A wheeled unmanned agricultural machine trajectory tracking control method based on whole-machine slip and slip ratio estimation

Abstract

The present application relates to the technical fields of agricultural machinery automation control and unmanned agricultural machinery navigation operation, and particularly relates to a wheeled unmanned agricultural machinery trajectory tracking control method based on whole machine slip and slippage estimation, and aims to solve the problem that the existing wheeled unmanned agricultural machinery trajectory tracking control method is easily affected by the interaction between wheels and soil in soft soil, wet and slippery ground, slope and non-flat farmland environment, leading to the difficulty in timely compensation of whole machine longitudinal slip, transverse unexpected slippage and heading deviation, thereby causing the problems of increased trajectory tracking error, delayed attitude adjustment and decreased driving stability, and provides a wheeled unmanned agricultural machinery trajectory tracking control method based on whole machine slip and slippage estimation, which comprises: motion state acquisition and trajectory error calculation, whole machine slip and slippage kinematics representation, trajectory tracking controller construction, whole machine slip and slippage online estimation, control amount compensation based on the estimation result, chassis execution and closed-loop update.

Description

Technical Field

[0001] This invention relates to the field of agricultural machinery automation control and unmanned agricultural machinery navigation technology, specifically to a trajectory tracking control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation. Background Technology

[0002] Wheeled unmanned agricultural machinery is an important form of equipment for achieving automated, precise, and less-manned operations in agricultural production. It can be widely used in various agricultural production processes such as sowing, fertilization, plant protection, weeding, field transportation, and inspection. In actual operation, wheeled unmanned agricultural machinery typically needs to travel continuously along a preset working trajectory. The accuracy of its trajectory tracking directly affects the consistency of row spacing, the occurrence of overlapping or missed work, operational efficiency, and agronomic quality. Therefore, establishing a trajectory tracking control method suitable for complex field environments is a crucial technological foundation for improving the autonomous operation capabilities of wheeled unmanned agricultural machinery.

[0003] Currently, most wheeled unmanned agricultural machinery relies on onboard sensors such as BeiDou / GNSS positioning, inertial measurement units, and wheel speed encoders for automatic navigation and trajectory tracking control. Common control methods include pure tracking control, PID control, fuzzy control, sliding mode control, and model predictive control. These methods can achieve certain control effects on flat, hard surfaces or under low-slip conditions. However, in soft soil, slippery surfaces, slopes, wheel ruts, and uneven farmland environments, the adhesion between tires and soil changes significantly, and the wheel-ground interaction becomes complex. This easily leads to longitudinal slippage, undesirable lateral slippage, and heading deviation, causing a deviation between the actual motion state of the agricultural machinery and the ideal motion model. Consequently, this results in increased trajectory tracking errors, delayed attitude adjustment, and decreased driving stability.

[0004] To address the problems of automatic navigation, sideslip estimation, and path tracking control for agricultural machinery, CN113721609B discloses a trajectory tracking control method for a 4WID high ground clearance sprayer under sideslip conditions. This method incorporates lateral slip velocity and front and rear axle steering deviations into the sprayer's pose error model and compensates for unknown sideslip through adaptive control to improve trajectory tracking accuracy under sideslip conditions. CN111238471B discloses a sideslip angle estimation method and estimator suitable for straight-line navigation of agricultural machinery, used for sideslip angle estimation during the straight-line navigation process of agricultural machinery. CN108693773B discloses an adaptive estimation method for landslide deviation in automatic driving of agricultural machinery. By estimating the heading angle and front wheel steering angle errors caused by sloping terrain, it improves the adaptability of the automatic driving system of agricultural machinery to terrain changes. CN109374315B discloses a method, control method, system, and device for obtaining instantaneous wheel slip rate of vehicles, used to improve the accuracy and robustness of slip rate calculation for wheeled vehicles on complex road surfaces.

[0005] While the aforementioned publicly available technologies have considered issues such as sideslip of agricultural machinery, slope deviation, or slip rate of wheeled vehicles, they still have certain limitations: First, existing sideslip control methods for agricultural machinery are mostly designed for specific equipment configurations, such as 4WID high-clearance sprayers. They primarily correct the posture error model by adjusting the lateral slip velocity or sideslip angle, and have not yet made a unified estimate of the longitudinal slip and undesired lateral slip of wheeled unmanned agricultural machinery. Second, existing methods for obtaining slip rate mostly calculate the instantaneous slip rate of each wheel from the wheel-level perspective, resulting in a large number of parameters and failing to form a closed-loop online compensation system for overall machine-level slip and slip for field trajectory tracking control. Third, existing agricultural machinery path tracking control methods are usually based on lateral deviation, heading error, path curvature, or vehicle kinematics models, lacking a technical solution to directly introduce the overall machine slip and slip estimation results into the control feedback loop for real-time compensation.

[0006] In summary, while existing trajectory tracking control technologies for wheeled unmanned agricultural machinery have explored aspects such as automatic navigation, feedback control, sideslip estimation, and model predictive control, a method is still lacking that can perform online estimation of the entire machine's slip and turn state under complex field conditions and incorporate the estimation results into the control feedback loop for real-time compensation. Therefore, it is necessary to propose a trajectory tracking control method for wheeled unmanned agricultural machinery based on overall machine slip and turn estimation to improve the trajectory tracking accuracy and driving stability of wheeled unmanned agricultural machinery in soft, slippery, sloping, and uneven farmland environments. Summary of the Invention

[0007] The problem that this invention aims to solve is that existing wheeled unmanned agricultural machinery trajectory tracking control methods are easily affected by wheel-ground interaction in soft soil, slippery ground, sloping land, and uneven farmland environments. This leads to difficulties in timely compensation for longitudinal slippage, undesirable lateral slippage, and heading deviation, resulting in increased trajectory tracking errors, delayed attitude adjustment, and decreased driving stability.

[0008] To address the aforementioned problems, this invention provides a trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation. The method includes the following steps: S1. Motion state acquisition and trajectory error calculation: Acquire the current motion state information of the wheeled unmanned agricultural machinery, and calculate and output the longitudinal position error, lateral position error and heading angle error in the vehicle coordinate system based on the difference between the actual pose of the wheeled unmanned agricultural machinery and the reference trajectory. S2. Kinematic characterization of the entire machine's sliding and turning; calculate the ideal longitudinal velocity and ideal angular velocity of the wheeled unmanned agricultural machinery under no-slip conditions, and determine the actual longitudinal velocity, lateral velocity, and heading angular velocity in the vehicle coordinate system by combining positioning and inertial measurement information. Based on the difference between the ideal motion state and the actual motion state, define the longitudinal sliding parameters and the lateral undesired turning parameters of the entire machine. The longitudinal slip parameter of the whole machine is used to characterize the longitudinal speed loss of wheeled unmanned agricultural machinery due to the tire-soil interaction. The lateral undesired slip parameter is used to characterize the lateral and heading deviation trends caused by lateral slip, attachment changes, or ground disturbances. S3. Trajectory tracking controller construction: Based on the trajectory error in the vehicle coordinate system and the kinematic characterization of the machine's slip and rotation, a dynamic model of the trajectory error of the wheeled unmanned agricultural machinery is constructed to generate the basic tracking control quantity of the wheeled unmanned agricultural machinery before slip and rotation compensation is performed. S4. Online estimation of overall machine slip and rotation: Based on the control timing information of the wheeled unmanned agricultural machinery in the continuous control cycle, the online estimation model of overall machine slip and rotation is constructed and the estimated longitudinal slip and the estimated lateral undesired rotation of the whole machine under the current control cycle are output. S5. Control quantity compensation based on estimation results: The longitudinal slip estimate and the lateral undesired slip estimate of the whole machine are obtained and introduced into the trajectory tracking control feedback loop to correct the uncompensated desired linear velocity and desired angular velocity; S6. Chassis execution and closed-loop update: The compensated chassis control commands are sent to the wheeled unmanned agricultural machinery chassis controller, and new motion state information is continuously collected to update the trajectory error and the overall machine slip and turn estimation results.

[0009] Furthermore, the acquisition of motion state and calculation of trajectory error also includes the following steps: S101. Acquisition of vehicle motion state information: Position and pose information is acquired through an inertial measurement unit, the rotational speed of each wheel is acquired through a wheel speed encoder, and the current linear velocity, angular velocity, or drive control commands are acquired through the chassis controller to construct a position and pose state describing the actual driving position and direction of the agricultural machinery. : ,in, and This represents the position coordinates of the wheeled unmanned agricultural machinery in the global coordinate system. Indicates the heading angle; S102, Multi-source information synchronization and alignment and control state definition; coordinate transformation, outlier removal and smoothing of pose data; conversion and smoothing of wheel speed encoder data to obtain wheel speed information within the current control cycle, and alignment of chassis control commands with sensor data according to timestamps to construct the first... Effective state vector for each control cycle : ,in, Indicates the current pose. Indicates attitude angle, Indicates angular velocity, Represents linear acceleration. Indicates the wheel speed. Indicates the current chassis control command; S103. Trajectory error calculation and coordinate transformation; obtain the current motion state of the unmanned wheeled agricultural machinery and get its current actual pose. and current reference pose Calculate the pose error of a wheeled unmanned agricultural machine relative to a reference trajectory. The pose error of the reference trajectory The coordinates are converted to the vehicle body coordinate system to calculate the longitudinal position error of the current wheeled unmanned agricultural machinery. Lateral position error Heading angle error ; The reference pose ,in and Indicates the position coordinates of the reference trajectory point. Indicates the reference heading angle; The pose error of the reference trajectory ,in, and This represents the positional deviation in the global coordinate system. Indicates the deviation of the heading angle; Based on the control parameters of wheeled unmanned agricultural machinery, the error will be... Further transformation to the vehicle coordinate system Below, the vehicle coordinate system : ,in, This represents the longitudinal positional error along the direction of vehicle movement. This represents the lateral positional error perpendicular to the direction of vehicle movement. This indicates the heading angle error.

[0010] Furthermore, the kinematic characterization of the entire machine's sliding and rotating motion also includes the following steps: S201, Calculation of ideal motion state; the calculation of ideal motion state includes left and right wheel speeds, ideal longitudinal speed, and ideal angular velocity; the calculation of left and right wheel speeds involves obtaining the rotational speeds of each drive wheel from the wheel speed encoder and equating the rotational speeds of wheels on the same side to the left wheel speed. and right wheel speed : ， ,in, and These represent the equivalent angular velocities of the left and right wheels, respectively. Indicates the first The angular velocity of each wheel and These represent the number of wheels on the left and right sides involved in the equivalent calculation, respectively. The calculation of the ideal longitudinal velocity and ideal angular velocity is based on the wheel radius. and left and right wheelbase and the speed of the left wheel and right wheel speed Calculate the ideal longitudinal speed of wheeled unmanned agricultural machinery and ideal angular velocity : , ， in, This represents the ideal longitudinal velocity calculated from the wheel speed under no-slip conditions. Represents the ideal angular velocity under no-slip conditions; S202. Determining the actual motion state; based on the rate of change of the wheeled unmanned agricultural machinery's position in the global coordinate system... and The heading angle is The actual longitudinal velocity in the vehicle coordinate system was calculated. and lateral speed The in, The actual longitudinal velocity along the direction of vehicle movement, the ,in, This represents the actual lateral velocity perpendicular to the direction of the vehicle's movement. S203. Definition of overall machine sliding and rotation parameters; Define the longitudinal sliding parameters of the entire machine based on the ideal motion state and the actual motion state. Deviation from lateral movement Establish a planar kinematic model of the vehicle; The longitudinal slip parameter ,in, A positive number is set to prevent the denominator from being too small at low speeds or when stationary; the aforementioned Used to characterize the degree of loss of actual longitudinal speed of wheeled unmanned agricultural machinery relative to ideal longitudinal speed. The larger the value, the more significant the longitudinal speed loss of the wheeled unmanned agricultural machinery due to the action of wheels on the ground. The deviation of the lateral movement Used to characterize the deviation of the actual lateral motion of wheeled unmanned agricultural machinery from the ideal lateral motion. ,in Represents the lateral velocity under ideal conditions; The construction of the vehicle planar kinematics model , , , , ， ,in, and This indicates the actual motion state of the wheeled unmanned agricultural machinery.

[0011] Furthermore, the construction of the trajectory tracking controller also includes the following steps: S301. Trajectory error dynamics construction; based on the error vector in the vehicle coordinate system obtained in step S1... In addition to the planar kinematics of wheeled unmanned agricultural machinery, a dynamic model of trajectory tracking error is established: , , ,in, , and These represent the rates of change of longitudinal error, lateral error, and heading error, respectively. S302. Sliding surface construction; constructing the sliding surface based on longitudinal and lateral position errors. and : , ,in, and These are the longitudinal and transverse sliding mode variables, respectively. and It is a positive control parameter; The longitudinal sliding mode variable and lateral sliding mode variables When the values ​​approach zero, the longitudinal and lateral position errors can converge, achieving stable tracking of the reference trajectory. S303. Generation of uncompensated control law: Based on the constructed sliding surface, the uncompensated control law is solved according to the error dynamics model and the sliding mode arrival condition. : ,in, Positive control gain, For boundary layer parameters, It is a saturation function; S304, Control Variable Smoothing and Constraint Handling; Based on the Generation of Uncompensated Control Variables For uncompensated expected linear velocity and uncompensated expected angular velocity Perform smoothing and amplitude limiting processing. , ,in, , The permissible linear velocity range for wheeled unmanned agricultural machinery. , This refers to the permissible range of angular velocities.

[0012] Furthermore, the online estimation of the overall machine's sliding and rotation also includes the following steps: S401. Construction of temporal input features; construction of estimation model input from sensor data within a continuous time window. : ,in, Indicates the first The input feature vector for each control cycle; The input features This includes information on wheel speed, uncompensated expected linear velocity, uncompensated expected angular velocity, attitude angle, angular velocity, linear acceleration, and changes in heading angle. S402. Construction of the overall machine sliding and rotation estimation model; based on the obtained time-series input sequence. Construct a sliding and rotation estimation model for the entire machine, the estimation model including a longitudinal sliding estimation model and a lateral undesired rotation estimation model; S403, Online estimation of longitudinal slip of the whole machine; input the timing sequence of the current control cycle. Input the longitudinal slip estimation model of the whole machine, and obtain the longitudinal slip estimate of the whole machine at the current time. : ,in, This represents the longitudinal slip estimation model for the entire machine. Indicates the first Estimated longitudinal slip of the entire machine in one control cycle; The longitudinal slip estimate of the whole machine is used to reflect the deviation between the ideal longitudinal speed calculated from the wheel speed and the actual longitudinal movement speed of the wheeled unmanned agricultural machinery; S404. Online estimation and result update of undesirable lateral slip: Using the same time-series input sequence as the longitudinal slip estimation, the undesirable lateral slip estimation model is input to obtain the estimator of the undesirable lateral slip for the current control cycle. : ,in This represents the lateral undesirable slip estimation model. Indicates the first The estimator of the lateral undesirable slip for each control cycle.

[0013] Furthermore, the control quantity compensation based on the estimation results also includes the following steps: S501. Determining the longitudinal slip compensation amount: Based on the uncompensated expected linear velocity output in step S3 and the smoothed longitudinal slip estimate of the whole machine obtained in step S4, the expected linear velocity is corrected to obtain the compensated linear velocity. : ,in, This represents the desired linear velocity after longitudinal slip compensation. A positive number is set to prevent the denominator from being too small; S502. Determination of lateral undesired slip compensation: Based on the uncompensated desired linear velocity output in step S3 and the smoothed longitudinal slip estimate of the entire machine obtained in step S4, the desired angular velocity is corrected to obtain the compensated angular velocity. : ,in, This represents the desired angular velocity after lateral undesired slip compensation. This represents the equivalent steering characteristic length of the wheeled unmanned agricultural machinery chassis. S503, generation of compensation control commands; after longitudinal slip and lateral undesired slip compensation, the compensation control quantity is obtained. : ,in This indicates the motion control command after compensation in the current control cycle; S504, Control Constraints and Outputs; Limiting the compensated linear velocity and angular velocity to output the linear velocity within a reasonable range. and angular velocity .

[0014] Furthermore, the chassis execution and closed-loop update also includes the following steps: S601, issue compensation control command; send the compensation control command generated in step S5 to the chassis controller of the wheeled unmanned agricultural machinery, the control command including the desired vehicle speed, steering angle, steering angular velocity or driving torque command; S602, Execution status feedback acquisition; Collect the position, heading angle, attitude angle, angular velocity, linear acceleration and wheel speed information of the wheeled unmanned agricultural machinery according to the preset control cycle, and simultaneously obtain the execution status feedback from the chassis controller; S603, Closed-loop control update: Obtain the latest collected motion state information, recalculate the error information of the wheeled unmanned agricultural machinery relative to the reference trajectory, input the calculation output result into the whole machine slip and turn estimation model, update the whole machine longitudinal slip estimate and lateral undesired slip estimate to generate a new basic trajectory tracking control quantity, combine the updated slip and turn estimation result for compensation, and form the chassis control command for the next control cycle.

[0015] Compared with existing technologies, this invention provides a unified estimation of longitudinal slippage and undesirable lateral slippage of wheeled unmanned agricultural machinery at the whole-machine level, significantly reducing the number of parameters to be estimated and lowering sensor configuration requirements and engineering implementation complexity. Simultaneously, it relies on onboard sensor information to identify the slippage and slippage state of the entire machine online and directly integrates the state estimate into the trajectory tracking control feedback loop to achieve real-time compensation for motion deviations. Furthermore, under complex working conditions representing soft, slippery, sloping, and uneven farmland, it accurately reflects the deviation between the actual motion state of the agricultural machinery and the ideal motion model, effectively reducing trajectory tracking errors, improving the overall driving stability and row spacing consistency in the field, and is compatible with existing wheeled unmanned agricultural machinery equipped with BeiDou / GNSS positioning, inertial measurement units, wheel speed encoders, and chassis control systems, making it easy to integrate into existing agricultural machinery automatic driving systems for widespread application. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and rotation estimation; Detailed Implementation

[0017] To better understand the purpose, structure, and function of this invention, the following description, in conjunction with the accompanying drawings, provides a more detailed account of a wheeled unmanned agricultural machinery trajectory tracking and control method based on whole-machine slip and turn estimation.

[0018] like Figure 1 The present invention provides a trajectory tracking control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation. The method includes steps S1 to S6, which are described in detail below: S1. Motion state acquisition and trajectory error calculation; Regarding step S1, it should be noted that when the wheeled unmanned agricultural machinery performs autonomous field operations, it is first necessary to obtain its current motion state and determine its deviation from the preset working trajectory. This step uses vehicle positioning, inertial measurement, wheel speed measurement, and chassis control information to form the state input for the current control cycle and calculate the error amount used for subsequent trajectory tracking control.

[0019] Specifically, the current motion status information of the wheeled unmanned agricultural machinery is obtained, and based on the difference between the actual posture of the wheeled unmanned agricultural machinery and the reference trajectory, the longitudinal position error, lateral position error and heading angle error in the vehicle coordinate system are calculated and output.

[0020] In specific implementation, step S1 also includes steps S101 to S103, which are described in detail below: S101. Acquisition of vehicle motion status information; Specifically, during operation, wheeled unmanned agricultural machinery obtains its position coordinates in the global coordinate system through the Beidou / GNSS positioning module, obtains heading angle, attitude angle, angular velocity and linear acceleration information through the inertial measurement unit, obtains the rotational speed of each wheel through the wheel speed encoder, and obtains the current linear velocity, angular velocity or drive control commands through the chassis controller.

[0021] Furthermore, in this embodiment, the pose state of the wheeled unmanned agricultural machinery is represented as follows: ,in and This represents the position coordinates of the wheeled unmanned agricultural machinery in the global coordinate system. The heading angle is indicated. The above pose information is used to describe the actual driving position and direction of the agricultural machinery. Wheel speed and chassis control information are used for subsequent calculations of the ideal motion state, and inertial measurement information is used to reflect the attitude changes and motion response during field driving.

[0022] S102, Multi-source information synchronization alignment and control state definition; It should be noted that because the sampling frequencies of the positioning module, inertial measurement unit, wheel speed encoder, and chassis controller are usually inconsistent, directly using the raw data can easily cause time misalignment between state variables. Therefore, after obtaining the raw sensor information, time synchronization and preprocessing are performed on various types of data.

[0023] Specifically, coordinate transformation, outlier removal, and smoothing are performed on GNSS position data; inertial measurement data is filtered to reduce the impact of vibration and noise on attitude, angular velocity, and acceleration information; wheel speed encoder data is converted and smoothed to obtain wheel speed information within the current control cycle; and chassis control commands are aligned with sensor data according to timestamps.

[0024] After synchronization and preprocessing, the first... Effective state vector for each control cycle: , in, Indicates the current pose. Indicates attitude angle, Indicates angular velocity, Represents linear acceleration. Indicates the wheel speed. This indicates the current chassis control command.

[0025] S103. Trajectory error calculation and coordinate transformation; Specifically, after obtaining the current motion state, the reference trajectory state corresponding to the current control cycle is determined according to the preset field operation path.

[0026] The preset field operation path can be composed of a series of discrete path points, or it can be represented by a straight line, circular arc, spline curve or other continuous curve.

[0027] Furthermore, based on the current position of the wheeled unmanned agricultural machinery, a target reference point or pre-aiming reference point matching the current position is selected from the preset operation path, and the corresponding position, heading angle, reference linear velocity, and reference angular velocity of the reference point are obtained. The current reference pose is represented as follows: : ,in and Indicates the position coordinates of the reference trajectory point. Indicates the reference heading angle.

[0028] Reference linear velocity and reference angular velocity Used to describe the expected motion state of agricultural machinery at this reference point.

[0029] Furthermore, based on the current actual pose and reference pose Calculate the pose error of the wheeled unmanned agricultural machinery relative to the reference trajectory: , in, and This represents the positional deviation in the global coordinate system. This indicates the deviation in heading angle.

[0030] Since the control variables of wheeled unmanned agricultural machinery are usually applied to the forward and turning directions of the vehicle, in order to facilitate subsequent controller control based on the machine's own movement direction, the above errors are further converted to the vehicle coordinate system: , in, This represents the longitudinal positional error along the direction of vehicle movement. This represents the lateral positional error perpendicular to the direction of vehicle movement. This indicates the heading angle error.

[0031] At this point, the state acquisition, reference trajectory matching, and trajectory error calculation within the current control cycle are completed. The resulting error will serve as the basic input for subsequent whole-machine sliding and turning characterization and trajectory tracking controller design.

[0032] S2. Kinematic characterization of the entire machine's sliding and rotation; It should be noted that in step S2, after acquiring the current state of the wheeled unmanned agricultural machinery and calculating the trajectory error, this step further establishes a slip and roll characterization method at the whole-machine level. This characterization method no longer describes the slip ratio or sideslip angle of each wheel separately, but treats the wheeled unmanned agricultural machinery as a whole, describing the deviation of the actual motion state from the ideal motion state through the longitudinal slip parameters and the lateral undesired roll parameters of the whole machine.

[0033] Specifically, the ideal longitudinal velocity and ideal angular velocity of the wheeled unmanned agricultural machinery under no-slip conditions are calculated. The actual longitudinal velocity, lateral velocity and heading angular velocity in the vehicle coordinate system are determined by combining positioning and inertial measurement information. Based on the difference between the ideal motion state and the actual motion state, the longitudinal slip parameters and the lateral undesired slip parameters of the whole machine are defined. The longitudinal slip parameter of the whole machine is used to characterize the longitudinal speed loss of wheeled unmanned agricultural machinery due to the tire-soil interaction. The lateral undesired slip parameter is used to characterize the lateral and heading deviation trends caused by lateral slip, attachment changes, or ground disturbances. In specific implementation, step S2 also includes steps S201 to S203, which are described in detail below: S201, Calculation of ideal motion state; For wheeled unmanned agricultural machinery that uses left and right side drive or four-wheel independent drive, the rotational speed of each drive wheel is obtained according to the wheel speed encoder, and the rotational speed of the wheel on the same side is equivalent to the left wheel speed and the right wheel speed: , in, and These represent the equivalent angular velocities of the left and right wheels, respectively. Indicates the first The angular velocity of each wheel and These represent the number of wheels on the left and right sides involved in the equivalent calculation, respectively.

[0034] Furthermore, under slip-free and rotation-free conditions, based on the wheel radius... and left and right wheelbase Calculate the ideal longitudinal velocity and ideal angular velocity of the wheeled unmanned agricultural machinery: ， ， in, This represents the ideal longitudinal velocity calculated from the wheel speed under no-slip conditions. This represents the ideal angular velocity under no-slip conditions. For wheeled unmanned agricultural machinery with steering mechanisms, the corresponding ideal angular velocity can also be calculated based on vehicle speed, turning angle, and wheelbase.

[0035] S202, Determination of actual motion state; Specifically, when wheeled unmanned agricultural machinery travels in the field, its actual motion state is usually not entirely equivalent to the calculated wheel speed due to the influence of loose soil, slippery ground, slopes, or wheel ruts. Therefore, this step determines the actual motion state in the vehicle coordinate system based on positioning and inertial measurement information.

[0036] Furthermore, let the rate of position change of the wheeled unmanned agricultural machinery in the global coordinate system be... and The heading angle is Then its actual longitudinal and lateral velocities in the vehicle coordinate system can be expressed as: ， , in, This indicates the actual longitudinal velocity along the direction of the vehicle's movement. This represents the actual lateral velocity perpendicular to the vehicle's forward direction. The actual yaw rate can be obtained from an inertial measurement unit or a yaw rate differential, denoted as . .

[0037] S203, Definition of overall machine sliding and rotation parameters; Specifically, after obtaining the ideal motion state and the actual motion state, the longitudinal sliding parameters of the whole machine are defined. This is used to characterize the degree of loss of the actual longitudinal speed of wheeled unmanned agricultural machinery relative to the ideal longitudinal speed. , in, A positive number is set to prevent the denominator from being too small when the speed is low or the state is stationary. The larger the value, the more significant the longitudinal speed loss of the wheeled unmanned agricultural machinery due to the action of the wheels on the ground.

[0038] Further define the lateral undesired slip parameters This is used to characterize the deviation of the actual lateral motion of a wheeled unmanned agricultural machine from its ideal lateral motion. , in, This represents the lateral velocity under ideal conditions. For wheeled unmanned agricultural machinery satisfying nonholonomic constraints, it can typically be taken as... For chassis with an equivalent instantaneous steering center offset, calculations can also be performed based on the chassis geometry. .

[0039] Therefore, the actual motion state of wheeled unmanned agricultural machinery under complex field conditions can be expressed as: ， , And further, the vehicle's planar kinematics model is substituted in: ， ， ， Based on the above definition, the slip and turn state of wheeled unmanned agricultural machinery is transformed from wheel-by-wheel parameter estimation to whole-machine longitudinal slip parameters. and lateral undesired slip parameters Two state variables. This representation method can reduce the number of parameters to be estimated and provide a unified state description for subsequent online estimation of the entire machine's slip and rotation and compensation of control variables.

[0040] S3, Construction of the trajectory tracking controller; It should be noted that in step S3, after obtaining the trajectory error in the vehicle coordinate system and the overall sliding and turning kinematics, this step constructs a trajectory tracking controller to generate the basic tracking control quantities for the wheeled unmanned agricultural machinery before sliding and turning compensation. This controller aims to reduce longitudinal position error, lateral position error, and heading angle error, enabling the agricultural machinery to travel stably along the reference trajectory and providing correctable desired linear velocity and desired angular velocity for subsequent sliding and turning compensation.

[0041] Specifically, based on the trajectory error in the vehicle coordinate system and the kinematic characterization of the machine's slip and rotation, a dynamic model of the trajectory error of the wheeled unmanned agricultural machinery is constructed to generate the basic tracking control quantities of the wheeled unmanned agricultural machinery before slip and rotation compensation is performed.

[0042] In specific implementation, step S3 also includes steps S301 to S304, which are described in detail below: S301, Trajectory Error Dynamics Construction; Specifically, based on the error vector in the vehicle coordinate system obtained in step S1 In addition to the planar kinematics of wheeled unmanned agricultural machinery, a dynamic model of trajectory tracking error is established. Let the reference linear velocity and reference angular velocity corresponding to the reference trajectory be respectively... and The actual longitudinal speed, lateral speed, and angular velocity of the wheeled unmanned agricultural machinery are as follows: , and Then the error dynamics can be expressed as: ， ， , in, , and These represent the rates of change of longitudinal error, lateral error, and heading error, respectively. The above model reflects the dynamic deviation between the actual motion state of the wheeled unmanned agricultural machinery and the reference trajectory, and forms the basis for subsequent control law design.

[0043] S302, Sliding surface construction; It should be noted that, in order to improve the robustness of trajectory tracking control to model uncertainties and field disturbances, this embodiment adopts the sliding mode control concept to construct the trajectory tracking controller.

[0044] Specifically, the sliding surface is constructed based on the longitudinal and lateral positional errors: ， ， in, and These are the longitudinal and transverse sliding mode variables, respectively. and This is a positive control parameter. When... and When the values ​​approach zero, the longitudinal and lateral position errors can converge, thereby enabling the wheeled unmanned agricultural machinery to stably track the reference trajectory.

[0045] In practice, the heading angle error can also be adjusted according to the controlled object and operating speed requirements. Introducing sliding surfaces, such as constructing combined sliding variables that include lateral and heading errors, can enhance attitude adjustment capabilities during turning or curved path tracking.

[0046] S303, Generation of uncompensated control law; Specifically, after constructing the sliding surface, the uncompensated control law is solved based on the error dynamics model and the sliding surface arrival condition.

[0047] Furthermore, let the desired control quantity output under uncompensated conditions be... ,in This indicates the uncompensated expected linear velocity. This indicates the uncompensated desired angular velocity.

[0048] Furthermore, this control quantity consists of an equivalent control term and a robust correction term, namely... ,in Used to make wheeled unmanned agricultural machinery move along a reference trajectory under an ideal motion model. It is used to suppress the impact of model error, ground disturbance and parameter uncertainty on tracking error.

[0049] In practical implementation, the following sliding mode arrival law can be set: , in, Positive control gain, For boundary layer parameters, It is a saturation function. According to the arrival law... and The solution is performed to make the trajectory error state converge toward the sliding surface.

[0050] S304, Smoothing and Constraint Handling of Control Variables; Specifically, due to physical limitations such as speed, angular velocity, and driving torque of the actuators in wheeled unmanned agricultural machinery chassis, directly outputting unconstrained control quantities may lead to abrupt changes in the actuator response or control chattering. Therefore, after generating the uncompensated control quantity, the following steps are taken: and Perform smoothing and limiting: ， ， in, , The permissible linear velocity range for wheeled unmanned agricultural machinery. , This refers to the permissible range of angular velocities.

[0051] By following the steps above, the basic control quantity for trajectory tracking without slip compensation can be obtained. This control quantity will then be combined with the overall longitudinal slip estimate and the lateral undesired slip estimate to further generate compensatory control commands adapted to the current field attachment conditions.

[0052] S4. Online estimation of overall machine sliding and rotation; It should be noted that in step S4, after obtaining the basic trajectory tracking control input, it is necessary to further determine whether the wheeled unmanned agricultural machinery exhibits significant longitudinal slippage or undesirable lateral slippage under the current field conditions. This step utilizes onboard sensor information to construct a time-series input and outputs the estimated longitudinal slippage and undesirable lateral slippage of the entire machine within the current control cycle through a whole-machine slippage estimation model, providing a basis for subsequent control input compensation.

[0053] Specifically, based on the control timing information of the wheeled unmanned agricultural machinery within the continuous control cycle, an online estimation model for the overall machine's slip and rotation is constructed as input, and the estimated longitudinal slip and the estimated lateral undesired rotation of the overall machine under the current control cycle are output. In specific implementation, step S4 also includes steps S401 to S404, which are described in detail below: S401, Construction of temporal input features; Specifically, during field operations, sensor data from a single moment is insufficient to fully reflect the slippage and turning state of wheeled unmanned agricultural machinery. Therefore, this embodiment uses sensor data within a continuous time window to construct the estimation model input. Let the current control cycle be the [number]th [time period]. Each cycle, select the previous one. The vehicle status information within each cycle forms a time-series input sequence: , in, Indicates the first The input feature vector for each control cycle. The input features may include information such as wheel speed, uncompensated desired linear velocity, uncompensated desired angular velocity, attitude angle, angular velocity, linear acceleration, and change in heading angle.

[0054] In practical implementation, the input feature vector can be represented as: , in, Indicates wheel speed information, and These represent the uncompensated desired linear velocity and desired angular velocity obtained in step S3, respectively. and Indicates attitude angle, Indicates the angular velocity of the heading. and Represents linear acceleration in the vehicle coordinate system. This indicates the change in heading angle between adjacent control cycles.

[0055] By constructing the aforementioned time-series input sequence, the estimation model can simultaneously utilize information on wheel speed response, attitude change, inertial response, and control command change, thereby improving its ability to identify slip and turn states in the field.

[0056] S402, Construction of the overall machine slip and rotation estimation model; Specifically, based on the time-series input sequence obtained in step S401, a whole-machine slip and rotation estimation model is constructed. The estimation model includes a longitudinal slip estimation model and a lateral undesired slip estimation model, which can adopt the same or different time-series modeling structures.

[0057] In one implementation, the longitudinal slip estimation model employs a time-series model combining a convolutional neural network and a long short-term memory network to extract wheel speed change, attitude disturbance, and acceleration response features from a continuous sensing sequence, and outputs an estimate of the longitudinal slip of the entire aircraft; the lateral undesired slip estimation model employs a regression model combining a convolutional neural network and a long short-term memory network to extract heading angle change, lateral acceleration, and angular velocity change features, and outputs an estimate of the lateral undesired slip.

[0058] The estimation model can be pre-trained based on field test data. The training samples consist of onboard sensor data collected by the wheeled unmanned agricultural machinery under different soil adhesion conditions, operating speeds, turning radii, and slopes. The training labels can be calculated based on the longitudinal slip parameters and lateral undesired slip parameters defined in step S2. After training, the model is deployed to the onboard computing unit of the wheeled unmanned agricultural machinery for online estimation.

[0059] S403, Online estimation of longitudinal slippage of the entire machine; Specifically, during the operation of wheeled unmanned agricultural machinery, the timing input sequence of the current control cycle is... Input the longitudinal slip estimation model of the whole machine, and obtain the longitudinal slip estimate of the whole machine at the current time. ,in This represents the longitudinal slip estimation model for the entire machine. Indicates the first Estimate of the longitudinal slip of the entire machine in each control cycle.

[0060] The estimated longitudinal slip of the entire machine is used to reflect the deviation between the ideal longitudinal speed calculated from the wheel speed and the actual longitudinal movement speed of the wheeled unmanned agricultural machinery. A larger value indicates poor ground adhesion or significant slippage of the drive wheels, requiring subsequent compensation of the desired linear velocity or drive control quantity.

[0061] S404. Online estimation and result update of undesirable lateral slip; Specifically, while performing longitudinal slip estimation, the same time-series input sequence or an extended time-series input sequence including the heading angle change is input into the lateral undesired slip estimation model to obtain the lateral undesired slip estimate for the current control cycle. ,in This represents the lateral undesirable slip estimation model. Indicates the first The estimator of the lateral undesirable slip for each control cycle.

[0062] Furthermore, to reduce the impact of field vibrations, sensor noise, or transient outliers on control compensation, the estimation results can be smoothed: ， ， in, and These represent the smoothed longitudinal slip estimate and the lateral undesired slip estimate of the entire machine, respectively. This is the smoothing coefficient. The updated estimate is input into the next step for real-time compensation of the trajectory tracking control input.

[0063] S5. Control quantity compensation based on estimation results; It should be noted that in step S5, after obtaining the longitudinal slip estimate and the lateral undesired slip estimate in step S4, this step introduces them into the trajectory tracking control feedback loop to correct the uncompensated control quantity generated in step S3. The purpose of compensation is not simply to reduce slip and rotation itself, but to adjust the linear velocity, angular velocity, or chassis execution commands according to the current wheel-ground interaction state, so that the wheeled unmanned agricultural machinery can still approach the reference trajectory movement under actual field ground conditions.

[0064] Specifically, the longitudinal slip estimate and the lateral undesired slip estimate of the whole machine are obtained and introduced into the trajectory tracking control feedback loop to correct the uncompensated desired linear velocity and desired angular velocity.

[0065] In specific implementation, step S5 also includes steps S501 to S505, which are described in detail below: S501. Determination of longitudinal slip compensation amount; Specifically, let the uncompensated desired linear velocity output in step S3 be... The smoothed longitudinal slip estimate of the whole machine obtained in step S4 is: When a wheeled unmanned agricultural machine slips longitudinally, there is a difference between the ideal speed corresponding to the wheel rotation speed and the actual forward speed of the vehicle body. Therefore, it is necessary to correct the expected linear velocity based on the slip estimation results.

[0066] In this embodiment, the compensated desired linear velocity can be expressed as: , in, This represents the desired linear velocity after longitudinal slip compensation. A positive number is set to prevent the denominator from being too small. Through this compensation method, when the longitudinal velocity loss increases due to soft or slippery field ground, the controller can correct the velocity command based on the slip estimation results, thereby reducing the deviation between the ideal motion model and the actual motion state.

[0067] S502, Determination of lateral undesired slip compensation amount; Specifically, let the uncompensated desired angular velocity output in step S3 be... The lateral undesirable slip estimate obtained in step S4 is: Undesired lateral slippage can cause wheeled unmanned agricultural machinery to produce additional lateral offset or heading deviation. If it continues to operate at the uncompensated angular velocity, it can easily lead to understeer, oversteer, or increased tracking error on curved paths.

[0068] This embodiment corrects the desired angular velocity based on the lateral undesired slip estimate: , in, This represents the desired angular velocity after lateral undesired slip compensation. The equivalent steering characteristic length of the wheeled unmanned agricultural machinery chassis can be determined based on the wheelbase, track width, or equivalent instantaneous steering center position.

[0069] Furthermore, this correction can convert the lateral motion deviation caused by lateral slippage into angular velocity compensation, thereby improving the heading tracking effect during turning, driving on curved paths or slopes.

[0070] S503, generation of compensation control commands; Specifically, after longitudinal slip and lateral undesired slip compensation, the compensation control quantity is obtained: , in, This represents the compensated motion control command for the current control cycle. Depending on the chassis configuration of the wheeled unmanned agricultural machinery, this control quantity can be further converted into corresponding execution commands.

[0071] Furthermore, for chassis with left and right differential drive or four-wheel independent drive, the wheel radius can be adjusted accordingly. and left and right wheelbase Command to calculate the equivalent wheel speed on the left and right sides: ， ， in, and These represent the equivalent wheel angular velocity commands for the left and right sides, respectively.

[0072] Furthermore, for wheeled unmanned agricultural machinery with steering mechanisms, the steering angle command can also be calculated based on the compensated linear velocity and angular velocity, for example: , in, This indicates the corrected steering angle command. Indicates wheelbase.

[0073] S504, Control Constraints and Outputs; It should be noted that, in order to avoid generating excessive control commands under conditions of high slip, strong sideslip, or low adhesion, this embodiment limits the compensated linear velocity, angular velocity, wheel speed, or steering angle: ， ， Furthermore, it can also be based on and The magnitude of the ...

[0074] After the above compensation and constraint processing, the chassis control command is output to adapt to the current field wheel attachment state. This command retains the tracking controller's ability to track the reference trajectory and explicitly considers the overall slip and roll estimation results, thereby improving the trajectory tracking stability of wheeled unmanned agricultural machinery in soft, slippery, sloping, and uneven farmland environments.

[0075] S6, Chassis Execution and Closed-Loop Update; It should be noted that in step S6, after receiving the compensated chassis control command, this step sends it to the wheeled unmanned agricultural machinery chassis execution system, enabling the agricultural machinery to perform trajectory tracking actions according to the corrected linear velocity, angular velocity, wheel speed, or steering angle. Simultaneously, new motion state information is continuously collected during execution, updating the trajectory error and the overall machine slippage and steering estimation results to form real-time closed-loop compensation control.

[0076] Specifically, the compensated chassis control commands are sent to the chassis controller of the wheeled unmanned agricultural machinery, and new motion state information is continuously collected to update the trajectory error and the estimation results of the overall machine slippage and rotation.

[0077] In specific implementation, step S6 also includes steps S601 to S603, which are described in detail below: S601, Compensation control command issued; Specifically, the compensation control command generated in step S5 is sent to the chassis controller of the wheeled unmanned agricultural machinery. For differential drive or four-wheel independent drive chassis, the control command may be the left and right wheel speed command, the speed command of each drive wheel, or the drive motor control command; for wheeled chassis with a steering mechanism, the control command may include the desired vehicle speed, steering angle, steering angular velocity, or driving torque command.

[0078] Furthermore, after receiving the compensation control command, the chassis controller, based on the actuator type of the wheeled unmanned agricultural machinery, drives the motor, steering mechanism, braking mechanism, or hydraulic actuator to complete the corresponding control actions, so that the agricultural machinery travels according to the compensated motion command.

[0079] S602, Execution status feedback collection; During chassis execution, the on-board sensing system continues to collect information such as the position, heading angle, attitude angle, angular velocity, linear acceleration, and wheel speed of the wheeled unmanned agricultural machinery according to the preset control cycle, and simultaneously obtains the execution status fed back by the chassis controller.

[0080] The execution status may include actual wheel speed, actual steering angle, drive motor output status, control command execution deviation, and chassis fault status. Based on this feedback information, it can be determined whether the compensation control command has been effectively executed, and new input data can be provided for trajectory error calculation and slip / steering estimation in the next control cycle.

[0081] S603, Closed-loop control update; Based on the latest collected motion state information, the longitudinal position error, lateral position error, and heading angle error of the wheeled unmanned agricultural machinery relative to the reference trajectory are recalculated. At the same time, the new wheel speed, inertial measurement information, and control commands are input into the overall machine slip and turn estimation model to update the overall machine longitudinal slip estimate and lateral undesired turn estimate.

[0082] Furthermore, a new basic trajectory tracking control quantity is generated based on the updated trajectory error, and compensation is performed by combining the updated slip and turn estimation results to form the chassis control command for the next control cycle.

[0083] Furthermore, through the aforementioned closed-loop update method, wheeled unmanned agricultural machinery can adjust chassis control commands in real time when the field ground adhesion conditions change, the attitude is disturbed, or the slip and turn state changes, thereby reducing the impact of motion model deviation on trajectory tracking accuracy and improving the stability of continuous operation in complex farmland environments.

[0084] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A trajectory tracking control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation, characterized in that, include: S1. Motion state acquisition and trajectory error calculation: Acquire the current motion state information of the wheeled unmanned agricultural machinery, and calculate and output the longitudinal position error, lateral position error and heading angle error in the vehicle coordinate system based on the difference between the actual pose of the wheeled unmanned agricultural machinery and the reference trajectory. S2. Kinematic characterization of the entire machine's sliding and turning; calculate the ideal longitudinal velocity and ideal angular velocity of the wheeled unmanned agricultural machinery under no-slip conditions, and determine the actual longitudinal velocity, lateral velocity, and heading angular velocity in the vehicle coordinate system by combining positioning and inertial measurement information. Based on the difference between the ideal motion state and the actual motion state, define the longitudinal sliding parameters and the lateral undesired turning parameters of the entire machine. The longitudinal slip parameter of the whole machine is used to characterize the longitudinal speed loss of wheeled unmanned agricultural machinery due to the tire-soil interaction. The lateral undesired slip parameter is used to characterize the lateral and heading deviation trends caused by lateral slip, attachment changes, or ground disturbances. S3. Trajectory tracking controller construction: Based on the trajectory error in the vehicle coordinate system and the kinematic characterization of the machine's slip and rotation, a dynamic model of the trajectory error of the wheeled unmanned agricultural machinery is constructed to generate the basic tracking control quantity of the wheeled unmanned agricultural machinery before slip and rotation compensation is performed. S4. Online estimation of overall machine slip and rotation: Based on the control timing information of the wheeled unmanned agricultural machinery in the continuous control cycle, the online estimation model of overall machine slip and rotation is constructed and the estimated longitudinal slip and the estimated lateral undesired rotation of the whole machine under the current control cycle are output. S5. Control quantity compensation based on estimation results: The longitudinal slip estimate and the lateral undesired slip estimate of the whole machine are obtained and introduced into the trajectory tracking control feedback loop to correct the uncompensated desired linear velocity and desired angular velocity; S6. Chassis execution and closed-loop update: The compensated chassis control commands are sent to the wheeled unmanned agricultural machinery chassis controller, and new motion state information is continuously collected to update the trajectory error and the overall machine slip and turn estimation results.

2. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The motion state acquisition and trajectory error calculation also include the following steps: S101. Acquisition of vehicle motion state information: Position and pose information is acquired through an inertial measurement unit, the rotational speed of each wheel is acquired through a wheel speed encoder, and the current linear velocity, angular velocity, or drive control commands are acquired through the chassis controller to construct a position and pose state describing the actual driving position and direction of the agricultural machinery. : ,in, and This represents the position coordinates of the wheeled unmanned agricultural machinery in the global coordinate system. Indicates the heading angle; S102, Multi-source information synchronization and alignment and control state definition; coordinate transformation, outlier removal and smoothing of pose data; conversion and smoothing of wheel speed encoder data to obtain wheel speed information within the current control cycle, and alignment of chassis control commands with sensor data according to timestamps to construct the first... Effective state vector for each control cycle : ,in, Indicates the current pose. Indicates attitude angle, Indicates angular velocity. Indicates linear acceleration. Indicates the wheel speed. Indicates the current chassis control command; S103. Trajectory error calculation and coordinate transformation; obtain the current motion state of the unmanned wheeled agricultural machinery and get its current actual pose. and current reference pose Calculate the pose error of a wheeled unmanned agricultural machine relative to a reference trajectory. The pose error of the reference trajectory The coordinates are then transformed to the vehicle body coordinate system to calculate the longitudinal position error of the current wheeled unmanned agricultural machinery. Lateral position error Heading angle error ; The reference pose ,in and Indicates the position coordinates of the reference trajectory point. Indicates the reference heading angle; The pose error of the reference trajectory ,in, and This represents the positional deviation in the global coordinate system. Indicates the deviation of the heading angle; Based on the control parameters of wheeled unmanned agricultural machinery, the error will be... Further transformation to the vehicle coordinate system Below, the vehicle coordinate system : ,in, This represents the longitudinal positional error along the direction of vehicle movement. This represents the lateral positional error perpendicular to the direction of vehicle movement. This indicates the heading angle error.

3. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The kinematic characterization of the entire machine's sliding and rotating motion also includes the following steps: S201, Calculation of ideal motion state; the calculation of ideal motion state includes left and right wheel speeds, ideal longitudinal speed, and ideal angular velocity; the calculation of left and right wheel speeds involves obtaining the rotational speeds of each drive wheel from the wheel speed encoder and equating the rotational speeds of wheels on the same side to the left wheel speed. and right wheel speed : ， ,in, and These represent the equivalent angular velocities of the left and right wheels, respectively. Indicates the first The angular velocity of each wheel and These represent the number of wheels on the left and right sides involved in the equivalent calculation, respectively. The calculation of the ideal longitudinal velocity and ideal angular velocity is based on the wheel radius. and left and right wheelbase and the speed of the left wheel and right wheel speed Calculate the ideal longitudinal speed of wheeled unmanned agricultural machinery and ideal angular velocity : ， ， in, This represents the ideal longitudinal velocity calculated from the wheel speed under no-slip conditions. Represents the ideal angular velocity under no-slip conditions; S202. Determining the actual motion state; based on the rate of change of the wheeled unmanned agricultural machinery's position in the global coordinate system... and The heading angle is The actual longitudinal velocity in the vehicle coordinate system was calculated. and lateral speed The in, The actual longitudinal velocity along the direction of vehicle movement, the ,in, This represents the actual lateral velocity perpendicular to the direction of the vehicle's movement. S203. Definition of overall machine sliding and rotation parameters; Define the longitudinal sliding parameters of the entire machine based on the ideal motion state and the actual motion state. Deviation from lateral movement Establish a planar kinematic model of the vehicle; The longitudinal slip parameter ,in, A positive number is set to prevent the denominator from being too small at low speeds or when stationary; the aforementioned Used to characterize the degree of loss of actual longitudinal speed of wheeled unmanned agricultural machinery relative to ideal longitudinal speed. The larger the value, the more significant the longitudinal speed loss of the wheeled unmanned agricultural machinery due to the action of wheels on the ground. The deviation of the lateral movement Used to characterize the deviation of the actual lateral motion of wheeled unmanned agricultural machinery from the ideal lateral motion. ,in Represents the lateral velocity under ideal conditions; The construction of the vehicle planar kinematics model , , , ， ， ,in, and This indicates the actual motion state of the wheeled unmanned agricultural machinery.

4. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The construction of the trajectory tracking controller also includes the following steps: S301. Trajectory error dynamics construction; based on the error vector in the vehicle coordinate system obtained in step S1... In addition to the planar kinematics of wheeled unmanned agricultural machinery, a dynamic model of trajectory tracking error is established: ， ， ,in, , and These represent the rates of change of longitudinal error, lateral error, and heading error, respectively. S302. Sliding surface construction; constructing the sliding surface based on longitudinal and lateral position errors. and : ， ,in, and These are the longitudinal and transverse sliding mode variables, respectively. and It is a positive control parameter; The longitudinal sliding mode variable and lateral sliding mode variables When the values ​​approach zero, the longitudinal and lateral position errors can converge, achieving stable tracking of the reference trajectory. S303. Generation of uncompensated control law: Based on the constructed sliding surface, the uncompensated control law is solved according to the error dynamics model and the sliding mode arrival condition. : ,in, Positive control gain, For boundary layer parameters, It is a saturation function; S304, Control Variable Smoothing and Constraint Handling; Based on the Generation of Uncompensated Control Variables For uncompensated expected linear velocity and uncompensated expected angular velocity Perform smoothing and amplitude limiting processing. ， ,in, , The permissible linear velocity range for wheeled unmanned agricultural machinery. , This refers to the permissible range of angular velocities.

5. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The online estimation of the overall machine's sliding and rotation also includes the following steps: S401. Construction of temporal input features; construction of estimation model input from sensor data within a continuous time window. : ,in, Indicates the first The input feature vector for each control cycle; The input features This includes information on wheel speed, uncompensated expected linear velocity, uncompensated expected angular velocity, attitude angle, angular velocity, linear acceleration, and changes in heading angle. S402. Construction of the overall machine sliding and rotation estimation model; based on the obtained time-series input sequence. Construct a sliding and rotation estimation model for the entire machine, the estimation model including a longitudinal sliding estimation model and a lateral undesired rotation estimation model; S403, Online estimation of longitudinal slip of the whole machine; input the timing sequence of the current control cycle. Input the longitudinal slip estimation model of the whole machine, and obtain the longitudinal slip estimate of the whole machine at the current time. : ,in, This represents the longitudinal slip estimation model for the entire machine. Indicates the first Estimated longitudinal slip of the entire machine in one control cycle; The longitudinal slip estimate of the whole machine is used to reflect the deviation between the ideal longitudinal speed calculated from the wheel speed and the actual longitudinal movement speed of the wheeled unmanned agricultural machinery; S404. Online estimation and result update of undesirable lateral slip: Using the same time-series input sequence as the longitudinal slip estimation, the undesirable lateral slip estimation model is input to obtain the estimator of the undesirable lateral slip for the current control cycle. : ,in This represents the lateral undesirable slip estimation model. Indicates the first The estimator of the lateral undesirable slip for each control cycle.

6. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The control quantity compensation based on the estimation results also includes the following steps: S501. Determining the longitudinal slip compensation amount: Based on the uncompensated expected linear velocity output in step S3 and the smoothed longitudinal slip estimate of the whole machine obtained in step S4, the expected linear velocity is corrected to obtain the compensated linear velocity. : ,in, This represents the desired linear velocity after longitudinal slip compensation. A positive number is set to prevent the denominator from being too small; S502. Determination of lateral undesired slip compensation: Based on the uncompensated desired linear velocity output in step S3 and the smoothed longitudinal slip estimate of the entire machine obtained in step S4, the desired angular velocity is corrected to obtain the compensated angular velocity. : ,in, This represents the desired angular velocity after lateral undesired slip compensation. This represents the equivalent steering characteristic length of the wheeled unmanned agricultural machinery chassis. S503, generation of compensation control commands; after longitudinal slip and lateral undesired slip compensation, the compensation control quantity is obtained. : ,in This indicates the motion control command after compensation in the current control cycle; S504, Control Constraints and Outputs; Limiting the compensated linear velocity and angular velocity to output the linear velocity within a reasonable range. and angular velocity .

7. The trajectory tracking and control method for wheeled unmanned agricultural machinery based on whole-machine slip and turn estimation according to claim 1, characterized in that, The chassis execution and closed-loop update also includes the following steps: S601, issue compensation control command; send the compensation control command generated in step S5 to the chassis controller of the wheeled unmanned agricultural machinery, the control command including the desired vehicle speed, steering angle, steering angular velocity or driving torque command; S602, Execution status feedback acquisition; Collect the position, heading angle, attitude angle, angular velocity, linear acceleration and wheel speed information of the wheeled unmanned agricultural machinery according to the preset control cycle, and simultaneously obtain the execution status feedback from the chassis controller; S603, Closed-loop control update: Obtain the latest collected motion state information, recalculate the error information of the wheeled unmanned agricultural machinery relative to the reference trajectory, input the calculation output result into the whole machine slip and turn estimation model, update the whole machine longitudinal slip estimate and lateral undesired slip estimate to generate a new basic trajectory tracking control quantity, combine the updated slip and turn estimation result for compensation, and form the chassis control command for the next control cycle.

Images