Row control tracking control method for row-crossing operation of crawler-type agricultural machine

Abstract

The invention discloses a row-control tracking control method for row-crossing operation of a crawler-type agricultural machine, and relates to the technical field of tracking control of the crawler-type agricultural machine, and the method comprises the steps: installing a plurality of high-precision sensors on the crawler-type agricultural machine, and collecting the position, posture and speed of the agricultural machine, and the information data of the surrounding environment of the agricultural machine; a multi-sensor data fusion technology is utilized to obtain accurate position, attitude, speed and environment information of the agricultural machine, and purification data is formed; extracting state features from the purified data by adopting a feature elimination method; constructing a path deviation prediction model by adopting a deep reinforcement learning DQN algorithm based on historical data; generating a target path by using map service, inputting the key agricultural machine state characteristics into the path deviation prediction model, and outputting a difference value; according to the difference value, the steering angle and the speed adjustment amount of the agricultural machine are calculated; a control strategy is given according to the steering angle and the speed adjustment amount of the agricultural machine, and the control strategy is executed through a control unit.

Description

Technical Field

[0001] The invention relates to the technical field of tracking control of crawler-type agricultural machinery, in particular to a row-by-row tracking control method for a crawler-type agricultural machinery in cross-row operation. Background Art

[0002] Tracked agricultural machinery tracking control refers to the technology that enables tracked agricultural machinery to automatically and accurately travel along a predetermined path through advanced sensors, data processing and control algorithms during agricultural operations. The main goal of this technology is to improve the accuracy, efficiency and automation level of agricultural operations and reduce the errors and labor intensity of human operations. The traditional cross-row operation tracking control method for crawler agricultural machinery uses a single sensor in use, and the data of a single sensor is easily disturbed by the environment, resulting in inaccurate data. At the same time, it is difficult for traditional methods to extract key features from a large amount of data, and it is easy to miss important information. The feature extraction method is fixed and cannot adapt to different operating environments and conditions. Secondly, the traditional path deviation prediction method relies on a fixed model, which is difficult to adapt to complex operating environments, lacks learning of historical data, and has low prediction accuracy. Therefore, how to use advanced technical means to improve the intelligence level and safety of the cross-row operation tracking control method for crawler agricultural machinery has become one of the current problems that need to be solved urgently. Summary of the invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a row-by-row tracking control method for crawler agricultural machinery operating across rows to solve the problem that traditional row-by-row tracking control methods for crawler agricultural machinery operating across rows use a single sensor during use, and the data of the single sensor is easily affected by environmental interference, resulting in inaccurate data; at the same time, traditional methods have difficulty extracting key features from large amounts of data and are prone to missing important information.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a method for tracking and controlling cross-row operations of a crawler agricultural machinery, which includes installing a variety of high-precision sensors on the crawler agricultural machinery to collect the position, posture, speed and environmental information data of the agricultural machinery; using multi-sensor data fusion technology in combination with a Kalman filter to remove noise and outliers in the collected data to obtain the precise position, posture, speed and environmental information of the agricultural machinery to form purified data; using a feature elimination method to extract key agricultural machinery position information, posture information, speed information and state characteristics of the agricultural machinery's surrounding environment information from the purified data; using a deep reinforcement learning DQN algorithm based on historical data to construct a path deviation prediction model; using a map service to generate a target path during the agricultural machinery operation process, inputting key agricultural machinery state characteristics into the path deviation prediction model, and outputting a difference value through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path; based on the difference value, introducing a PID control algorithm and an adaptive control algorithm to calculate the steering angle and speed adjustment amount of the agricultural machinery; giving a control strategy based on the steering angle and speed adjustment amount of the agricultural machinery, sending the control strategy to the control unit of the agricultural machinery through a communication interface, and the control unit executing the control strategy.

[0006] As a preferred solution of the crawler agricultural machinery cross-row operation row tracking control method of the present invention, wherein: a plurality of high-precision sensors are installed on the crawler agricultural machinery to collect the position, posture, speed and surrounding environment information data of the agricultural machinery, and the specific steps are as follows: Install a GPS receiver, an IMU inertial measurement unit, an RGB-D camera, and a LiDA laser radar on the crawler agricultural machinery; start the sensors to collect the position, orientation angle, forward tilt angle, side tilt angle, speed, color images of the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data.

[0007] As a preferred solution of the crawler-type agricultural machinery cross-row operation row tracking control method of the present invention, wherein: the multi-sensor data fusion technology is combined with the Kalman filter to remove the noise and abnormal values ​​in the collected data to obtain the precise position, posture, speed and environmental information of the agricultural machinery. The specific steps are: The data from different sensors are fused using the weighted average method to form a comprehensive data set, which is expressed as: ; in, For a comprehensive dataset, is the number of sensors installed on agricultural machinery, is the index variable, For the The weight of each sensor, For the The data collected by each sensor; the extended Kalman filter is used to remove noise and abnormal values ​​in the comprehensive data set to obtain the precise longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data to form purified data .

[0008] As a preferred solution of the crawler-type agricultural machinery cross-row operation row tracking control method of the present invention, wherein: the state characteristics of key agricultural machinery position information, posture information, speed information and agricultural machinery surrounding environment information are extracted from the purified data by using the return feature elimination method, and the specific steps are as follows: The recursive feature elimination method is used to extract the longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color image around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and state features of three-dimensional point cloud data of key agricultural machinery from the purified data; the support vector machine is used as the base model to evaluate the importance score of each state feature; according to the importance score of each state feature, the weight of each state feature is dynamically adjusted; the state feature extraction function is set , according to the characteristics of agricultural machinery in different operation stages, the feature extraction function is adaptively adjusted, and the expression is: ; in, is the extracted state feature, It is The extraction function of state features, It is to purify the data. is the number of state features extracted.

[0009] As a preferred solution of the crawler-type agricultural machinery cross-row operation row tracking control method of the present invention, wherein: the path deviation prediction model is constructed based on historical data using the deep reinforcement learning DQN algorithm, and the specific steps include: Select the DQN algorithm in deep reinforcement learning as the basic model of the path deviation prediction model; set the state space and action space Provide decision options for the DQN model, expressed as: ; ; in, For agricultural machinery at time step of Axis horizontal axis and Axis ordinate, For agricultural machinery at time step The orientation angle, For agricultural machinery at time step Linear speed, For agricultural machinery at time step information about the surrounding environment, For agricultural machinery at time step The state set, For agricultural machinery at time step The steering angle, For agricultural machinery at time step The speed adjustment amount, For agricultural machinery at time step The action set.

[0010] As a preferred solution of the row tracking control method for cross-row operation of crawler-type agricultural machinery of the present invention, wherein: the target path in the agricultural machinery operation process is generated by using a map service, key agricultural machinery state characteristics are input into the path deviation prediction model, and the difference value is output through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path. The specific steps are as follows: Use the Google Maps API map service to obtain the coordinate points of the target path. The target path is represented as a series of coordinate points. The expression is: ; in, represents the first point on the path, where It's the first point Axis horizontal axis, It's the first point Axis ordinate, represents the second point on the path, where It's the second point Axis horizontal axis, It's the second point Axis ordinate, represents the middle point on the path, represents the last point on the path, where It's the last point. Axis horizontal axis, It's the last point. Axis ordinate, Indicates the target path; obtains the current position of the agricultural machinery through the GPS receiver ; For each point on the target path, calculate the Euclidean distance to the current position of the agricultural machine; calculate the difference value by the Euclidean distance between the current position of the agricultural machine and the nearest point on the target path, the expression is: ; in, is the difference between the current position of the agricultural machinery and the target path, and is the coordinate of the point on the target path closest to the current position of the agricultural machine, It represents the Euclidean distance between the nearest point on the target path and the current position of the agricultural machine.

[0011] As a preferred solution of the row tracking control method for the crawler-type agricultural machinery in the present invention, wherein: according to the difference value, the PID control algorithm and the adaptive control algorithm are introduced to calculate the steering angle and speed adjustment amount of the agricultural machinery, and the specific steps are: The difference value output by the path deviation prediction model As input, the PID control algorithm and adaptive control algorithm are used to calculate the steering angle and speed adjustment of the agricultural machinery. The expression is: ; in, is the steering angle adjustment of the agricultural machinery, is the speed adjustment of the agricultural machinery, is the proportional gain of the PID control algorithm, is the integral gain of the PID control algorithm, is the differential gain of the PID control algorithm, is the proportional gain of the adaptive control algorithm, used to dynamically adjust the speed, is the integral gain of the adaptive control algorithm, is the derivative gain of the adaptive control algorithm, is the difference between the current position of the agricultural machinery and the target path, is the current time, is the control input vector containing the steering angle and speed adjustment, Represents small increments of time.

[0012] As a preferred solution of the row tracking control method for the crawler-type agricultural machinery in the present invention, wherein: the control strategy is given according to the steering angle and speed adjustment amount of the agricultural machinery, and the control strategy is sent to the control unit of the agricultural machinery through the communication interface, and the control unit executes the control strategy. The specific steps are: Based on the calculated steering angle and speed adjustment , integrated into the control input vector , and generate control strategies; control input vectors are transmitted through the communication interface The control unit of the agricultural machinery receives the control strategy and executes the corresponding control instructions, including: normal mode, when the agricultural machinery is operating normally, the steering angle is , the speed adjustment is ; Emergency obstacle avoidance mode, when an obstacle is detected in front, the steering angle is , the speed adjustment is =0, the machine should stop moving forward immediately and adjust the direction to avoid obstacles; in the slow fine adjustment mode, when the machine approaches the target path, the steering angle is , the speed adjustment is When the speed is slow, fine-tune the adjustment.

[0013] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the row tracking control method for cross-row operation of a crawler-type agricultural machinery as described in the first aspect of the present invention is implemented.

[0014] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the row tracking control method for cross-row operation of a crawler-type agricultural machinery as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: by installing a GPS receiver, an IMU inertial measurement unit, an RGB-D camera and a LiDAR laser radar, various state information and surrounding environment information of agricultural machinery can be collected comprehensively and with high precision; data from different sensors are fused by a weighted average method, and an extended Kalman filter is used to remove noise and outliers in the data; the use of data fusion technology can integrate the advantages of different sensors and improve the accuracy and robustness of the data; the Kalman filter can effectively remove noise and outliers and ensure the purity of the data; the importance score of each state feature is evaluated by a recursive feature elimination method and a support vector machine, the most critical information can be screened out from a large amount of data, data redundancy can be reduced, and the training efficiency and prediction accuracy of the model can be improved; the deep reinforcement learning DQN algorithm can learn the optimal path deviation prediction strategy through historical data and improve the accuracy and robustness of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0017] Figure 1 It is a flow chart of the row tracking control method for the crawler-type agricultural machinery in cross-row operation in Example 1; Figure 2 This is a flow chart of the row tracking control method for cross-row operation of a crawler-type agricultural machinery in Example 1. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0021] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for controlling row tracking of a crawler-type agricultural machine during cross-row operation, comprising the following steps: S1. Install a variety of high-precision sensors on crawler agricultural machinery to collect information data on the position, posture, speed and surrounding environment of the agricultural machinery; Furthermore, a variety of high-precision sensors are installed on the crawler agricultural machinery to collect the position, posture, speed and information data of the surrounding environment of the agricultural machinery. The specific steps are: install a GPS receiver, IMU inertial measurement unit, RGB-D camera and LiDA laser radar on the crawler agricultural machinery; start the sensor to collect the position, orientation angle, forward tilt angle, roll angle, speed, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery and three-dimensional point cloud data.

[0022] It should be noted that installing high-precision sensors can not only improve the automation level of agricultural machinery during field operations, but also enhance its adaptability to complex environments. By collecting these detailed information, it can provide a solid foundation for subsequent data processing and analysis.

[0023] S2. Use multi-sensor data fusion technology and Kalman filter to remove noise and outliers in the collected data, obtain the precise position, attitude, speed and environmental information of the agricultural machinery, and form purified data; Furthermore, multi-sensor data fusion technology is used in combination with Kalman filter to remove noise and outliers in the collected data, and the precise position, attitude, speed and environmental information of the agricultural machinery are obtained. The specific steps are to use the weighted average method to fuse the data from different sensors to form a comprehensive data set, which is expressed as: ; in, For a comprehensive dataset, is the number of sensors installed on agricultural machinery, is the index variable, For the The weight of each sensor, For the The data collected by each sensor; the extended Kalman filter is used to remove noise and abnormal values ​​in the comprehensive data set to obtain the precise longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data to form purified data .

[0024] It should be noted that the use of multi-sensor data fusion technology and Kalman filter can significantly improve the quality of data, reduce errors caused by single sensor failure or environmental factors, and ensure the effectiveness of subsequent path planning and control strategies.

[0025] S3, using the return feature elimination method to extract key state features of agricultural machinery position information, posture information, speed information and surrounding environment information of agricultural machinery from the purified data; Furthermore, the state characteristics of the key agricultural machinery position information, posture information, speed information and the surrounding environment information of the agricultural machinery are extracted from the purified data by using the recursive feature elimination method. The specific steps are: using the recursive feature elimination method to extract the longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery and three-dimensional point cloud data state characteristics of the key agricultural machinery from the purified data; using the support vector machine as the base model to evaluate the importance score of each state feature; dynamically adjusting the weight of each state feature according to the importance score of each state feature; setting the state feature extraction function , according to the characteristics of agricultural machinery in different operation stages, the feature extraction function is adaptively adjusted, and the expression is: ; in, is the extracted state feature, It is The extraction function of state features, It is to purify the data. is the number of state features extracted.

[0026] It should be noted that the feature elimination method can help identify which data features are most critical to agricultural machinery operations, thereby reducing unnecessary consumption of computing resources and improving the response speed and efficiency of the system. In addition, by dynamically adjusting feature weights, the system can better adapt to different operating environments and conditions.

[0027] S4. Build a path deviation prediction model based on historical data using the deep reinforcement learning DQN algorithm; Furthermore, based on historical data, a path deviation prediction model is constructed using the deep reinforcement learning DQN algorithm. The specific steps include: selecting the DQN algorithm in deep reinforcement learning as the basic model of the path deviation prediction model; setting the state space and action space Provide decision options for the DQN model, expressed as: ; ; in, For agricultural machinery at time step of Axis horizontal axis and Axis ordinate, For agricultural machinery at time step The orientation angle, For agricultural machinery at time step Linear speed, For agricultural machinery at time step information about the surrounding environment, For agricultural machinery at time step The state set, For agricultural machinery at time step The steering angle, For agricultural machinery at time step The speed adjustment amount, For agricultural machinery at time step action set.

[0028] S4. Build a path deviation prediction model based on historical data using the deep reinforcement learning DQN algorithm; Furthermore, a path deviation prediction model is constructed based on historical data using a deep reinforcement learning DQN algorithm. The specific steps include selecting the DQN algorithm in deep reinforcement learning as the basic model of the path deviation prediction model; Setting the state space and action space Provide decision options for the DQN model, expressed as: ; ; in, For agricultural machinery at time step of Axis horizontal axis and Axis ordinate, For agricultural machinery at time step The orientation angle, For agricultural machinery at time step Linear speed, For agricultural machinery at time step information about the surrounding environment, For agricultural machinery at time step The state set, For agricultural machinery at time step The steering angle, For agricultural machinery at time step The speed adjustment amount, For agricultural machinery at time step The action set.

[0029] It should be noted that the choice of the DQN algorithm is based on its advantages in dealing with sequential decision-making problems. Especially when faced with complex and uncertain environments, by constructing a path deviation prediction model, possible path deviations can be predicted in advance, and timely measures can be taken to correct them to ensure the continuity and accuracy of the operation.

[0030] S5. Use map services to generate the target path during the agricultural machinery operation process, input key agricultural machinery status characteristics into the path deviation prediction model, and output the difference value through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path; Furthermore, the target path of the agricultural machinery operation process is generated using the map service, and the key agricultural machinery status characteristics are input into the path deviation prediction model. The difference value is output through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path. The specific steps are as follows: Use the Google Maps API map service to obtain the coordinate points of the target path. The target path is represented as a series of coordinate points. The expression is: ; in, represents the first point on the path, where It's the first point Axis horizontal axis, It's the first point Axis ordinate, represents the second point on the path, where It's the second point Axis horizontal axis, It's the second point Axis ordinate, represents the middle point on the path, represents the last point on the path, where It's the last point. Axis horizontal axis, It's the last point. Axis ordinate, Indicates the target path; obtains the current position of the agricultural machinery through the GPS receiver ; For each point on the target path, calculate the Euclidean distance to the current position of the agricultural machine; calculate the difference value by the Euclidean distance between the current position of the agricultural machine and the nearest point on the target path, the expression is: ; in, is the difference between the current position of the agricultural machinery and the target path, and is the coordinate of the point on the target path closest to the current position of the agricultural machine, It represents the Euclidean distance between the nearest point on the target path and the current position of the agricultural machine.

[0031] It should be noted that using map services to generate target paths not only makes it easier for operators to intuitively understand the operating routes of agricultural machinery, but also helps to achieve the goal of precision agriculture. By comparing the current position of the agricultural machinery with the target path in real time, the route of the agricultural machinery can be adjusted immediately, reducing human intervention and improving operating efficiency.

[0032] S6. According to the difference value, the PID control algorithm and the adaptive control algorithm are introduced to calculate the steering angle and speed adjustment of the agricultural machinery; Furthermore, according to the difference value, the PID control algorithm and the adaptive control algorithm are introduced to calculate the steering angle and speed adjustment of the agricultural machinery. The specific steps are as follows: the difference value output by the path deviation prediction model As input, the PID control algorithm and adaptive control algorithm are used to calculate the steering angle and speed adjustment of the agricultural machinery. The expression is: ; in, is the steering angle adjustment of the agricultural machinery, is the speed adjustment of the agricultural machinery, is the proportional gain of the PID control algorithm, is the integral gain of the PID control algorithm, is the differential gain of the PID control algorithm, is the proportional gain of the adaptive control algorithm, used to dynamically adjust the speed, is the integral gain of the adaptive control algorithm, is the derivative gain of the adaptive control algorithm, is the difference between the current position of the agricultural machinery and the target path, is the current time, is the control input vector containing the steering angle and speed adjustment, Represents small increments of time.

[0033] It should be noted that the introduction of PID control algorithm and adaptive control algorithm is to cope with various uncertainties that may arise during the operation, such as terrain changes, external interference, etc. The combination of the two control methods can improve the stability and robustness of the system while ensuring the operation accuracy.

[0034] S7, providing a control strategy according to the steering angle and speed adjustment amount of the agricultural machine, sending the control strategy to the control unit of the agricultural machine through the communication interface, and the control unit executing the control strategy; Furthermore, a control strategy is given according to the steering angle and speed adjustment of the agricultural machinery, and the control strategy is sent to the control unit of the agricultural machinery through the communication interface. The control unit executes the control strategy. The specific steps are as follows: and speed adjustment , integrated into the control input vector , and generate control strategies; control input vectors are transmitted through the communication interface The control unit of the agricultural machinery receives the control strategy and executes the corresponding control instructions, including: normal mode, when the agricultural machinery is operating normally, the steering angle is , the speed adjustment is ; Emergency obstacle avoidance mode, when an obstacle is detected in front, the steering angle is , the speed adjustment is =0, the machine should stop moving forward immediately and adjust the direction to avoid obstacles; in the slow fine adjustment mode, when the machine approaches the target path, the steering angle is , the speed adjustment is When the speed is slow, fine-tune the adjustment.

[0035] It should be noted that formulating reasonable control strategies and effectively communicating them to the agricultural machinery control unit is the key to ensuring the smooth progress of the entire automated operation process. Flexibly adjusting the control strategy according to the needs of different operation scenarios can not only improve the operation quality, but also effectively protect the safety of agricultural machinery and operators.

[0036] This embodiment also provides a computer device, which is suitable for the case of a method for controlling row-to-row tracking of a crawler agricultural machinery during cross-row operation, and includes: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the method for controlling row-to-row tracking of a crawler agricultural machinery during cross-row operation as proposed in the above embodiment.

[0037] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0038] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for controlling row tracking of a crawler agricultural machinery during cross-row operation as proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a disk or an optical disk.

[0039] In summary, the present invention can comprehensively and accurately collect various status information and surrounding environment information of agricultural machinery by installing GPS receivers, IMU inertial measurement units, RGB-D cameras and LiDAR laser radars, fuse data from different sensors through weighted averaging, and use extended Kalman filters to remove noise and outliers in the data. The use of data fusion technology can integrate the advantages of different sensors and improve the accuracy and robustness of data. The Kalman filter can effectively remove noise and outliers to ensure the purity of data. The importance score of each state feature is evaluated through recursive feature elimination and support vector machines, which can screen out the most critical information from a large amount of data, reduce data redundancy, and improve the training efficiency and prediction accuracy of the model. The deep reinforcement learning DQN algorithm can learn the optimal path deviation prediction strategy through historical data to improve the accuracy and robustness of the prediction.

[0040] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a row tracking control method for a crawler-type agricultural machinery operating across rows are provided.

[0041] A GPS receiver, an IMU inertial measurement unit, an RGB-D camera and a LiDAR laser radar are installed on the crawler tractor. The sensors are used to collect the tractor's position, orientation angle, forward tilt angle, side tilt angle, speed, surrounding color images, depth information of the environment and three-dimensional point cloud data.

[0042] The sensor was activated, and a set of data was collected every 1 second when the tractor was performing cross-row operations. A total of 100 sets of data were collected, each of which included the tractor's position, posture, speed, and environmental information.

[0043] The weighted average method is used to fuse the data from different sensors to form a comprehensive data set. The extended Kalman filter is used to remove noise and outliers in the comprehensive data set to obtain the tractor's precise longitude, latitude, altitude, heading angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, surrounding color images, environmental depth information, and three-dimensional point cloud data to form purified data. The recursive feature elimination method is used to extract the key state features of the tractor's position information, posture information, speed information and surrounding environment information from the purified data. The specific steps include using support vector machine as the base model, evaluating the importance score of each state feature, dynamically adjusting the weight of each state feature according to the importance score of each state feature, setting the state feature extraction function, and adaptively adjusting the feature extraction function according to the characteristics of the tractor in different operation stages.

[0044] The DQN algorithm in deep reinforcement learning is selected as the basic model of the path deviation prediction model, and the state space and action space are set to provide decision options for the DQN model.

[0045] Use the Google Maps API map service to obtain the coordinate points of the target path, and represent the target path as a series of coordinate points.

[0046] The current position of the tractor is obtained through the GPS receiver. For each point on the target path, the Euclidean distance to the current position of the tractor is calculated. The difference value is calculated by the Euclidean distance between the current position of the tractor and the nearest point on the target path.

[0047] The difference value output by the path deviation prediction model is used as input, and the steering angle and speed adjustment of the tractor are calculated using the PID control algorithm and the adaptive control algorithm.

[0048] The calculated steering angle and speed adjustment are integrated into a control input vector and a control strategy is generated. The control input vector is sent to the tractor control unit through the communication interface. The control unit executes the corresponding control instructions, including normal mode, emergency obstacle avoidance mode and slow fine adjustment mode. The details are shown in Table 1 below: Table 1 Performance comparison of row tracking control methods for crawler agricultural machinery cross-row operation

[0049] By analyzing the experimental data in the table, the experimental group using the self-calibration method showed significant advantages in multiple dimensions compared to the control group: The position accuracy of the present invention is significantly improved, reduced from 48 cm to 6 cm, ensuring that the tractor can track the predetermined path more accurately and reduce operation errors; the posture accuracy is significantly improved, reduced from 4.2° to 1.1°, ensuring that the tractor maintains a correct posture during operation and improves the operation quality; meanwhile, the speed accuracy is significantly improved, reduced from 9.5 cm / s to 2.3 cm / s, ensuring that the tractor maintains a stable driving speed during operation and improves the operation efficiency; secondly, the path deviation is significantly reduced, reduced from 28 cm to 4.5 cm, ensuring that the tractor can travel along the predetermined path more accurately and improving the operation accuracy.

[0050] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for tracking and controlling a crawler-type agricultural machine during cross-row operation, characterized in that: include, Install a variety of high-precision sensors on crawler agricultural machinery to collect information about the position, posture, speed and surrounding environment of the agricultural machinery; Using multi-sensor data fusion technology and Kalman filter to remove noise and outliers in the collected data, the precise position, posture, speed and environmental information of the agricultural machinery can be obtained to form purified data; The key state features of the agricultural machinery position information, posture information, speed information and the surrounding environment information of the agricultural machinery are extracted from the purified data by using the regression feature elimination method; Based on historical data, a path deviation prediction model is constructed using the deep reinforcement learning DQN algorithm; Use map services to generate the target path during the operation of agricultural machinery, input key agricultural machinery status characteristics into the path deviation prediction model, and output the difference value through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path; According to the difference value, PID control algorithm and adaptive control algorithm are introduced to calculate the steering angle and speed adjustment of the agricultural machinery; A control strategy is given according to the steering angle and speed adjustment amount of the agricultural machinery, and the control strategy is sent to the control unit of the agricultural machinery through the communication interface, and the control unit executes the control strategy.

2. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The method of installing a plurality of high-precision sensors on the crawler agricultural machinery to collect the position, posture, speed and surrounding environment information data of the agricultural machinery is as follows: Install GPS receivers, IMU inertial measurement units, RGB-D cameras and LiDA laser radars on crawler agricultural machinery; Start the sensor to collect the position, orientation angle, forward tilt angle, side tilt angle, speed, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data.

3. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The multi-sensor data fusion technology is combined with the Kalman filter to remove noise and outliers in the collected data to obtain the precise position, posture, speed and environmental information of the agricultural machinery. The specific steps are: The data from different sensors are fused using the weighted average method to form a comprehensive data set, which is expressed as: ； in, For a comprehensive dataset, is the number of sensors installed on agricultural machinery, is the index variable, For the The weight of the sensor, For the Data collected by sensors; The extended Kalman filter is used to remove noise and abnormal values ​​in the comprehensive data set to obtain the precise longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data to form purified data. .

4. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The feature elimination method is used to extract the key state features of the agricultural machinery position information, posture information, speed information and the surrounding environment information of the agricultural machinery from the purified data. The specific steps are as follows: The recursive feature elimination method is used to extract the longitude, latitude, altitude, orientation angle, forward tilt angle, roll angle, angular velocity, acceleration, linear velocity, color images around the agricultural machinery, depth information of the surrounding environment of the agricultural machinery, and three-dimensional point cloud data status features of the key agricultural machinery from the purified data; Support vector machine is used as the base model to evaluate the importance score of each state feature; Dynamically adjust the weight of each state feature according to the importance score of each state feature; Setting state feature extraction function , according to the characteristics of agricultural machinery in different operation stages, the feature extraction function is adaptively adjusted, and the expression is: ； in, is the extracted state feature, It is The extraction function of state features, It is to purify the data. is the number of state features extracted.

5. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The path deviation prediction model is constructed based on historical data using the deep reinforcement learning DQN algorithm. The specific steps include: Select the DQN algorithm in deep reinforcement learning as the basic model of the path deviation prediction model; Setting the state space and action space Provide decision options for the DQN model, expressed as: ； ； in, For agricultural machinery at time step of Axis horizontal axis and Axis ordinate, For agricultural machinery at time step The orientation angle, For agricultural machinery at time step Linear speed, For agricultural machinery at time step information about the surrounding environment, For agricultural machinery at time step The state set, For agricultural machinery at time step The steering angle, For agricultural machinery at time step The speed adjustment amount, For agricultural machinery at time step The action set.

6. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The target path of the agricultural machinery operation process is generated by using the map service, key agricultural machinery status features are input into the path deviation prediction model, and the difference value is output through the Euclidean distance between the current position of the agricultural machinery and the nearest point on the target path. The specific steps are as follows: Use the Google Maps API map service to obtain the coordinate points of the target path. The target path is represented as a series of coordinate points. The expression is: ； in, represents the first point on the path, where It's the first point Axis horizontal axis, It's the first point Axis ordinate, represents the second point on the path, where It's the second point Axis horizontal axis, It's the second point Axis ordinate, represents the middle point on the path, represents the last point on the path, where It's the last point. Axis horizontal axis, It's the last point. Axis ordinate, Indicates the target path; Get the current location of the agricultural machinery through the GPS receiver ; For each point on the target path, calculate the Euclidean distance to the current position of the agricultural machine; The difference value is calculated by the Euclidean distance between the current position of the agricultural machine and the nearest point on the target path. The expression is: ； in, is the difference between the current position of the agricultural machinery and the target path, and is the coordinate of the point on the target path closest to the current position of the agricultural machine, It represents the Euclidean distance between the nearest point on the target path and the current position of the agricultural machine.

7. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: According to the difference value, the PID control algorithm and the adaptive control algorithm are introduced to calculate the steering angle and speed adjustment of the agricultural machinery. The specific steps are: The difference value output by the path deviation prediction model As input, the PID control algorithm and adaptive control algorithm are used to calculate the steering angle and speed adjustment of the agricultural machinery. The expression is: ； in, is the steering angle adjustment of the agricultural machinery, is the speed adjustment of the agricultural machinery, is the proportional gain of the PID control algorithm, is the integral gain of the PID control algorithm, is the differential gain of the PID control algorithm, is the proportional gain of the adaptive control algorithm, used to dynamically adjust the speed, is the integral gain of the adaptive control algorithm, is the derivative gain of the adaptive control algorithm, is the difference between the current position of the agricultural machinery and the target path, is the current time, is the control input vector containing the steering angle and speed adjustment, Represents small increments of time.

8. The row tracking control method for crawler-type agricultural machinery in cross-row operation according to claim 1, characterized in that: The control strategy is given according to the steering angle and speed adjustment of the agricultural machine, and the control strategy is sent to the control unit of the agricultural machine through the communication interface, and the control unit executes the control strategy. The specific steps are: Based on the calculated steering angle and speed adjustment , integrated into the control input vector , and generate control strategies; The control input vector is transmitted through the communication interface sent to the control unit of the agricultural machinery; After receiving the control strategy, the control unit of the agricultural machinery executes the corresponding control instructions, including: In normal mode, when the agricultural machinery is operating normally, the steering angle is , the speed adjustment is ; In the emergency obstacle avoidance mode, when an obstacle is detected ahead, the steering angle is , the speed adjustment is =0, you should stop moving immediately and adjust the direction to avoid obstacles; In the slow fine adjustment mode, when the agricultural machine approaches the target path, the steering angle is , the speed adjustment is When the speed is slow, fine-tune the adjustment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the row tracking control method for cross-row operation of a crawler-type agricultural machinery as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the row tracking control method for cross-row operation of a crawler-type agricultural machinery as described in any one of claims 1 to 7 are implemented.

Images