An unmanned system and control method for a medium and shallow rhizome medicinal material combined harvester

Abstract

The present application provides a kind of medium and shallow rhizome type medicinal material combined harvester unmanned system and control method, the system is composed of perception system, control system and executing mechanism, integrates binocular camera, Beidou RTK, laser radar and other multi-type sensor to collect multi-source data;First, the data is compensated dynamically for lever error, corrected inversely for three-dimensional point cloud tilt and extracted for ridge center line;Then, based on the differential steering instruction generated by hydraulic resistance correction track slip rate, combined with laser radar feedforward ranging and attitude solution, realize the self-adaptive constant control of digging depth;Resistance feedforward and rotational speed feedback compound power allocation strategy is adopted, and automatic unloading, unbalanced load alarm, hydraulic temperature control and hardware level anti-collision protection mechanism are integrated. Can automatically complete the whole process operation of medicinal material excavation, separation, conveying and collection, significantly improve the precision, efficiency and safety of harvesting operation.

Description

Technical Field

[0001] This invention relates to the field of agricultural machinery automation and intelligent control technology, and in particular to an unmanned driving system and control method for a combine harvester of shallow and medium-sized rhizomes medicinal herbs. Background Technology

[0002] Shallow-rooted medicinal herbs (such as Ophiopogon japonicus, Codonopsis pilosula, and Panax notoginseng) are important raw materials for the traditional Chinese medicine industry. Their planting area is expanding, mainly in hilly and mountainous areas such as Yunnan, Guizhou, Sichuan, and Fujian, where the soil is heavy clay and the plots are small and scattered. The harvesting process has long been plagued by pain points and technical bottlenecks, such as high labor intensity, low operating efficiency, high harvesting costs, poor adaptability of machinery, and low level of automation and intelligence.

[0003] Traditional harvesting relies primarily on manual labor, involving multiple steps such as digging, shaking off the soil, collecting, and transporting—a time-consuming and labor-intensive process. With the increasing shortage of rural labor, traditional manual harvesting severely restricts the large-scale and modern development of the Chinese medicinal herb industry. Existing harvesting machinery almost entirely depends on manual operation by the driver, requiring simultaneous control of multiple parameters such as driving speed, digging depth, and shaking intensity. This makes it difficult to ensure the optimization and consistency of operational parameters, and prolonged fatigue driving further increases operational risks and quality instability. Summary of the Invention

[0004] The purpose of this invention is to provide an unmanned driving system and control method for a combined harvester of shallow and medium-sized rhizomes medicinal materials, so as to automatically complete the combined harvesting process of digging, separating, conveying and collecting shallow and medium-sized rhizomes medicinal materials.

[0005] To achieve the above objectives, the present invention provides the following solution: An unmanned driving system for a combine harvester of shallow and medium-sized rhizomes medicinal herbs includes: a sensing system, a control system, and an actuator; The sensing system includes: a binocular camera, a BeiDou satellite navigation module, an attitude sensor, a lidar, a speed sensor, a displacement sensor, a pressure sensor, a weight sensor, a temperature sensor, a rotary encoder, and a magnetic switch; the sensing system is used to acquire absolute positioning data, geodetic images, 3D depth data, and attitude data; The control system includes: a touch screen, an industrial computer, a programmable logic controller, a closed-loop stepper motor driver, and intermediate relays; the control system is used to perform dynamic compensation of lever arm error, reverse correction of 3D point cloud tilt, evaluation of transient working resistance, theoretical linear velocity correction, and generation of control commands based on data collected by the sensing system; The actuators include: a starter motor, an electromagnetic clutch, a radiator, a closed-loop stepper rod, a throttle controller, an electromagnetic directional valve, a steering hydraulic cylinder, a lifting hydraulic cylinder for the excavating device, and a tilting hydraulic cylinder for the collecting device; the actuators are used to provide feedback responses according to control commands.

[0006] Optionally, the binocular camera is fixed to the front of the harvester with bolts and connected to the industrial control computer via a network cable; the Beidou satellite navigation module is fixed to the top of the harvester with a bracket and connected to the industrial control computer via an industrial serial port. The attitude sensor is located at the centroid of the tracked chassis. The lidar is symmetrically mounted on both sides of the depth limiting wheel of the excavator through anti-vibration brackets. The speed sensor is fixed to the engine and the electromagnetic clutch respectively with bolts. The displacement sensor is located in the internal cavity of the lifting hydraulic cylinder of the excavator and the tilting hydraulic cylinder of the collection device. The pressure sensor is fixed to the lifting hydraulic cylinder of the excavator through a threaded pressure test connector. The weight sensor is fixed to the bottom of the collection box with bolts. The temperature sensor is embedded in the hydraulic oil tank through a threaded interface. All sensors are connected to the programmable logic controller through shielded twisted pair cables. Rotary encoders are symmetrically mounted on the tracked chassis and connected to the track drive wheels via a synchronous belt drive mechanism; magnetic switches are installed at the extension limit position, intermediate reference position, and retraction limit position of all cylinders of the harvester and are connected to the programmable logic controller via shielded twisted pair cables.

[0007] Optionally, the touch screen is connected to the industrial computer via a Type-C cable. The industrial computer is fixed inside the control cabinet via a vibration damping bracket. The programmable logic controller is fixed on a standard guide rail inside the control cabinet. The closed-loop stepper motor driver is fixed inside the control cabinet. The intermediate relay is fixed on a standard guide rail inside the control cabinet.

[0008] A control method for an unmanned driving system of a combine harvester for shallow and medium-sized rhizomes medicinal herbs, based on the aforementioned unmanned driving system for a combine harvester of shallow and medium-sized rhizomes medicinal herbs, includes the following steps: Dynamic compensation for lever error, reverse correction of 3D point cloud tilt, and removal of ridge centerline are performed on the collected absolute positioning data, ridge image, 3D depth data, and attitude data to obtain unified data; The theoretical linear velocity of the track is corrected based on unified data and hydraulic cylinder pressure data to obtain differential steering commands. Based on the differential steering command, spatial kinematics calculations are performed according to the track chassis pitch angle and ground distance data to obtain the feedforward displacement compensation amount, and dynamic adaptive constant control of digging depth is performed in combination with absolute stroke feedback. When the pressure in the hydraulic chamber surges and exceeds the set threshold, feedforward compensation is performed, and the throttle controller is controlled to increase the throttle opening; when the actual engine speed drops sharply, feedback control based on speed deviation and rate of change is performed. The data from the weight sensor is smoothed and filtered, and the off-center load difference is calculated. When the effective total weight is greater than the preset full load threshold, the hydraulic cylinder of the collection device is driven to unload the material. When the off-center load difference is greater than the safety threshold, the system speed reduction alarm is triggered. When the hydraulic oil temperature exceeds the limit, the radiator is forcibly activated to cool down; when the actuator triggers the magnetic switch on the closed-loop stepper push rod, the underlying logic triggers a hardware interrupt and cuts off the closed-loop stepper motor driver signal.

[0009] Optionally, the collected absolute positioning data, ridge image, 3D depth data, and attitude data are subjected to dynamic compensation for lever error, reverse correction of 3D point cloud tilt, and removal of ridge centerline to obtain unified data, including: Extract the normalized RGB components from the ridge-shaped image and calculate the anti-lighting feature index; The ridge-shaped image is binarized using the Otsu method to obtain two-dimensional pixel coordinates. These coordinates are then combined with pixel depth data and camera intrinsic parameters to calculate the three-dimensional spatial coordinates. The formula for calculating the three-dimensional spatial coordinates is as follows: , ;in,( X cL , Y cL , Z cL ) represents the three-dimensional spatial coordinate components of the left furrow in the camera coordinate system. X cR , Y cR , Z cR ) represents the three-dimensional spatial coordinate components of the right furrow in the camera coordinate system. u L , v L ) represents the two-dimensional pixel coordinates of the left furrow edge on the image plane. u R , v R () represents the two-dimensional pixel coordinates of the right furrow edge on the image plane. d This represents the true spatial depth measured by the binocular camera at this pixel. f x , f y They are binocular cameras X shaft and Y Axial equivalent focal length, ( c x , c y () represents the principal pixel coordinates of the optical center of the binocular camera on the image plane; Transform the 3D depth data into the centroid coordinate system of the tracked chassis; The spatial midpoint is calculated based on the transformed 3D spatial coordinates, and outliers are removed using the 3D RANSAC algorithm to obtain the 3D ridge centerline equation; the expression for the 3D ridge centerline equation is: ;in, L ridge This is a 3D tracking path for a local target generated in the chassis centroid coordinate system. P baseL , P baseR These are the three-dimensional spatial coordinate vectors of the left and right furrows, respectively, transformed into the centroid coordinate system of the tracked chassis. Based on the attitude angles fed back by the attitude sensor, coordinate transformation is performed on the BeiDou satellite navigation module; the transformation formula is: ;in, P center This is the three-dimensional spatial coordinate vector of the centroid of the harvester track chassis after projection correction. P GNSS This refers to the absolute positioning three-dimensional spatial coordinate vector obtained by the BeiDou satellite navigation module. R ( i , F , ψ (The pitch angle is the factor) i Roll angle F and heading angle ψ The three-dimensional Euler direction cosine rotation matrix is ​​formed. L arm The phase center of the BeiDou navigation satellite module relative to the centroid origin of the chassis is measured for calibration. O The fixed three-dimensional space lever arm vector; The local target 3D tracking path and the harvester track chassis centroid 3D spatial coordinate vector are unified into the track chassis centroid coordinate system to obtain unified data.

[0010] Optionally, the theoretical linear velocity of the track is corrected based on unified data and hydraulic cylinder pressure data to obtain differential steering commands, including: When the hydraulic oil pressure exceeds the reference value, the dynamic slip compensation coefficient is calculated, and the centerline velocity and yaw rate of the entire machine are corrected accordingly; the correction formula is: ;in, l For dynamic slip compensation coefficient, P act Hydraulic oil pressure, P 0 The calibrated no-load reference hydraulic oil pressure, k This is the drag-slip ratio conversion factor. C This is the soil viscosity compensation constant. V This is the corrected actual linear velocity of the chassis center. ohThis is the corrected actual yaw rate of the entire machine. V L , V R These are the theoretical linear velocities fed back by the left and right track encoders, respectively. B The center distance between the left and right tracks of the harvester; The corrected kinematic data and heading angle are fused using Kalman filtering to obtain the differential steering command.

[0011] Optionally, the formula for calculating the feedforward displacement compensation is: ;in, D Z This represents the actual ground height deviation. H 0 The vertical calibration installation height of the lidar above the ground. d lidar The oblique detection range value obtained by the lidar. i The pitch angle of the harvester obtained by the attitude sensor. Dh This refers to the feedforward displacement compensation amount required for the lifting hydraulic cylinder of the excavator. L The horizontal axis distance from the lidar mounting point to the swing axis of the excavator. D target The target excavation depth is set.

[0012] Optionally, the formula for calculating the engine speed deviation in feedforward compensation is: ;in, U ( t () is the analog control voltage output to the throttle controller. e ( t The real-time speed deviation between the engine target speed and the actual feedback speed. ，K p0 , K i0 , K d0 These are the initial proportional, integral, and derivative gain parameters of the PID controller. K p , K i , K d Fuzzy PID controllers are based on e ( t )and de ( t ) / dt Dynamically output proportional, integral, and derivative gain correction parameters.

[0013] Optionally, the formula for calculating the off-center load difference is: ;in, Wtotal ( t The total instantaneous weight of the medicinal materials collected in the collection box at the current sampling moment is denoted as . w i ( t ) represents the current sampling time. i Instantaneous detection values ​​from each weight sensor W f ( t The effective total weight is the smoothed weight after first-order hysteresis filtering. β These are the first-order hysteresis filter coefficients. W f ( t -1) represents the smoothed effective total weight of the previous sampling period. ΔW This represents the difference in weight between the left and right sides due to eccentric loading. w 1 , w 2 This is the data collected by the weight sensor group on the left. w 3 , w 4 This is the data collected by the weight sensor group on the right.

[0014] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The unmanned driving system for a combined harvester of shallow and medium-sized rhizomes medicinal herbs provided by the present invention performs dynamic compensation for lever error, reverse correction of 3D point cloud tilt, and removal of ridge centerline processing on the collected absolute positioning data, ridge image, 3D depth data, and attitude data to obtain unified data; based on the unified data and hydraulic cylinder pressure data, the theoretical linear velocity of the tracks is corrected to obtain differential steering commands; based on the differential steering commands, spatial kinematics calculations are performed according to the track chassis pitch angle and ground surface distance data to obtain feedforward displacement compensation, and the digging depth is determined by combining absolute stroke feedback. Dynamic adaptive constant control; when the pressure in the hydraulic chamber surges and exceeds a set threshold, feedforward compensation is performed, and the throttle controller is controlled to increase the throttle opening; when the actual engine speed drops sharply, feedback control based on speed deviation and rate of change is implemented; the data from the weight sensor is smoothed and filtered, and the off-center load difference is calculated. When the effective total weight is greater than the preset full-load threshold, the hydraulic cylinder of the collection device is driven to tilt and unload the material; when the off-center load difference is greater than the safety threshold, the system speed reduction alarm is triggered; when the hydraulic oil temperature exceeds the limit, the radiator is forcibly activated to cool down; when the actuator triggers the magnetic switch on the closed-loop stepper push rod, the underlying logic triggers a hardware interrupt and cuts off the closed-loop stepper motor driver signal. This system can automatically complete the combined harvesting process of digging, separating, conveying, and collecting shallow and medium-sized rhizomes. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the overall mechanical structure and sensor layout of an embodiment of the present invention; Figure 2 This is a left view of the overall mechanical structure and sensor layout of an embodiment of the present invention; Figure 3 This is a front view of the collection device according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the electronic and electrical hardware communication topology of the unmanned driving system according to an embodiment of the present invention; Figure 5 This is a flowchart of the control method for the unmanned harvester driving system according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the composite dynamic regulation control principle of anti-caking load resistance feedforward and speed feedback in an embodiment of the present invention.

[0017] Reference numerals: 1. Engine; 2. Collection box; 3. Closed-loop stepper rod; 4. Track; 5. Control cabinet; 6. Excavator lifting hydraulic cylinder; 7. Excavator; 8. LiDAR; 9. Depth limiting wheel; 10. Binocular camera; 11. Radiator; 12. Hydraulic oil tank; 13. Temperature sensor; 14. Beidou satellite navigation module; 15. Attitude sensor; 16. Electromagnetic clutch; 17. Rotary encoder; 18. Track drive wheel; 19. Starter motor; 20. Speed ​​sensor; 21. Weight sensor; 22. Collection device tilting hydraulic cylinder. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1~Figure 4As shown, this invention provides an unmanned driving system for a combined harvester of shallow-rooted medicinal herbs, comprising: a perception system, a control system, and actuators. The perception system includes: a binocular camera 10, a Beidou satellite navigation module 14, an attitude sensor 15, a lidar 8, a speed sensor 20, a displacement sensor, a pressure sensor, a weight sensor 21, a temperature sensor 13, a rotary encoder 17, and a magnetic switch; the control system includes: a touchscreen, an industrial computer, a programmable logic controller, a closed-loop stepper motor driver, and intermediate relays; the actuators include: a starter motor 19, an electromagnetic clutch 16, a radiator 11, a closed-loop stepper push rod 3, a throttle controller, an electromagnetic reversing valve, a steering hydraulic cylinder, a digging device lifting hydraulic cylinder 6, and a collecting device tilting hydraulic cylinder 22.

[0021] In this embodiment, the binocular camera 10 is fixed to the front of the harvester with bolts, and its lens faces the direction of the harvester's movement. It is used to perform binocular visual perception of the working environment in front to identify the edge features of the medicinal herb ridges and simultaneously acquire the three-dimensional depth information of the corresponding ridge edges. The binocular camera 10 is connected to the industrial control computer via a gigabit network cable and transmits the collected two-dimensional image data and three-dimensional depth information to the industrial control computer based on the GigEVision protocol, providing visual and spatial location basis for the accurate planning of local paths.

[0022] The Beidou satellite navigation module 14 (preferably a dual-antenna RTK module) is rigidly fixed to the top of the harvester by a bracket and located directly above the centroid of the tracked chassis. It is used to acquire the absolute latitude and longitude and heading information of the harvester. The Beidou satellite navigation module 14 is connected to the industrial control computer through an industrial serial port and continuously outputs NMEA standard navigation messages using the RS232 protocol. The industrial control computer parses the message in real time and maps the absolute latitude and longitude information into relative position and heading information with the centroid of the tracked chassis as the coordinate origin through lever compensation and coordinate transformation algorithms, thereby aligning with the spatial coordinate system of the vision sensor.

[0023] The attitude sensor 15 is preferably a high-precision inertial measurement unit, which is firmly installed at the centroid of the tracked chassis. It is used to acquire the three-dimensional spatial attitude data of the harvester in complex farmland environments, including pitch angle, roll angle and yaw angle. The attitude sensor 15 is connected to the programmable logic controller through a shielded twisted pair cable and uses the CANopen bus communication protocol to transmit high-frequency attitude data to the programmable logic controller. It serves as the core attitude reference for the dynamic compensation of the arm error of the Beidou satellite navigation module 14 and the three-dimensional point cloud tilt correction of the binocular camera 10, ensuring the attitude fusion accuracy of the harvester in undulating terrain.

[0024] Two lidar radars (8) are configured and symmetrically installed on both sides of the depth-limiting wheel of the digging device via anti-vibration brackets. Their detection surfaces face vertically downward toward the furrow surface to detect the undulations of the ground height in front of the harvester. The lidar radars (8) are connected to the programmable logic controller (PLC) via shielded twisted-pair cables and use the CANopen bus communication protocol to transmit the collected ground distance data to the PLC, providing high-precision distance feedback for dynamic adaptive adjustment of digging depth and ground contour control.

[0025] The speed sensor 20 is configured as two Hall effect gear speed sensors 20, which are respectively bolted to the outside of the flywheel ring gear of the engine 1 and the output sprocket side of the electromagnetic clutch 16. By capturing the pulse frequency generated by the alternating rotation of the gear teeth, the main shaft speed of the engine 1 and the actual output speed after the electromagnetic clutch 16 is engaged are obtained. The speed sensor group 20 is connected to the programmable logic controller via shielded twisted pair cable and adopts the CANopen bus communication protocol to transmit the high-speed sampled speed signal to the programmable logic controller, providing accurate data feedback for the dynamic load monitoring and control algorithm of the harvester.

[0026] The displacement sensors are preferably built-in magnetostrictive displacement sensors, with a total of 3 sensors. They are integrated and installed in the internal cavities of the lifting hydraulic cylinder 6 of the excavating device and the tilting hydraulic cylinder 22 of the collecting device, respectively, to obtain the real-time stroke and absolute extension of the hydraulic cylinder piston rod. The displacement sensors are connected to the programmable logic controller via shielded twisted-pair cables and use the CANopen bus communication protocol to transmit the precise displacement data of the hydraulic cylinder to the programmable logic controller, providing closed-loop feedback signals for the precise lifting of the excavating device 7 and the smooth unloading of the collecting device.

[0027] Two pressure sensors are configured and directly integrated into the pressure ports of the two lifting hydraulic cylinders 6 of the excavating device via threaded pressure test connectors. They are used to obtain instantaneous pressure changes inside the hydraulic system to characterize the physical operating resistance encountered by the excavating device 7 in the soil. The pressure sensors are connected to the programmable logic controller via shielded twisted-pair cables and use the CANopen bus communication protocol to transmit the collected pressure data to the programmable logic controller, providing the most direct soil information sensing source for the resistance feedforward and feedback composite control strategy.

[0028] Five weight sensors 21 are configured and are securely installed at the four corners and the center stress point of the bottom of the collection box 2 by bolts. The multi-point distributed layout effectively eliminates the off-center load error and is used to obtain the weight of the medicinal materials that are constantly accumulating in the collection box 2. The weight sensor group 21 is connected to the programmable logic controller through shielded twisted pair cable and adopts the CANopen bus communication protocol to transmit dynamic load data to the programmable logic controller, providing an accurate quality feedback benchmark for the system's full load warning and automatic unloading decision.

[0029] The temperature sensor 13 is preferably an industrial-grade immersion temperature sensor 13, which is sealed and embedded in the hydraulic oil tank 12 through a threaded interface, and is used to obtain the real oil temperature of the hydraulic system during continuous operation. The temperature sensor 13 is connected to the programmable logic controller through a shielded twisted pair cable and uses the CANopen bus communication protocol to transmit the real-time temperature to the programmable logic controller, providing key data feedback for the thermodynamic state monitoring and heat dissipation control of the hydraulic system.

[0030] Two rotary encoders 17 are configured and symmetrically mounted on the brackets of the left and right track chassis. They are connected to the track drive wheel 18 through a synchronous belt drive mechanism and are used to collect the real-time linear velocity and rotation angle of the left and right tracks 4. The rotary encoder 17 group is connected to the programmable logic controller through shielded twisted pair cables and adopts the CANopen bus communication protocol to transmit the feedback data of the two track tracks 4 to the programmable logic controller, providing accurate data support for the differential steering control and straight-line driving correction of the harvester.

[0031] Three magnetic switches are configured and arranged along the movement trajectory of the closed-loop stepper rod 3 at the extension limit position, intermediate reference position, and retraction limit position of the cylinder body, respectively. They accurately identify the mechanical position of the push rod by capturing the magnetic field changes of the internal piston. The magnetic switches are connected to the programmable logic controller via shielded twisted-pair cables and use a switch signal transmission method to feed back the stroke status of the push rod to the programmable logic controller, providing a physical limit basis for the precise control of the actuator and preventing the mechanical structure from overtravel collision.

[0032] In this embodiment, the touch screen is configured as an industrial-grade touch display device, which is firmly embedded in the control cabinet 5 panel and serves as the human-machine interaction terminal for the entire autonomous driving system. The touch screen is directly connected to the industrial computer via a Type-C cable, and the DP alternating mode of the Type-C interface is used to realize the "one-line" bidirectional transmission of high-definition video signals and touch operation signals. This is used to assist operators in scheduling and running the autonomous driving program, quickly setting operation parameters, and visually monitoring the data of the entire machine's perception system.

[0033] The industrial control computer is securely mounted in the control cabinet 5 using a vibration damping bracket, serving as the high-performance decision-making center for the overall control system. At the perception and decision-making end, the industrial control computer acquires image depth information from the binocular camera 10 and positioning and orientation data from the Beidou navigation module via the GigE Vision protocol and RS232 protocol, respectively. It then combines the attitude data to perform spatial coordinate transformation and multimodal information fusion, thereby planning a high-precision local tracking trajectory. At the interaction end, the industrial control computer provides a visual system monitoring interface for the touch screen via a Type-C interface. At the control and coordination end, the industrial control computer establishes high-speed bidirectional communication with the programmable logic controller (PLC) via the TCP / IP protocol, acquires mechanical condition data uploaded by various sensors in real time, and after comprehensive calculation, issues precise global control commands such as target heading, working depth, and vehicle speed to the PLC.

[0034] The programmable logic controller (PLC) is fixed on a standard guide rail inside control cabinet 5 and serves as the underlying control and signal processing layer of the autonomous driving system. At the sensing aggregation end, the PLC collects high-frequency data from sensors such as speed, pressure, displacement, and attitude in real time via the CANopen bus communication protocol, and performs preliminary filtering and data encapsulation. At the communication logic end, the PLC acts as a bridge between the industrial control computer and the actuators, uploading the aggregated sensor data to the industrial control computer via the TCP / IP protocol and receiving control commands issued by the industrial control computer. At the execution drive end, the PLC has built-in PID control and resistance feedforward compensation programs, which precisely control the throttle controller, stepper motor driver, and solenoid directional valve through analog signals, pulse signals, and switch signals to ensure the smoothness and response speed of the actuators.

[0035] The closed-loop stepper motor driver is securely installed inside control cabinet 5, serving as the dedicated power execution unit for the closed-loop stepper actuator 3. The driver connects to the high-speed I / O port of the programmable logic controller (PLC) to receive precise pulse and direction control signals and output phase current to the motor of the closed-loop stepper actuator 3. It should be noted that during operation, the driver reads the actual mechanical angle signal fed back from the encoder at the tail of the closed-loop stepper actuator 3 motor in real time. Through an internal vector control algorithm, it performs instantaneous closed-loop compensation for step loss caused by sudden changes in field load, ensuring absolute accuracy and smoothness in gear shifting. Furthermore, the position closed-loop logic built into the driver, in conjunction with the limit signal of the magnetic switch installed on the cylinder of the closed-loop stepper actuator 3, constructs a dual redundancy safety mechanism of "software compensation + hardware limit" for the actuator.

[0036] Ten intermediate relays are configured and uniformly installed on standard guide rails inside control cabinet 5. They serve as electrical isolation and power amplification units between the programmable logic controller and the high-power actuator. The intermediate relays receive switching control commands from the programmable logic controller I / O ports and use the closing and opening of their internal contacts to precisely control the operation of high-current power circuits such as engine 1 starting, electromagnetic clutch 16 engaging, electromagnetic reversing valve switching, and radiator 11 starting and stopping, ensuring physical isolation between the underlying control signals and the power execution circuit.

[0037] In this embodiment, under the unified scheduling of the programmable logic controller, the actuator, driven by the closed-loop stepper motor driver, intermediate relay and analog output module, collaboratively completes the start and stop of the engine 1, the dynamic adjustment of the walking speed, the precise tracking of the steering path, the adaptive adjustment of the digging depth and the automatic flipping and unloading of the medicinal material collection box 2, thereby realizing the fully automated joint harvesting operation of shallow and medium-sized rhizomes from digging, separation, transportation to collection.

[0038] like Figure 5 As shown, the present invention also provides a control method for an unmanned driving system of a combine harvester for shallow and medium-sized rhizomes medicinal herbs, which is based on the above-mentioned unmanned driving system for a combine harvester for shallow and medium-sized rhizomes medicinal herbs, and includes the following steps: Step 100: Perform dynamic compensation for lever error, reverse correction of 3D point cloud tilt, and removal of ridge centerline on the collected absolute positioning data, ridge image, 3D depth data, and attitude data to obtain unified data; Step 200: Based on unified data and hydraulic cylinder pressure data, the theoretical linear speed of the track is corrected to obtain the differential steering command; Step 300: Based on the differential steering command, perform spatial kinematics calculations according to the track chassis pitch angle and ground distance data to obtain the feedforward displacement compensation amount, and combine it with absolute stroke feedback to perform dynamic adaptive constant control of the digging depth. Step 400: When the pressure in the hydraulic chamber surges and exceeds the set threshold, feedforward compensation is performed, and the throttle controller is controlled to increase the throttle opening; when the actual engine speed drops sharply, feedback control based on speed deviation and rate of change is performed. Step 500: Smooth the data from the weight sensor and calculate the off-center load difference. When the effective total weight is greater than the preset full load threshold, drive the hydraulic cylinder of the collection device to unload the material. When the off-center load difference is greater than the safety threshold, trigger the system speed reduction alarm. Step 600: When the hydraulic oil temperature exceeds the limit, the radiator is forcibly started to cool down; when the actuator triggers the magnetic switch on the closed-loop stepper push rod, the underlying logic triggers a hardware interrupt and cuts off the closed-loop stepper motor driver signal.

[0039] It should be noted that, in order to achieve accurate fusion and kinematic calculation of multi-source sensing data, this embodiment first establishes a whole-machine rectangular coordinate system O-XYZ with the geometric centroid of the harvester track chassis as the origin O: the Y-axis points to the forward direction of the harvester, the X-axis points to the right of the forward direction of the harvester, and the Z-axis is perpendicular to the horizontal surface of the track chassis and pointing downwards. Based on this coordinate system, the three-dimensional spatial attitude angles are uniformly defined in the system control logic as follows: pitch angle i (Rotation around the X-axis): Positive for the front of the car lifting up, negative for it tilting down; Roll angle F (Rotation around the Y-axis): The right side of the aircraft is lower, which is positive; the left side is lower, which is negative; yaw angle ψ (Rotation around the Z-axis): Right rotation is positive, left rotation is negative.

[0040] In the specific implementation process, the industrial control computer in step 100 acquires in real time the absolute positioning data of the Beidou satellite navigation module, the ridge image and 3D depth data of the binocular camera, and the attitude data transmitted by the programmable logic controller. The industrial control computer uses the attitude data to dynamically compensate for the lever arm error of the Beidou satellite navigation module and performs inverse tilt correction on the 3D point cloud of the binocular camera, extracting the smooth ridge centerline and unifying the heterogeneous sensing data into a local coordinate system with the centroid of the tracked chassis as the origin. Specifically, when processing the 2D image data acquired by the binocular camera, the industrial control computer extracts the normalized RGB components to calculate the anti-lighting characteristic index. I The calculation formula is: ; in, I It is a light resistance characteristic index. r , g , b These represent the normalized red, green, and blue channel components of an image pixel, respectively. α These are the illumination compensation coefficients dynamically calculated based on the global histogram of the image.

[0041] Next, the feature index map is binarized using Otsu's method to extract the two-dimensional pixel coordinates of the left and right furrow edges. u L , v L )and( u R , v R This is combined with the depth data of the corresponding pixels synchronously output by the binocular cameras. d The three-dimensional spatial coordinates of the left and right furrows in the camera coordinate system are calculated using the camera intrinsic parameter model. P cL ( X cL , Y cL , ZcL )and P cR ( X cR , Y cR , Z cR The calculation formula is: ; ; in,( X cL , Y cL , Z cL ) represents the three-dimensional spatial coordinate components of the left furrow in the camera coordinate system. X cR , Y cR , Z cR ) represents the three-dimensional spatial coordinate components of the right furrow in the camera coordinate system. u L , v L ) represents the two-dimensional pixel coordinates of the left furrow edge on the image plane. u R , v R () represents the two-dimensional pixel coordinates of the right furrow edge on the image plane. d This represents the true spatial depth measured by the binocular camera at this pixel. f x , f y They are binocular cameras X shaft and Y Axial equivalent focal length, ( c x , c y () represents the principal pixel coordinates of the optical center of the binocular camera on the image plane.

[0042] To achieve spatial unification of multi-source sensing data, the industrial control computer then transformed the 3D point cloud of furrows in the camera coordinate system to the centroid coordinate system of the tracked chassis. O - XYZ Below. Set the mounting angle of the stereo camera to [value missing]. c c Its relative to the origin of the chassis centroid O The translation vector is T cam The transformed centroid coordinates of the left and right furrows P baseL andP baseR The calculation is as follows: ; in, P baseL , P baseR These are the three-dimensional spatial coordinate vectors of the left and right furrows, respectively, transformed into the centroid coordinate system of the tracked chassis. R cam ( c c The angle between the binocular camera and the horizontal plane (mounting angle) is the angle of attack. c c The pitch and rotation matrix is ​​composed of ) T cam The origin of the optical center of the binocular camera to the centroid of the tracked chassis. O A fixed three-dimensional translation vector.

[0043] Then, based on the transformed three-dimensional spatial coordinates of the left and right furrows, the spatial midpoint is calculated, and the three-dimensional RANSAC algorithm is used to remove outliers and noise points, thus fitting a smooth three-dimensional ridge centerline equation. L ridge The expression is: ; in, L ridge This is a 3D tracking path for a local target generated in the chassis centroid coordinate system.

[0044] Simultaneously, using the attitude angles fed back by the attitude sensor, coordinate transformation is performed on the BeiDou satellite navigation module. The transformation formula is as follows: ; in, P center This is the three-dimensional spatial coordinate vector of the centroid of the harvester track chassis after projection correction. P GNSS This refers to the absolute positioning three-dimensional spatial coordinate vector obtained by the BeiDou satellite navigation module. R ( i , F , ψ (The pitch angle is the factor) i Roll angle F and heading angle ψ The three-dimensional Euler direction cosine rotation matrix is ​​formed. L arm The phase center of the BeiDou navigation satellite module relative to the centroid origin of the chassis is measured for calibration. O The fixed three-dimensional space arm vector is then used. Finally, through the aforementioned multi-matrix transformation algorithm, the local visual planning path is transformed. L ridgeGlobal absolute positioning P center Unified to a single tracked chassis centroid coordinate system O - XYZ This fundamentally eliminates navigation deviations caused by sensor heterogeneity and vehicle shaking.

[0045] In the specific implementation process, step 200 addresses track slippage caused by loose soil in the field. The programmable logic controller (PLC) acquires real-time pressure data from the lifting hydraulic cylinder of the excavator to assess transient operating resistance and dynamically generates a slip compensation coefficient based on this resistance. The industrial control computer uses this slip compensation coefficient to correct the theoretical linear velocity of the left and right track rotary encoders, calculates the actual chassis linear velocity and yaw rate, and integrates the sensing data from step 100 to output differential steering commands to control the tracks to accurately track the local path. Specifically, in soft soil, the PLC monitors the hydraulic oil pressure of the pressure sensor in real time. If the pressure exceeds the reference value, it determines that a high-resistance zone has been entered and calculates the dynamic slip compensation coefficient. l And calculate the actual centerline velocity of the entire machine. V and yaw rate oh The calculation formula is: ; in, l For dynamic slip compensation coefficient, P act The pressure sensor acquires in real time the hydraulic oil pressure of the lifting hydraulic cylinder of the excavator. P 0 The calibrated no-load reference hydraulic oil pressure, k This is the drag-slip ratio conversion factor. C This is the soil viscosity compensation constant. V This is the corrected actual linear velocity of the chassis center. oh This is the corrected actual yaw rate of the entire machine. V L , V R These are the theoretical linear velocities fed back by the left and right track encoders, respectively. B This is the center distance between the left and right tracks of the harvester. The industrial control computer will then compare the corrected kinematic data with the heading angle. ψ Kalman filtering is performed to fuse the signals and output high-precision differential steering commands.

[0046] In the specific implementation process, step 300 involves the programmable logic controller (PLC) acquiring real-time ground distance data scanned by the lidar during operation. This data, combined with the pitch angle of the tracked chassis fed back by the attitude sensor, is used to perform spatial kinematic calculations to eliminate distance measurement errors caused by harvester vibrations. The PLC also calculates the required feedforward displacement compensation for the digging device's lifting hydraulic cylinder. Combined with the absolute stroke feedback from the displacement sensor, a closed-loop drive is used to operate the digging device's lifting hydraulic cylinder, achieving dynamic adaptive constant control of the digging depth. Specifically, the lidar is installed on both sides of the depth-limiting wheels of the digging device. Z In a coordinate system with the axis pointing downwards, the actual ground elevation deviation ΔZ And the target feedforward displacement compensation amount required for the lifting hydraulic cylinder of the excavating device Dh The calculation formula is: ; in, ΔZ This represents the actual ground height deviation. H 0 The vertical calibration installation height of the lidar above the ground. d lidar The oblique detection range value obtained by the lidar. i The pitch angle of the harvester obtained by the attitude sensor. Dh The target feedforward displacement compensation amount required for the lifting hydraulic cylinder of the excavator. L The horizontal axis distance from the lidar mounting point to the swing axis of the excavator. D target The programmable logic controller (PLC) drives the electromagnetic directional valve in real time based on the calculation results to ensure that the digging shovel penetrates the soil at an absolutely constant depth in undulating terrain, thus ensuring that the target digging depth is set.

[0047] like Figure 6 As shown, in the specific implementation process, during constant-depth excavation in step 400, when the pressure sensor first detects a surge in pressure within the hydraulic chamber exceeding a set threshold, the programmable logic controller triggers feedforward compensation, proactively controls the throttle controller, and increases the throttle opening. If the platen resistance causes a sudden drop in the actual engine speed monitored by the speed sensor, feedback control based on speed deviation and rate of change is superimposed to further enhance engine power output, thereby achieving composite power distribution and preventing stalling under heavy load. Specifically, when the hydraulic oil pressure mutation rate exceeds the threshold, the feedforward channel is triggered to proactively increase the throttle opening. Simultaneously, the feedback channel calculates the engine speed deviation and outputs the throttle control voltage to achieve power response and stall prevention control under large load mutations. The calculation formula is as follows: ; in, U ( t () is the analog control voltage output to the throttle controller.e ( t This represents the real-time speed deviation between the engine target speed and the actual feedback speed. de ( t ) / dt Real-time rate of change of engine speed deviation ，K p0 , K i0 , K d0 These are the initial proportional, integral, and derivative gain parameters of the PID controller. K p , K i , K d Fuzzy PID controllers are based on e ( t )and de ( t ) / dt Dynamically output proportional, integral, and derivative gain correction parameters.

[0048] In the specific implementation process, step 500 addresses the issues of field bumps and uneven loading of medicinal herbs. The programmable logic controller (PLC) acquires data from multiple weight sensor groups in real time and performs smoothing filtering and uneven loading difference calculation. When the effective total weight exceeds a preset full-load threshold, the collection device's tilting hydraulic cylinder is automatically driven to unload the material. When the uneven loading difference exceeds a safety threshold, a system deceleration alarm is triggered to prevent the center of gravity from shifting. Specifically, the formulas for smoothing filtering and uneven loading calculation are: ; in, W total ( t The total instantaneous weight of the medicinal materials collected in the collection box at the current sampling moment is denoted as . w i ( t ) represents the current sampling time. i Instantaneous detection values ​​of each weight sensor ( i =1, 2, ..., 5), W f ( t The effective total weight is the smoothed weight after first-order hysteresis filtering. β These are the first-order hysteresis filter coefficients. W f ( t -1) represents the smoothed effective total weight of the previous sampling period. ΔW This represents the difference in weight between the left and right sides due to eccentric loading. w 1 , w 2This is the data collected by the weight sensor group on the left. w 3 , w 4 This is the data collected by the weight sensor group on the right. When the effective total weight... W f ( t When the load difference exceeds the preset full load threshold, the programmable logic controller automatically drives the collecting device to tilt the hydraulic cylinder to unload the material; when the load difference is greater than the preset full load threshold, the programmable logic controller automatically drives the collecting device to tilt the hydraulic cylinder to unload the material. ΔW When the speed exceeds the safety threshold, a system speed reduction alarm is triggered.

[0049] In the specific implementation process, during the entire operation, step 600 will forcibly start the radiator to cool down when the temperature sensor detects that the hydraulic oil temperature exceeds the limit; when the actuator triggers the magnetic control switch on the closed-loop stepper push rod, the underlying logic triggers a hardware interrupt and cuts off the closed-loop stepper motor driver signal, forming an insurmountable mechanical anti-collision bottom line to achieve mechanical anti-collision limit protection for the whole machine.

[0050] The beneficial effects of this invention are as follows: 1. A resistance assessment mechanism based on transient perception of hydraulic pressure in the lifting hydraulic cylinder of the excavating device was introduced, which dynamically corrected the track slip ratio. At the same time, the attitude sensor was combined to perform spatial lever compensation for the Beidou satellite navigation module and tilt correction for the binocular visual point cloud. This completely eliminated the false displacement error caused by complex terrain and slippage, and achieved extremely high-precision path tracking between complex medicinal herb ridges. It broke through the bottleneck of navigation failure and trajectory deviation in soft medicinal herb fields. 2. By coupling the laser radar feedforward ranging with the pitch angle depth of the tracked chassis, the artifacts caused by the tilt of the harvester are eliminated through spatial kinematics calculation, and the true ground height undulation is obtained. In addition, with the absolute displacement sensor inside the hydraulic cylinder, the absolute adaptive constant depth of the digging shovel is achieved, avoiding deep digging that damages the roots or shallow digging that misses the harvest. 3. By using a composite power distribution control strategy of "resistance feedforward + speed feedback", the pressure sensor captures the transient surge of hydraulic pressure as a feedforward signal, and the speed drop rate is used as a fuzzy PID feedback signal for compensation. This achieves coordinated power response and track speed reduction under large load changes, and greatly improves the system's resistance to load impact. 4. Hardware-level protection mechanisms are set up at critical nodes: unloading is automatically triggered by a multi-point weight sensor group; the temperature sensor is linked to the radiator to ensure the thermodynamic stability of the hydraulic system; magnetic switches are deployed at the extreme positions of the actuator as the highest priority hardware interrupt signal, forming a double anti-collision protection with the software closed loop of the closed-loop stepper actuator, which greatly improves the overall mechanical safety of the unmanned operation; all sensors of the machine uniformly adopt shielded twisted pair cables and the CANopen bus communication protocol with strong anti-interference, resisting electromagnetic noise from the engine, etc.

[0051] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0052] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An unmanned driving system for a combine harvester of shallow-rooted medicinal herbs, characterized in that, include: Sensing systems, control systems, and actuators; The sensing system includes: a binocular camera, a BeiDou satellite navigation module, an attitude sensor, a lidar, a rotation speed sensor, a displacement sensor, a pressure sensor, a weight sensor, a temperature sensor, a rotary encoder, and a magnetic switch; the sensing system is used to acquire absolute positioning data, morphological images, three-dimensional depth data, and attitude data; The control system includes: a touch screen, an industrial computer, a programmable logic controller, a closed-loop stepper motor driver, and intermediate relays; the control system is used to perform dynamic compensation of lever arm error, reverse correction of three-dimensional point cloud tilt, evaluation of transient working resistance, correction of theoretical linear velocity, and generation of control commands based on the data collected by the sensing system. The actuator includes: a starter motor, an electromagnetic clutch, a radiator, a closed-loop stepper rod, a throttle controller, an electromagnetic directional valve, a steering hydraulic cylinder, a lifting hydraulic cylinder for the excavating device, and a tilting hydraulic cylinder for the collecting device; the actuator is used to provide feedback response according to the control command.

2. The unmanned driving system for the combined harvester of shallow and medium-sized rhizomes medicinal materials according to claim 1, characterized in that, The binocular camera is fixed to the front of the harvester with bolts and connected to the industrial control computer via a network cable; the Beidou satellite navigation module is fixed to the top of the harvester with a bracket and connected to the industrial control computer via an industrial serial port. The attitude sensor is located at the centroid of the tracked chassis. The lidar is symmetrically mounted on both sides of the depth-limiting wheel of the excavating device via anti-vibration brackets. The speed sensor is fixed to the engine and the electromagnetic clutch respectively by bolts. The displacement sensor is located in the internal cavity of the lifting hydraulic cylinder of the excavating device and the tilting hydraulic cylinder of the collecting device. The pressure sensor is fixed to the lifting hydraulic cylinder of the excavating device via a threaded pressure test connector. The weight sensor is fixed to the bottom of the collecting box by bolts. The temperature sensor is embedded in the hydraulic oil tank via a threaded interface. All sensors are connected to the programmable logic controller via shielded twisted-pair cables. The rotary encoder is symmetrically mounted on the tracked chassis and connected to the track drive wheel via a synchronous belt transmission mechanism; the magnetic control switch is installed at the extension limit position, intermediate reference position and retraction limit position of all cylinders of the harvester and is connected to the programmable logic controller via shielded twisted pair cables.

3. The unmanned driving system for the combined harvester of shallow and medium-sized rhizomes medicinal materials according to claim 1, characterized in that, The touch screen is connected to the industrial computer via a Type-C cable. The industrial computer is fixed inside the control cabinet by a vibration damping bracket. The programmable logic controller is fixed on a standard guide rail inside the control cabinet. The closed-loop stepper motor driver is fixed inside the control cabinet. The intermediate relay is fixed on a standard guide rail inside the control cabinet.

4. A control method for an unmanned driving system of a combine harvester for shallow-rooted medicinal herbs, implemented based on the unmanned driving system for a combine harvester of shallow-rooted medicinal herbs as described in any one of claims 1-3, characterized in that... Includes the following steps: Dynamic compensation for lever error, reverse correction of 3D point cloud tilt, and removal of ridge centerline are performed on the collected absolute positioning data, ridge image, 3D depth data, and attitude data to obtain unified data; Based on the unified data and hydraulic cylinder pressure data, the theoretical linear velocity of the track is corrected to obtain the differential steering command. Based on the differential steering command, spatial kinematics calculations are performed according to the track chassis pitch angle and ground distance data to obtain the feedforward displacement compensation amount, and dynamic adaptive constant control of the digging depth is performed in combination with absolute stroke feedback. When the pressure in the hydraulic chamber surges and exceeds the set threshold, feedforward compensation is performed, and the throttle controller is controlled to increase the throttle opening; when the actual engine speed drops sharply, feedback control based on speed deviation and rate of change is performed. The data from the weight sensor is smoothed and filtered, and the off-center load difference is calculated. When the effective total weight is greater than the preset full load threshold, the hydraulic cylinder of the collection device is driven to unload the material. When the off-center load difference is greater than the safety threshold, the system speed reduction alarm is triggered. When the hydraulic oil temperature exceeds the limit, the radiator is forcibly activated to cool down; when the actuator triggers the magnetic switch on the closed-loop stepper push rod, the underlying logic triggers a hardware interrupt and cuts off the closed-loop stepper motor driver signal.

5. The control method for the unmanned driving system of the combined harvester for shallow and medium-sized rhizomes medicinal materials according to claim 4, characterized in that, The collected absolute positioning data, ridge images, 3D depth data, and attitude data were processed using dynamic compensation for lever errors, reverse correction of 3D point cloud tilt, and removal of ridge centerlines to obtain unified data, including: Extract the normalized RGB components from the ridge-shaped image and calculate the anti-lighting feature index; The ridge-shaped image is binarized using the Otsu method to obtain two-dimensional pixel coordinates, and then three-dimensional spatial coordinates are calculated by combining pixel depth data and camera intrinsic parameter models; the formula for calculating the three-dimensional spatial coordinates is as follows: , ;in,( X cL , Y cL , Z cL ) represents the three-dimensional spatial coordinate components of the left furrow in the camera coordinate system. X cR , Y cR , Z cR ) represents the three-dimensional spatial coordinate components of the right furrow in the camera coordinate system. u L , v L ) represents the two-dimensional pixel coordinates of the left furrow edge on the image plane. u R , v R () represents the two-dimensional pixel coordinates of the right furrow edge on the image plane. d This represents the true spatial depth measured by the binocular camera at this pixel. f x , f y They are binocular cameras X shaft and Y Axial equivalent focal length, ( c x , c y () represents the principal pixel coordinates of the optical center of the binocular camera on the image plane; The three-dimensional depth data is converted to the centroid coordinate system of the tracked chassis; The spatial midpoint is calculated based on the transformed three-dimensional spatial coordinates, and outliers are removed using the three-dimensional RANSAC algorithm to obtain the three-dimensional ridge centerline equation; the expression of the three-dimensional ridge centerline equation is: ;in, L ridge This is a 3D tracking path for a local target generated in the chassis centroid coordinate system. P baseL , P baseR These are the three-dimensional spatial coordinate vectors of the left and right furrows, respectively, transformed into the centroid coordinate system of the tracked chassis. Based on the attitude angles fed back by the attitude sensor, coordinate transformation is performed on the BeiDou satellite navigation module; the transformation formula is: ;in, P center This is the three-dimensional spatial coordinate vector of the centroid of the harvester track chassis after projection correction. P GNSS This refers to the absolute positioning three-dimensional spatial coordinate vector obtained by the BeiDou satellite navigation module. R ( θ , Φ , ψ (The pitch angle is the factor) θ Roll angle Φ and heading angle ψ The three-dimensional Euler direction cosine rotation matrix is ​​formed. L arm The phase center of the BeiDou navigation satellite module relative to the centroid origin of the chassis is measured for calibration. O The fixed three-dimensional space lever arm vector; The three-dimensional tracking path of the local target and the three-dimensional spatial coordinate vector of the centroid of the harvester track chassis are unified into the centroid coordinate system of the track chassis to obtain the unified data.

6. The control method for the unmanned driving system of the combined harvester for shallow and medium-sized rhizomes medicinal materials according to claim 4, characterized in that, Based on the unified data and hydraulic cylinder pressure data, the theoretical linear velocity of the track is corrected to obtain differential steering commands, including: When the hydraulic oil pressure exceeds the reference value, the dynamic slip compensation coefficient is calculated, and the centerline velocity and yaw rate of the entire machine are corrected accordingly; the correction formula is: ;in, λ For dynamic slip compensation coefficient, P act Hydraulic oil pressure, P 0 The calibrated no-load reference hydraulic oil pressure, k This is the drag-slip ratio conversion factor. C This is the soil viscosity compensation constant. V This is the corrected actual linear velocity of the chassis center. ω This is the corrected actual yaw rate of the entire machine. V L , V R These are the theoretical linear velocities fed back by the left and right track encoders, respectively. B The center distance between the left and right tracks of the harvester; The corrected kinematic data and heading angle are fused using Kalman filtering to obtain the differential steering command.

7. The control method for the unmanned driving system of the combined harvester for shallow and medium-sized rhizomes medicinal materials according to claim 4, characterized in that, The formula for calculating the feedforward displacement compensation is as follows: ;in, ΔZ This represents the actual ground height deviation. H 0 The vertical calibration installation height of the lidar above the ground. d lidar The oblique detection range value obtained by the lidar. θ The pitch angle of the harvester obtained by the attitude sensor. Δh This refers to the feedforward displacement compensation amount required for the lifting hydraulic cylinder of the excavator. L The horizontal axis distance from the lidar mounting point to the swing axis of the excavator. D target The target excavation depth is set.

8. The control method for the unmanned driving system of the combined harvester for shallow and medium-sized rhizomes medicinal materials according to claim 4, characterized in that, The formula for calculating engine speed deviation in feedforward compensation is: ;in, U ( t () is the analog control voltage output to the throttle controller. e ( t The real-time speed deviation between the engine target speed and the actual feedback speed. ，K p0 , K i0 , K d0 These are the initial proportional, integral, and derivative gain parameters of the PID controller. K p , K i , K d Fuzzy PID controllers are based on e ( t )and de ( t ) / dt Dynamically output proportional, integral, and derivative gain correction parameters.

9. The control method for the unmanned driving system of the combined harvester for shallow and medium-sized rhizomes medicinal materials according to claim 4, characterized in that, The formula for calculating the off-center load difference is: ; in, W total ( t The total instantaneous weight of the medicinal materials collected in the collection box at the current sampling moment is denoted as . w i ( t ) represents the current sampling time. i Instantaneous detection values ​​from each weight sensor W f ( t The effective total weight is the smoothed weight after first-order hysteresis filtering. β These are the first-order hysteresis filter coefficients. W f ( t -1) represents the smoothed effective total weight of the previous sampling period. ΔW This represents the difference in weight between the left and right sides due to eccentric loading. w 1 , w 2 This is the data collected by the weight sensor group on the left. w 3 , w 4 This is the data collected by the weight sensor group on the right.

Images