difference between the average values of the lawn and obstacle texture features is calculated;

[0073] 4) Statistically calculate the curve of the average difference of the lawn and obstacle texture features, and select the parameter combination with large distinction between the two;

[0074] 5) Through statistical calculation, the gray level L and the step d are selected to be 16 and 1 respectively.

[0075] The gray level co-occurrence matrix is very sensitive to the direction of image texture, and strict requirements are required for the direction angle during the calculation of the gray level co-occurrence matrix, and the present application selects 0°, 45°, 90° and 135° for scanning calculation; according to the parameter combination, the data set of the texture characteristics of the lawn and the obstacle is made, the support vector machine (SVM) model is trained, 0°, 45°, 90°, 135° and 16 combinations composed of energy, entropy, contrast and inverse difference matrix are used to form 16-dimensional data on the lawn and obstacle image, and the data is normalized to the interval, a plurality of training samples and a plurality of test samples are obtained; the SVM model is classified and trained by using a plurality of training samples.

[0076] The mowing robot obstacle avoidance range exceeds the obstacle range by at least 10%, when the mowing robot follows the initial path and detects that there is an obstacle on the initial path, it is determined whether it can bypass the obstacle and drive in the effective area corresponding to the initial path, and the effective path corresponding to the initial path is the area formed by expanding outward within the range of 180° left and right in front of the mowing robot before the initial path;

[0077] If it is determined that it can bypass, the mowing robot is controlled to bypass the obstacle and drive in the effective area corresponding to the initial path;

[0078] If it is determined that it cannot bypass, the path is re-planned, and the mowing robot is controlled to bypass the obstacle and drive on the new path.

[0079] The working principle of the present application is: by controlling the effective shooting distance of the real-time working environment image according to the real-time working environment light intensity of the mowing robot, when the environment light is stronger, the effective shooting distance of the real-time working environment image is farther, and vice versa, when the environment light is dimmer, the effective shooting distance of the real-time working environment image is closer; the stronger the environment light represents the clearer the shooting of the real-time working environment image, and the control system is more identifiable when identifying obstacles, so the effective shooting distance can be appropriately increased, the dimmer the environment light represents the more blurred the shooting of the real-time working environment image, and the control system is more difficult to identify, which requires longer time and more system computing power to identify, so the effective shooting distance needs to be appropriately reduced; the amount of effective shooting distance can be directly reflected in the shooting content of the real-time working environment image, the less the shooting content of the real-time working environment image, the easier it is to process, and the more the shooting content of the real-time working environment image, the more difficult it is to process; in this way, the computing power consumption of the adaptive control room of the mowing robot is relieved, and the working safety of the mowing robot is improved.

[0080] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive mowing control of a mowing robot, characterized in that, The method comprises the following steps: S1, acquiring real-time working environment images of the mowing robot and motion information; S2, adaptively controlling the effective shooting distance of the real-time working environment images of the mowing robot according to the external light environment at the current working time; the effective shooting distance of the real-time working environment images in step S2 comprises the following specific contents: a light-sensitive sensor arranged on the mowing robot is used to collect the front light intensity information at the current mowing working time, and the current light intensity information is compared with a standard light intensity range to find standard light intensity information with the same light intensity in the standard light intensity range, and the effective shooting distance of the real-time working environment images of the mowing robot is adaptively controlled according to the distance data corresponding to the standard light intensity information; if the current light intensity information is lower than the lowest critical value of the standard light intensity range, the acquisition of the real-time working environment images of the mowing robot is stopped and the mowing work of the mowing robot is also stopped; if the current light intensity information is higher than the highest critical value of the standard light intensity range, the effective shooting distance of the real-time working environment images of the mowing robot is controlled according to the distance data corresponding to the highest critical value of the standard light intensity range; S3, dividing the real-time working environment images into block planning for near environment images and far environment images according to the working driving speed of the mowing robot and the effective shooting distance as the division parameter, wherein the near environment images are images with an area not more than one fourth of the real-time working environment images radiating outward from the center of the mowing robot, the far environment images are all the real-time working environment images except the near environment images, the maximum working driving speed is proportional to the area ratio of the near environment images in the real-time environment images, and the area ratio of the near environment images in the real-time environment images is not more than 25%, the area ratio of the near environment images is related to the external light intensity, and the higher the light intensity, the larger the area ratio of the near environment images; S4, processing and analyzing the real-time working environment images by using image processing technology to identify and judge obstacles; S5, when obstacles are identified in the real-time working environment images, the type of the obstacles is acquired, and the obstacle avoidance range corresponding to the type of the obstacles is also acquired; S6, controlling the mowing robot to actively avoid obstacles according to the obstacle avoidance range; The acquisition of the real-time working environment images of the mowing robot and the motion information is completed by a camera and a position sensor, the camera is used to collect the front fan-shaped image area of the mowing robot during working, and the collected image information is sent to a trained SVM model to identify and classify the images, so as to finally identify the type of the obstacles in front of the mowing robot during working and obtain the bounding box and confidence information; the position sensor is used to collect the real-time position data of the mowing robot, so as to plan the driving path. The effective shooting distance is proportional to the front light intensity information data during operation, the higher the front light intensity information data, the farther the effective shooting distance, and vice versa, the lower the front light intensity information data, the closer the effective shooting distance, the operation driving speed of the lawn mower is limited by the effective shooting distance, the lawn mower is self-adaptive to control the operation driving speed based on the completion of the mowing standard, the operation driving speed is limited by the shooting distance of the real-time operation environment image, the farther the effective shooting distance, the faster the maximum operation driving speed of the lawn mower, when the effective shooting distance reaches the critical distance, the maximum operation driving speed of the lawn mower reaches the fastest; The control system is arranged in the lawn mower to control the operation of the lawn mower, the self-adaptive control of the lawn mower is completed by the control system, the real-time operation environment image processing weight distribution is performed before the real-time operation environment image is processed and analyzed in step S4, which includes the following contents: the weight distribution is performed according to the proportion of the near environment image area and the far environment image area in the real-time operation environment image, the higher the proportion, the higher the weight, the higher the weight, the more system resources are allocated to the image processing, and the sum of the near environment image processing weight and the far environment image processing weight is equal to 1. The control system processes the near environment image and the far environment image at the same time.

2. The adaptive mowing control method of a mowing robot according to claim 1, characterized in that, The real-time operation environment image is processed and analyzed by using the image processing technology, which includes the following contents: S4.1, the real-time operation environment image is divided into multiple network blocks, and the texture features of each network block are extracted respectively; S4.2, each network block is classified and judged by using the trained SVM model; S4.3, the image is segmented according to the classification of the network block, and the binary image is obtained, the real-time operation environment image presents obvious black and white effect, and the identification of the obstacle is realized. 3.The adaptive mowing control method of the mowing robot according to claim 2, wherein, The image feature is extracted based on the texture feature in the step S4.1, a problem feature extraction method of a gray level co-occurrence matrix is adopted, and a gray level co-occurrence matrix is set according to two gray levels i and j, the directions of the two gray levels in the image plane are θ and d respectively, and the gray level co-occurrence matrix is represented as . A real-time working environment image, the gray value of which is L, satisfies the gray level co-occurrence matrix The element of the gray level co-occurrence matrix is shown as formula (1): , wherein represents the solution of the number of elements x in the set, 。 4.The adaptive mowing control method of the mowing robot according to claim 3, wherein, When the texture features are calculated by using the gray level co-occurrence matrix, the more the texture features are obtained by the larger gray level, therefore, the image gray level is degraded, L and d of the gray level co-occurrence matrix are determined by establishing an experimental model, which includes the following steps: 1) select several lawn images and images with obstacles, randomly select 10 sample image blocks from them, and obtain sample image library with gray level L of 8, 16 and 32; 2) extract effective texture features from the sample image library, and provide 4×4 problem features from 0°, 45°, 90° and 135°; 3) set the gray level to be 8, 16 and 32 in turn, and the step d increases between [0, 15], the average values of the lawn and obstacle texture features under different combinations of L and d are calculated, and the difference between the two is calculated; based on different gray levels and statistical steps, the difference between the average values of the lawn and obstacle texture features is calculated; 4) the curve of the average difference of the lawn and obstacle texture features is calculated, and the parameter combination with large difference is selected; 5) through statistical calculation, the gray level L and the step d are selected to be 16 and 1 respectively.

5. The adaptive mowing control method of the mowing robot according to claim 4, characterized in that, According to the parameter combination, a data set of lawn and obstacle texture features is made, a support vector machine (SVM) model is trained, 16-dimensional data is obtained by using 0°, 45°, 90°, 135° and 16 combinations of energy, entropy, contrast and inverse difference moment on the lawn and obstacle images, and the data is normalized to an interval, a plurality of training samples and a plurality of test samples are obtained; and the SVM model is classified and trained by using the plurality of training samples. 6.The adaptive mowing control method of the mowing robot according to claim 5, wherein, The obstacle avoidance range of the mowing robot is at least 10% greater than the obstacle range, and when the mowing robot travels along an initial path and detects that there is an obstacle on the initial path, it is determined whether the mowing robot can travel in the effective area corresponding to the initial path under the condition of bypassing the obstacle, and the effective path corresponding to the initial path is an area formed by expanding outward within a range of 180° left and right in front of the mowing robot before traveling; If it is determined that the obstacle can be bypassed, the mowing robot is controlled to bypass the obstacle and travel in the effective area corresponding to the initial path; If it is determined that the obstacle cannot be bypassed, the path is re-planned, and the mowing robot is controlled to bypass the obstacle and travel on the new path.

Images