An Adaptive Control Method for Operating Parameters of an RTK-based Autonomous Obstacle Avoidance Yam Harvester

By using a multi-source sensing data acquisition and adaptive control method for an RTK-navigated autonomous obstacle avoidance yam harvester, the problems of operating depth and obstacle avoidance coordination in complex field conditions were solved, and real-time adjustment of operating parameters and stability of harvesting quality were achieved.

CN123113398APending Publication Date: 2026-11-17聊城大学东昌学院
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611595151.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-15
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing yam harvesters are unable to adapt to the dynamic changes in ridge conditions under complex field conditions. The coordination between driving and operating parameters is insufficient in working depth and obstacle avoidance conditions, resulting in damage to tubers and missed harvesting.

Method used

An autonomous obstacle avoidance yam harvester based on RTK navigation is adopted. Through multi-source operation perception data acquisition, autonomous navigation obstacle avoidance trajectory planning and working condition segmentation, adaptive matching decision-making of operation parameters and collaborative control execution, adaptive closed-loop control of operation parameters is achieved. It adapts to changes in ridge elevation and coordinates the adjustment of digging depth, travel speed and conveyor chain speed.

Benefits of technology

It enables real-time adjustment of the operating depth of yam harvesters under complex field conditions, reducing the probability of tuber damage and missed harvesting, and ensuring the accuracy and continuity of harvesting operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113398A_ABST
    Figure CN123113398A_ABST
Patent Text Reader

Abstract

This invention relates to the field of intelligent control technology for agricultural harvesting machinery, specifically to an adaptive control method for operating parameters of a yam harvester based on RTK navigation and autonomous obstacle avoidance. The method includes the following steps: multi-source operational sensing data acquisition; autonomous navigation obstacle avoidance trajectory planning and operational condition segmentation; adaptive matching decision-making of operating parameters, coordinating the adjustment of digging depth, travel speed, and conveyor chain speed based on the relative ridge operating depth benchmark; and coordinated control execution of operating parameters. This invention achieves dynamic adaptation of operating depth to field conditions by inverting the ridge benchmark drift based on digging resistance and shovel geometry parameters in stable straight operating sections, and correcting the depth benchmark by combining heading deviation. This manages machine load, reduces tuber damage and missed harvesting, and improves harvesting accuracy. By segmenting operational conditions, identifying and coordinating operating parameters, and executing coordinated control of the entire vehicle, it avoids conveyor accumulation and depth fluctuations, ensuring operational continuity and stable harvesting quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for agricultural harvesting machinery, and more specifically, to an adaptive control method for operating parameters of a yam harvester based on RTK navigation and autonomous obstacle avoidance. Background Technology

[0002] Mechanized harvesting of root and tuber crops is a key link in improving the quality and efficiency of the yam planting industry. With the continuous improvement of the scale of yam planting in China, self-propelled yam harvesting equipment has become the mainstream harvesting equipment. It relies on the digging shovel to cut into the soil and the conveyor chain to clear and transfer the soil to complete the harvesting operation. It is the core equipment support for reducing reliance on manual labor and improving production efficiency.

[0003] Currently, there are two common technical approaches used in field production. The first is the manual driving, fixed-depth harvesting scheme, which uses a powered chassis paired with a fixed digging shovel and conveying mechanism. The driver controls the driving path and pre-sets the digging depth, and the core operating parameters remain essentially constant during operation. This scheme is structurally mature and reliable, with low manufacturing costs, and is widely used in contiguous planting areas in plains. The second is the RTK (Real-Time Kinematic) linear navigation assisted harvesting scheme, which adds an RTK positioning unit to the fixed-depth model to achieve automatic navigation for straight-line travel. The driver is responsible for fine-tuning the depth and avoiding obstacles. This scheme can reduce the driver's labor intensity and improve the alignment accuracy of rows and the utilization rate of the working width.

[0004] Although the two aforementioned solutions have a mature foundation for industrial application, they still face common technical limitations in actual operations under complex field conditions. Firstly, the working depth is difficult to adapt to dynamic changes in ridge surface conditions. Affected by ridge surface subsidence, terrain undulations, and uneven soil properties, fixed depth settings or manual fine-tuning cannot promptly match changes in the ridge surface baseline, easily leading to situations where excessive digging increases machine load, or insufficient digging causes damage and missed harvesting of tubers. Secondly, the coordination between driving and operational parameters in obstacle avoidance is insufficient. During manual obstacle avoidance, the driving state dynamically changes, and the parameters of the operating mechanism cannot be adjusted synchronously, easily causing problems such as material accumulation in the conveyor chain and instantaneous fluctuations in working depth, affecting operational continuity and harvesting quality. As the requirements for harvesting accuracy and operational stability in yam cultivation continue to increase, developing adaptive control technology adapted to complex field conditions has become a practical need for industrial upgrading. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive control method for the operating parameters of an RTK-based autonomous obstacle avoidance yam harvester, in order to solve the problems mentioned in the background art, such as the difficulty in adapting the operating depth of the yam harvester to the dynamic changes in the working conditions of the ridge surface and the insufficient coordination between the driving and operating parameters under obstacle avoidance conditions.

[0006] To solve the above-mentioned technical problems, this invention provides an adaptive control method for operating parameters of an RTK-navigated autonomous obstacle avoidance yam harvester, comprising the following steps: S1. Multi-source operation perception data acquisition: Taking a yam harvester equipped with an RTK positioning unit, an environmental perception unit and an operation status detection unit as the object, multi-source operation perception data is acquired through synchronous acquisition of multiple units, and a standardized multi-source perception dataset is generated after standardized processing. S2. Autonomous navigation obstacle avoidance trajectory planning and working condition segmentation: Using the standardized multi-source perception dataset as input, combined with the preset working width and obstacle avoidance safety distance constraints, the target driving trajectory and the corresponding trajectory tracking control quantity are generated; at the same time, the working condition segmentation is identified according to the trajectory operation characteristics, and the working condition segment identifiers of stable straight working segment and obstacle avoidance turning segment are output. S3. Adaptive Matching Decision of Operation Parameters: Using the target travel trajectory, working condition segment identifier, and standardized multi-source sensing dataset as input, the ridge surface reference drift is inverted based on the correspondence between digging resistance and shovel geometry parameters within a stable straight working segment; combined with the currently effective ridge surface reference drift and the heading deviation between the target travel trajectory and the real-time heading of the harvester, a relative ridge surface working depth reference is obtained; based on the relative ridge surface working depth reference, the digging depth, travel speed, and conveyor chain speed are coordinated and adjusted to generate operation parameter control commands; S4. Coordinated control execution of operation parameters: Taking the trajectory tracking control quantity and operation parameter control command as input, the driving system and operation execution system are synchronously driven through the coordinated control of the whole vehicle; the operation status data after execution is sent back to S1 to participate in the construction and control iteration of the next round of standardized multi-source perception dataset, so as to realize the adaptive closed-loop control of operation parameters under the autonomous obstacle avoidance of the harvester.

[0007] As a further improvement to this technical solution, step S1 involves collecting and acquiring multi-source operation sensing data and generating a standardized multi-source sensing dataset, including the following steps: S1.1. By synchronously collecting data from the RTK positioning unit, environmental perception unit and operation status detection unit mounted on the yam harvester, multi-source operation perception data including the harvester's real-time pose, field obstacle information, digging resistance, travel distance, travel speed, conveyor chain load and conveyor chain speed are obtained. S1.2 Perform time synchronization, filtering and noise reduction, and dimension normalization on the multi-source operation sensing data to generate a standardized multi-source sensing dataset.

[0008] As a further improvement to this technical solution, step S2 involves generating the target driving trajectory and the corresponding trajectory tracking control quantity, and performing work condition segment identification and outputting work condition segment identifiers, including the following steps: S2.1 Using the standardized multi-source perception dataset as input, and combining the preset operating width and obstacle avoidance safety distance constraints, plan and generate a target driving trajectory that includes path point coordinates, heading angle and trajectory curvature, and calculate the corresponding trajectory tracking control quantity. S2.2 Extract three types of trajectory operation features from the target driving trajectory: trajectory curvature, heading change rate, and driving speed fluctuation. By setting thresholds, distinguish between stable straight operation segments and obstacle avoidance turning segments. The stable straight operation segment is the effective trigger condition for subsequent ridge surface reference drift inversion, and the obstacle avoidance turning segment is the corresponding condition for subsequent ridge surface reference drift maintenance and attitude coupling compensation. Output the corresponding condition segment identifier.

[0009] As a further improvement to this technical solution, in step S3, the target driving trajectory, working condition segment identifier, and standardized multi-source sensing dataset are used as inputs. The process involves triggering benchmark inversion and multi-parameter collaborative matching according to working condition levels to generate operation parameter control instructions, including the following steps: S3.1. Using the working condition section identifier as the trigger condition, a dual gating mechanism of working condition verification and cause discrimination is adopted. Within the stable straight working section, the excavation resistance data in the standardized multi-source sensing dataset is called, and the force-geometry mapping inversion is performed in combination with the preset shovel geometric parameters to output the ridge surface reference drift. S3.2. Using the drift amount of the ridge surface reference, the expected heading corresponding to the target travel trajectory and the heading deviation calculated by the real-time heading of the harvester as the correction input, the reference translation correction and attitude coupling compensation are performed on the depth reference, and the output is the relative ridge surface operation depth reference after superposition. S3.3. Using the relative ridge surface working depth benchmark as the core constraint, the digging load is characterized by real-time digging resistance. The relationship between digging load and conveying capacity is matched, and the digging depth, travel speed and conveying chain speed are adjusted in linkage, and the operation parameter control command is output.

[0010] As a further improvement to this technical solution, the specific control process of the dual gating mechanism for operating condition verification and cause determination in S3.1 includes the following steps: S3.11. Receive the working condition segment identifier and perform geometric working condition verification. When the working condition segment identifier is a stable straight working section, pass the first level of gating. S3.12. Call the excavation resistance data in the standardized multi-source sensing dataset to verify the physical causes, and determine whether the resistance change comes from the ridge elevation drift or the local soil mutation, through the second gating. S3.13 When both the geometric condition verification and the physical cause verification pass, the starting force-geometric mapping inversion calculation is performed in combination with the preset shovel-shaped geometric parameters; if either verification fails, the drift amount of the previous effective ridge surface reference is maintained.

[0011] As a further improvement to this technical solution, in S3.12, the physical cause verification is performed based on the dynamic soil-machine coupling stability factor; the dynamic soil-machine coupling stability factor is a quantitative index used to measure the credibility of the current excavation resistance change being dominated by the uniform drift of the ridge elevation, and its value is calculated by fusing three physical dimension components, including the following steps: S3.121. Call the mining resistance and driving distance data in the standardized multi-source sensing dataset, extract the corresponding data sequence within the short time window, calculate the linear fit goodness of resistance with driving distance, and obtain the linearity component. S3.122. Extract the resistance gradient values ​​of the current detection segment and the adjacent segments before and after it, calculate the resistance gradient difference between segments, and obtain the spatial continuity component. S3.123. Call the pose data in the standardized multi-source sensing dataset, extract the heading angle and pitch angle sequences within the short time window, calculate the attitude fluctuation amplitude, and obtain the attitude decoupling components.

[0012] As a further improvement to this technical solution, in S3.12, after completing the calculation of the three physical dimension components, the dynamic soil-machine coupling stability factor is obtained through weighted fusion and the physical cause is determined. After the determination is passed, local outlier removal is performed and the effective resistance data is output to S3.13, including the following steps: S3.124. Perform weighted fusion calculation on the linearity component, spatial continuity component and attitude decoupling component to obtain the comprehensive value of the dynamic soil-machine coupling stability factor; S3.125. Compare the comprehensive value with the preset calibration trigger threshold range. When the comprehensive value is within the acceptable range, it is determined that the current change in excavation resistance is dominated by the uniform drift of the ridge elevation, and the physical cause is verified. After the physical cause verification in S3.126 is passed, the detected local resistance peaks are eliminated by using amplitude limiting combined with cross-verification of adjacent segments to obtain effective excavation resistance data, which is then output to S3.13 for the inversion calculation of the ridge surface reference drift.

[0013] As a further improvement to this technical solution, in step S3.2, the depth reference is subjected to reference translation correction and attitude coupling compensation, and the output relative to the ridge surface working depth reference is superimposed, including the following steps: S3.21. Perform benchmark translation correction, using the ridge surface benchmark drift as the overall correction amount, and perform a same-direction translation adjustment on the initial depth benchmark to correct the systematic depth deviation introduced by the overall settlement or lifting of the ridge surface. S3.22. Perform attitude coupling compensation, using the heading deviation as input, and calculate the compensation amount of the effective digging depth in combination with the cutting geometry of the shovel blade, and correct the instantaneous depth deviation introduced by the tilt of the body when avoiding obstacles. S3.23. The depth reference after the reference translation correction is superimposed with the attitude coupling compensation amount to obtain the final relative ridge surface operation depth reference.

[0014] As a further improvement to this technical solution, in S3.3, the digging load and conveying capacity are matched with the relative ridge surface operation depth benchmark as a constraint, and the operation parameters are adjusted in conjunction with the output control command, including the following steps: S3.31. Determine the depth constraint boundary. Taking the relative ridge surface operation depth benchmark as the core and combining it with the preset operation depth allowable fluctuation threshold, define the adjustable range of the excavation depth. S3.32 Match the load and conveying relationship, use real-time excavation resistance to represent the excavation load and conveying chain load data to represent the conveying capacity, match the excavation load and conveying capacity relationship, and obtain the coordinated matching range of travel speed and conveying chain speed. S3.33 Multi-parameter linkage adjustment: Within the adjustable range of digging depth, the digging depth, travel speed and conveyor chain speed are adjusted synchronously in combination with the load and conveying relationship to maintain a dynamic balance between digging load and conveying capacity. S3.34. Generate operation parameter control instructions, integrating the adjusted digging depth, travel speed and conveyor chain speed parameters into operation parameter control instruction output.

[0015] As a further improvement to this technical solution, in step S4, the coordinated driving of the entire vehicle is executed based on the trajectory tracking control quantity and the operation parameter control command, and the operation status data is returned to achieve closed-loop iteration, including the following steps: S4.1 Receive trajectory tracking control quantity and operation parameter control command, parse them and allocate them to the driving control unit and the operation execution control unit respectively, and determine the target control parameters corresponding to each unit; S4.2 The control system performs path tracking according to the trajectory tracking control quantity, and synchronously controls the excavating mechanism and the conveying mechanism to perform operation actions according to the operation parameter control instructions, so as to complete the coordinated drive of driving and operation; S4.3 Real-time acquisition of the actual position and speed of the driving system, as well as the actual digging depth, conveyor chain speed, and real-time digging resistance of the operation execution system; S4.4 The collected operation status data is sent back to the multi-source operation perception data acquisition stage of S1 and incorporated into the construction of the standardized multi-source perception dataset to participate in the next round of control parameter iteration, forming a closed loop of full-process adaptive control.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention solves the problem that existing solutions are difficult to adapt to dynamic changes in ridge elevation by inverting the ridge surface reference drift based on digging resistance and shovel geometry parameters within a stable straight working section, and combining it with heading deviation correction to obtain a relative ridge surface working depth reference. This enables real-time adjustment of working depth according to field conditions, effectively controls the load on machinery, reduces the probability of tuber damage and missed harvesting, and improves harvesting accuracy. 2. This invention solves the problem of asynchronous driving and operating parameters during obstacle avoidance by segmenting and identifying stable straight operating sections and obstacle avoidance turning sections, and by linking and matching digging depth, driving speed and conveyor chain speed and performing whole-vehicle coordinated control. It realizes dynamic coordinated adjustment of operating parameters under obstacle avoidance conditions, avoids material accumulation and depth fluctuations in the conveyor chain, and ensures continuous operation and stable harvesting quality. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall method steps of the adaptive control method for operating parameters of the yam harvester based on RTK navigation in this invention; Figure 2 This is a schematic diagram illustrating the application scenario of the adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance according to the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figures 1-2 As shown, this embodiment provides an adaptive control method for the operating parameters of an RTK-navigated autonomous obstacle avoidance yam harvester, including the following steps: S1. Multi-source operation perception data acquisition: Taking a yam harvester equipped with an RTK positioning unit, an environmental perception unit and an operation status detection unit as the object, multi-source operation perception data is acquired through synchronous acquisition of multiple units, and a standardized multi-source perception dataset is generated after standardized processing. In this embodiment, the vehicle controller of the yam harvester is connected to the RTK positioning unit, environmental sensing unit, and operation status detection unit via a CAN bus. The vehicle controller sends a synchronization trigger signal, enabling the three units to start data acquisition under the same clock source, and adds a timestamp to each frame of data. The acquired data is sent to the data processing module within the vehicle controller, where time synchronization, filtering and noise reduction, and dimension normalization are performed sequentially to generate a standardized multi-source sensing dataset, which is then output to S2. The specific acquisition content of each unit is described in S1.1, and the standardization process is described in S1.2.

[0020] S1.1. By synchronously collecting data from the RTK positioning unit, environmental perception unit and operation status detection unit mounted on the yam harvester, multi-source operation perception data including the harvester's real-time pose, field obstacle information, digging resistance, travel distance, travel speed, conveyor chain load and conveyor chain speed are obtained. Specifically, the RTK positioning unit combines a GNSS receiver and an inertial measurement unit (IMU). The GNSS receiver receives satellite navigation signals and differential correction information from the base station, while the IMU acquires the harvester's heading, pitch, and roll angles. In this embodiment, the RTK positioning unit uses a sampling frequency of 20Hz, achieving a positioning accuracy of ±2cm, a heading accuracy of ±0.2°, and pitch and roll accuracy of ±0.1°, outputting the harvester's real-time attitude.

[0021] The environmental perception unit employs a combination of lidar and a vision camera. The lidar is mounted on top of the harvester's front hood, scanning the working area in front of and to the sides of the harvester to detect the distance, orientation, and size of obstacles. The vision camera is mounted beside the lidar, acquiring field images and identifying obstacle types. In this embodiment, the lidar has a detection range of 0–10 m, a horizontal field of view of 360°, and a sampling frequency of 20 Hz; the vision camera has a resolution of 1920 × 1080 and a frame rate of 30 fps. The environmental perception unit outputs field obstacle information, including obstacle distance, azimuth, height, and width.

[0022] The operational status detection unit includes a digging resistance sensor, a travel distance sensor, a travel speed sensor, a conveyor chain load sensor, and a conveyor chain speed sensor. The digging resistance sensor is a strain gauge force sensor mounted on the connecting pin of the digging shovel, with a range of 0–50 kN, an accuracy of ±1% FS, and a sampling frequency of 200 Hz. The travel distance sensor is a rotary encoder mounted on the drive wheel shaft, with a resolution of 1024 pulses / revolution and a sampling frequency of 50 Hz. The travel speed sensor is a Hall effect speed sensor mounted on the drive wheel, with a range of 0–5 m / s, an accuracy of ±0.05 m / s, and a sampling frequency of 50 Hz. The conveyor chain load sensor is a torque sensor mounted on the conveyor chain drive shaft, with a range of 0–500 N·m, an accuracy of ±0.5% FS, and a sampling frequency of 100 Hz. The conveyor chain speed sensor is a rotary encoder mounted on the conveyor chain drive shaft, with a range of 0–3 m / s, an accuracy of ±0.02 m / s, and a sampling frequency of 100 Hz. The aforementioned sensors output digging resistance, travel distance, travel speed, conveyor chain load, and conveyor chain speed, respectively.

[0023] S1.2 Perform time synchronization, filtering and noise reduction, and dimension normalization on the multi-source operation sensing data to generate a standardized multi-source sensing dataset.

[0024] Specifically, time synchronization uses the sampling time of the RTK positioning unit as a reference, and performs linear interpolation on the data of other units to align all data to the same time series. Let the reference time be... Adjacent sampling times are and The corresponding original data value is and Then the interpolated data value at the reference time for: ; in, Indicates the reference time The interpolated data values; Indicates the adjacent previous sampling time The original data values; Indicates the next adjacent sampling time The original data values; Indicates the reference time; Indicates the adjacent previous sampling time; This indicates the next adjacent sampling time. The time unit is seconds. The dimensions are consistent with the data type of the interpolation.

[0025] In the filtering and denoising process, position coordinates and attitude angle data are smoothed using Kalman filtering, and the process noise covariance and observation noise covariance are determined based on sensor accuracy calibration. Digging resistance, travel distance, and travel speed are filtered using moving average filtering with a window length of 5. Conveyor chain load and conveyor chain speed are filtered using first-order low-pass filtering with a cutoff frequency of 5Hz. Field obstacle information is denoised using clustering and threshold filtering, with a cluster radius of 0.3m, and false targets less than 0.2m or greater than 10m are removed. The filtered data retains the low-frequency trend reflecting changes in ridge elevation while suppressing high-frequency noise caused by vibration and electromagnetic interference.

[0026] Dimensional normalization employs Min-Max normalization, mapping each data point to the [0,1] interval. The calibration ranges for each data point are determined based on actual operating conditions. In this embodiment, the calibration ranges are: excavation resistance 0–50 kN, travel speed 0–5 m / s, conveyor chain speed 0–3 m / s, conveyor chain load 0–500 N·m, obstacle distance 0–10 m, obstacle azimuth 0–180°, obstacle height 0–3 m, and obstacle width 0–5 m. For attitude angle data such as heading, pitch, and roll angles, which have positive and negative directions, Z-score normalization is used.

[0027] Finally, the time-synchronized, filtered, and normalized data sequences are fused together using a unified timestamp to generate a standardized multi-source sensing dataset. This dataset is a two-dimensional matrix, with each row corresponding to a timestamp and each column corresponding to a data feature. The column content includes position coordinates, heading angle, pitch angle, roll angle, obstacle distance, obstacle azimuth angle, obstacle height, obstacle width, digging resistance, travel distance, travel speed, conveyor chain load, and conveyor chain speed. In this embodiment, the sampling frequency of the standardized multi-source sensing dataset is set to 20Hz, that is, a set of data vectors is generated every 50ms, and this dataset is output to S2.

[0028] S2. Autonomous navigation obstacle avoidance trajectory planning and working condition segmentation: Using the standardized multi-source perception dataset as input, combined with the preset working width and obstacle avoidance safety distance constraints, the target driving trajectory and the corresponding trajectory tracking control quantity are generated; at the same time, the working condition segmentation is identified according to the trajectory operation characteristics, and the working condition segment identifiers of stable straight working segment and obstacle avoidance turning segment are output. In this embodiment, the vehicle controller receives the standardized multi-source sensing dataset output by S1 and extracts the real-time pose of the harvester and information on obstacles in the field from it. Preset working width. Based on the agronomic requirements for yam ridge planting, a distance of 1.2m is used in this embodiment; obstacle avoidance safety distance. Based on the harvester's external dimensions and operational safety requirements, this embodiment uses 1.5m. The target travel trajectory is generated with the row direction as the reference, and the path point spacing is... Based on the requirements for RTK positioning accuracy and trajectory smoothness, this embodiment uses 0.1m. The trajectory tracking control quantities include front wheel steering angle commands and driving speed commands. The working condition segment identifier uses 0 to represent a stable straight-line working segment and 1 to represent an obstacle avoidance turning segment, outputting point by point to S3 along the trajectory. Specific planning and segmentation processes are shown in S2.1 and S2.2: S2.1 Using the standardized multi-source perception dataset as input, and combining the preset operating width and obstacle avoidance safety distance constraints, plan and generate a target driving trajectory that includes path point coordinates, heading angle and trajectory curvature, and calculate the corresponding trajectory tracking control quantity. Specifically, path planning is performed in a field plane coordinate system. This coordinate system has the harvester's starting point as the origin, the row direction as the X-axis, and the direction perpendicular to the row as the Y-axis. The harvester's current position is extracted from a standardized multi-source sensing dataset. Current heading angle and obstacle distance Azimuth ,high ,width .

[0029] In the absence of obstacles, the target's trajectory is a straight line parallel to the X-axis, with path point coordinates spaced at equal intervals. Generated along the X-axis, the first heading angle of each path point Set to 0, trajectory curvature Take 0.

[0030] When an obstacle is detected and the obstacle's lateral distance relative to the current path is... Less than When this occurs, obstacle avoidance path planning is triggered. Lateral distance. Calculate using the following formula: ; in, Indicates the lateral distance of the obstacle relative to the current path; Indicates the distance to the obstacle; Indicates the azimuth angle of the obstacle. When When the obstacle is located to the left of the current path, the obstacle avoidance path detours to the right; when When the obstacle is located to the right of the current path, the obstacle avoidance path will detour to the left.

[0031] The obstacle avoidance path consists of an entry arc segment, a bypass straight line segment, and an exit arc segment. The bypass straight line segment represents the lateral offset relative to the original path. for: ; in, This indicates the lateral offset of the bypass straight segment relative to the original path; Represents a symbolic function; Indicates the safe distance for obstacle avoidance; Indicates the working width.

[0032] Obstacle avoidance radius for: ; in, Indicates the radius of the obstacle avoidance arc; This indicates the angle between the detour path and the ridge direction. This angle is determined based on the balance between the turning radius and the lateral space occupied. In this embodiment, it is taken as 45°.

[0033] The center of the tangent arc is located in the direction of the current point's normal, at a distance of... The direction is determined by the direction of detour; the tangent arc is tangent to the detour straight line segment, and the point of tangency is translated from the center of the arc along the detour direction. We obtain the result. The center of the tangent arc is located at the normal direction of the end of the straight line segment, at a distance of... The cut arc is tangent to the original path, and the point of tangency is translated from the center of the cut arc along the direction of the original path. The length of the straight segment around the obstacle is obtained. The sum of the safety margin and twice the safety margin, which is determined based on the lidar detection error and obstacle size estimation error, is taken as 0.5m in this embodiment. The path point coordinates are generated by simultaneously solving the circular arc equation and the straight line equation, and the heading angle is... Take the tangent direction at the path point and the trajectory curvature. Take the arc segment , take 0 on the straight line segment.

[0034] The trajectory tracking control quantity is calculated using a pure tracking algorithm. For the target path point... Front wheel steering command for: ; in, Indicates the first Front wheel steering angle instructions at each path point; Indicates the wheelbase of the harvester; This indicates the angle between the harvester's current heading and the direction of the target waypoint; This represents the forward sight distance, which is determined based on the operating speed and trajectory tracking stability. In this embodiment, it is taken as 1.0m.

[0035] in Calculate using the following formula: ;

[0036] in, This indicates the angle between the harvester's current heading and the direction of the target waypoint; , Indicates the coordinates of the target path point; , This indicates the current coordinates of the harvester; Indicates the current heading angle of the harvester; This represents the arctangent function in the four quadrants. The calculated value is... Normalize to the interval [-π, π].

[0037] Driving speed command Based on the trajectory curvature Adjustment: ; in, Indicates the first Speed ​​instructions at each waypoint; This indicates the preset operating speed, which is determined based on the excavation load and the conveyor chain's handling capacity; in this embodiment, it is set to 0.6 m / s. This represents the speed attenuation coefficient, with the dimension in meters. This coefficient is determined based on the correspondence between the allowable speed in a curve and the curvature of the trajectory. In this embodiment, it is taken as 2.0m. This represents the absolute value of the trajectory curvature.

[0038] S2.2 Extract three types of trajectory operation features from the target driving trajectory: trajectory curvature, heading change rate, and driving speed fluctuation. By setting thresholds, distinguish between stable straight operation segments and obstacle avoidance turning segments. The stable straight operation segment is the effective trigger condition for subsequent ridge surface reference drift inversion, and the obstacle avoidance turning segment is the corresponding condition for subsequent ridge surface reference drift maintenance and attitude coupling compensation. Output the corresponding condition segment identifier.

[0039] Specifically, the trajectory curvature is extracted point by point from the target driving trajectory generated in S2.1. Calculate the rate of change of heading from the heading angle sequence of the target trajectory. : ; in, Indicates the first Rate of change of heading at each waypoint; This represents the time interval between adjacent path points. This time interval is consistent with the sampling period of the standardized multi-source sensing dataset in S1, and in this embodiment, it is taken as 0.05s.

[0040] Driving speed fluctuation In length Calculation of driving speed command sequence based on target driving trajectory within the sliding window: ; in, Indicates the first Speed ​​fluctuations at each waypoint; This represents the length of the sliding window, which corresponds to a 1-second time window and matches the sampling frequency of 20Hz. In this embodiment, it is set to 20. This represents the average speed command within the window.

[0041] Set curvature threshold This threshold corresponds to a turning radius of 20m; when the trajectory curvature is less than this value, the trajectory is approximately a straight line; heading rate of change threshold. This threshold is determined based on the range of heading angle fluctuations during stable straight-line travel; speed fluctuation threshold This threshold is determined based on the permissible fluctuation range of travel speed during stable operation. When and and At that time, the judgment of the first If a path point is in a stable straight section, the output condition segment identifier is 0; otherwise, it is determined to be an obstacle avoidance turning section, and the output condition segment identifier is 1. The condition segment identifier is output to S3 point by point along the trajectory. The stable straight section is used to trigger subsequent ridge surface reference drift inversion, and the obstacle avoidance turning section is used to maintain the ridge surface reference drift amount and perform attitude coupling compensation.

[0042] S3. Adaptive Matching Decision of Operation Parameters: Using the target travel trajectory, working condition segment identifier, and standardized multi-source sensing dataset as input, the ridge surface reference drift is inverted based on the correspondence between digging resistance and shovel geometry parameters within a stable straight working segment; combined with the currently effective ridge surface reference drift and the heading deviation between the target travel trajectory and the real-time heading of the harvester, a relative ridge surface working depth reference is obtained; based on the relative ridge surface working depth reference, the digging depth, travel speed, and conveyor chain speed are coordinated and adjusted to generate operation parameter control commands; In existing technologies, digging depth is typically adjusted based on fixed settings or manual experience, with depth control independent of travel and conveying control. This approach differs in that it uses the working condition segment identifier as a trigger, first performing geometric condition verification, and then physical cause verification based on dynamic soil-machine coupling stability factors. Force-geometric mapping inversion is only initiated when resistance changes are determined to be dominated by uniform drift in the ridge elevation, thus suppressing misadjustments caused by local soil abrupt changes and attitude disturbances. The inverted ridge reference drift is further combined with the heading deviation, and after reference translation correction and attitude coupling compensation, a relative ridge working depth reference is formed, which constrains the coordinated adjustment of digging depth, travel speed, and conveyor chain speed.

[0043] It should be noted that this method is applicable to yam planting fields where soil moisture content and compaction vary little within a short time window and the ridge shape is basically regular. RTK positioning accuracy should be no less than ±2cm, excavation resistance sensor accuracy no less than ±1%FS, and lidar detection error no greater than 0.1m. If soil or topographic conditions exceed the above ranges, a dual gating mechanism and outlier removal mechanism can reduce the probability of misadjustment, but the inversion accuracy may decrease. When soil parameters show significant spatiotemporal differences, soil specific resistivity... and coefficient of friction It can be updated through field calibration or zonal calibration before operation.

[0044] Specifically, the vehicle controller receives the target driving trajectory and working condition segment identifier output by S2, as well as the standardized multi-source sensing dataset output by S1. Within the stable straight working segment, it calls up the digging resistance data and, in conjunction with the preset shovel geometry parameters, inverts the ridge surface reference drift; after comparing and updating with the current effective value, it sequentially performs reference translation correction and attitude coupling compensation based on the heading deviation, and superimposes them to obtain the relative ridge surface working depth reference; then, combined with the real-time digging resistance and conveyor chain load, it adjusts the digging depth, driving speed, and conveyor chain speed in a coordinated manner, generates working parameter control commands, and outputs them to S4. The specific implementation of each step is shown in S3.1 to S3.3.

[0045] S3.1. Using the working condition segment identifier as the trigger condition, a dual gating mechanism of working condition verification and cause discrimination is adopted. Within the stable straight working section, the digging resistance data in the standardized multi-source sensing dataset is called, and force-geometric mapping inversion is performed in combination with preset shovel geometry parameters to output the ridge surface reference drift. Specifically, the vehicle controller uses the working condition segment identifier output by S2 as the trigger condition. When the working condition segment identifier is 0, the dual gating mechanism is activated. Among them, geometric working condition verification is used to confirm that the current working condition meets the inversion condition, and physical cause verification is used to determine whether the change in digging resistance comes from the ridge surface elevation drift rather than local soil abrupt change. After both verifications pass, force-geometric mapping inversion is performed in combination with preset shovel geometry parameters to output the ridge surface reference drift; if either verification fails, the previous valid ridge surface reference drift is retained. See S3.11 to S3.13 for the specific process: S3.11 Receive the working condition segment identifier and perform geometric working condition verification. When the working condition segment identifier is a stable straight working segment, the first gate is passed. Specifically, the vehicle controller receives the working condition segment identifier output by S2. The working condition segment identifier defined in S2 uses 0 to represent a stable straight working segment and 1 to represent an obstacle avoidance turning segment. If the working condition segment identifier is 0, it is determined that the current situation is a stable straight working segment, the first gate is passed, and subsequent physical cause verification is allowed; if the working condition segment identifier is 1, it is determined that the current situation is an obstacle avoidance turning segment, the first gate is not passed, and force-geometric mapping inversion is not initiated.

[0046] S3.12. The physical cause verification is performed using excavation resistance data from the standardized multi-source sensing dataset to determine whether the resistance change originates from ridge elevation drift or local soil abrupt changes, passing a second gating control. The physical cause verification is based on a dynamic soil-machine coupling stability factor. This dynamic soil-machine coupling stability factor is a quantitative indicator used to measure the credibility of the current excavation resistance change being dominated by uniform ridge elevation drift; its value is calculated by fusing three physical dimension components. Specifically, after passing the first gating control, the excavation resistance data from the standardized multi-source sensing dataset is used to calculate the dynamic soil-machine coupling stability factor. This factor is obtained by weighted fusing of linearity, spatial continuity, and attitude decoupling components. The specific calculations of each component are shown in S3.121 to S3.123, and the weighted fusing and determination are shown in S3.124 to S3.126. S3.121. Call the mining resistance and driving distance data in the standardized multi-source sensing dataset, extract the corresponding data sequence within the short time window, calculate the linear fit goodness of resistance with driving distance, and obtain the linearity component. Specifically, extracting the mining resistance within a short time window from a standardized multi-source sensing dataset. and driving distance Data sequence. Short window length taken Each sampling point corresponds to a 1-second time window. This sampling frequency is consistent with the 20Hz sampling frequency of the standardized multi-source sensing dataset in S1. At an operating speed of 0.6 m / s, the 1-second time window corresponds to approximately 0.6 m of travel distance, which is smaller than the typical elevation change scale of a ridge surface and can reflect local resistance trends. If the travel speed varies significantly within a short time window, a travel distance window can be used instead of the time window, with a window length of 0.5 m. The number of sampling points is adaptively adjusted according to the speed.

[0047] A univariate linear fit is performed on the data sequence within the window, and the fitting model is as follows: ; in, Indicates the first Fitted values ​​of excavation resistance at each sampling point; Indicates the first The driving distance at each sampling point; Indicates the slope of the fit; This represents the fitting intercept. and Determined by the least squares method.

[0048] Goodness of linear fit Calculate using the following formula: ; in, Indicates the linearity component; Indicates the first Measured values ​​of excavation resistance at each sampling point; Indicates the first Fitted values ​​of excavation resistance at each sampling point; This represents the average excavation resistance within the window; Indicates the length of the short time window. Dimensionless, with a value range of [0,1]. The closer the value is to 1, the more linear the change in resistance with the distance traveled.

[0049] When the denominator is zero, meaning there is no change in digging resistance within the window, it is necessary to distinguish between normal operating conditions and sensor malfunction. If the travel distance changes normally and the digging resistance does not differ significantly between adjacent windows, it is determined to be a normal operating condition with no ridge drift. Set to 1; if the travel distance does not change, or the digging resistance remains constant for a long time and differs significantly from the adjacent window, then it is determined that the digging resistance sensor is faulty, outputs a fault flag and maintains the drift amount of the previous effective ridge surface reference, and does not start the force-geometric mapping inversion.

[0050] S3.122. Extract the resistance gradient values ​​of the current detection segment and the adjacent segments before and after it, calculate the resistance gradient difference between segments, and obtain the spatial continuity component. Specifically, the travel distance is divided into test segments of equal length, and the length of each segment is taken as... The resistance gradient of the current detection segment. Calculate using the following formula: ; in, This represents the resistance gradient of the current detection segment; This indicates the excavation resistance at the end of the current detection section; This indicates the excavation resistance at the beginning of the current detection section; Indicates the length of the detection segment. Resistance gradient of the preceding adjacent segment. and the resistance gradient of the adjacent segment Calculate in the same way.

[0051] Spatial continuity components Calculate using the following formula: ; in, Represents spatial continuity components; This represents the preset maximum allowable value of the resistance gradient, determined based on the statistical upper bound of the resistance gradient between adjacent test sections within the experimental field. In this embodiment, it is taken as... ; This represents the function that takes the maximum value, used to... The lower limit is truncated to 0. Dimensionless, with a value range of [0,1]. The closer the value is to 1, the smaller the difference in resistance gradient between the current segment and the adjacent segment.

[0052] S3.123. Call the pose data in the standardized multi-source sensing dataset, extract the heading angle and pitch angle sequences within the short time window, calculate the attitude fluctuation amplitude, and obtain the attitude decoupling components.

[0053] Specifically, the heading angle within a short time window is extracted from the standardized multi-source sensing dataset. With pitch angle Sequence, short time window length taken =20 sampling points. Heading angle fluctuation amplitude and pitch angle fluctuation amplitude Calculate based on standard deviation: ; ; in, Indicates the range of heading angle fluctuation; Indicates the first The heading angle at each sampling point; This represents the mean heading angle within the window; Indicates the amplitude of pitch angle fluctuation; Indicates the first Pitch angle at each sampling point; This represents the average pitch angle within the window.

[0054] Attitude decoupling components Calculate using the following formula: ; in, Indicates attitude decoupling components; This represents the preset attitude fluctuation reference value, determined based on the accuracy of the RTK positioning unit's heading and pitch angles and the measured values ​​of attitude fluctuations in the field. In this embodiment, it is taken as... ; This represents the natural exponential function. Dimensionless, with a value range of (0,1]. The closer the value is to 1, the smaller the attitude fluctuation.

[0055] After completing the calculation of the three physical dimension components, the dynamic soil-machine coupling stability factor is obtained through weighted fusion, and the physical cause is determined. After the determination is passed, local outlier removal is performed, and the effective resistance data is output to S3.13, including the following steps: S3.124. Perform weighted fusion calculation on the linearity component, spatial continuity component and attitude decoupling component to obtain the comprehensive value of the dynamic soil-machine coupling stability factor; Specifically, dynamic soil-machine coupling stability factor Calculated using the following weighted fusion formula: ; in, This represents the comprehensive value of the dynamic soil-machine coupling stability factor; These represent the weighting coefficients of the linearity component, spatial continuity component, and attitude decoupling component, respectively. In this embodiment, based on the statistical results of the response amplitude and discrimination of the three components under the condition of uniform drift of the ridge surface within a stable straight working section, the following values ​​are taken: And satisfy This weight can be recalibrated based on soil conditions during actual operation. Dimensionless, with a value range of [0,1].

[0056] S3.125. Compare the comprehensive value with the preset calibration trigger threshold range. When the comprehensive value is within the acceptable range, it is determined that the current change in excavation resistance is dominated by the uniform drift of the ridge elevation, and the physical cause is verified. Specifically, the preset calibration trigger threshold range is: The stability factor distribution is determined based on the separation between the uniform drift of the ridge surface and the local abrupt change in soil conditions within a stable straight-line operation section. In this embodiment, we take... .when If the change in current excavation resistance is determined to be dominated by a uniform drift in ridge elevation, the second gating is passed; otherwise, it is determined to be dominated by a local soil mutation or attitude disturbance, the second gating is not passed, and the drift amount of the previous effective ridge reference is maintained.

[0057] After the physical cause verification in S3.126 is passed, the detected local resistance peaks are eliminated by using amplitude limiting combined with cross-verification of adjacent segments to obtain effective excavation resistance data, which is then output to S3.13 for the inversion calculation of the ridge surface reference drift.

[0058] After the physical cause verification is passed, local outliers are removed from the excavation resistance data within the short time window. The amplitude limiting rule is: if the excavation resistance at a certain sampling point... satisfy: ; This point is then determined to be a local resistance peak. Among them, This represents the standard deviation of excavation resistance within the window; This represents the amplitude limiting coefficient, which is taken in this embodiment. .

[0059] The cross-validation rule for adjacent segments is as follows: if a sampling point is initially determined to be a peak, then it is further checked whether the resistance gradients of the adjacent segments before and after that point all exceed the preset gradient threshold. In this embodiment, If the gradients of both adjacent segments exceed... If the peak is confirmed to be caused by a local abrupt change in the soil, it should be removed; otherwise, the point should be retained.

[0060] After removing outliers, the number of sampling points removed from the statistics window. .like Exceeding window length If 50% of the data is lost, the current window's data is deemed unusable, the inversion is not performed, the previous valid ridge surface reference drift is maintained, and the determination is delayed until the next window. If the remaining excavation resistance data does not exceed 50% of the window length, then a moving average filter is applied to the remaining data, with the moving average window length set to [value missing]. To obtain effective excavation resistance data The output is sent to S3.13 for the inversion calculation of the ridge surface reference drift.

[0061] S3.13 When both the geometric condition verification and the physical cause verification pass, the starting force-geometric mapping inversion calculation is performed in combination with the preset shovel-shaped geometric parameters; if either verification fails, the drift amount of the previous effective ridge surface reference is maintained.

[0062] Specifically, when both the first and second gates are passed, the initiating force-geometric mapping inversion calculation is performed. Preset shovel-shaped geometric parameters include the shovel face inclination angle. blade width Length of shovel face Coefficient of friction between soil and shovel surface and soil resistivity In this embodiment, , .

[0063] Excavation resistance With excavation depth The force-geometric mapping relationship between them is: ; in, Indicates the excavation depth; This represents the cotangent function.

[0064] This mapping relationship assumes that the soil is homogeneous and the cutting process is in a steady state. When the soil parameters vary significantly in time and space, and It can be updated through field calibration or zonal calibration before operation. This mapping relationship provides a baseline conversion for inversion, and the inversion results, after double gating and outlier removal, are still subject to the correction constraints of S3.2.

[0065] make ,but This represents the drag-depth coefficient, measured in N / m. In this embodiment, .

[0066] Initial depth benchmark Based on the pre-set agronomic requirements for yam cultivation, this embodiment takes... Benchmark excavation resistance corresponding to the initial depth benchmark. Determined based on the force-geometric mapping relationship: ; In this embodiment, .

[0067] Ridge reference drift Inverse the process using the following formula: ; in, The dimension is m, a positive value indicates ridge surface lifting, and a negative value indicates ridge surface settling.

[0068] The excavation resistance after outlier removal and moving average filtering in S3.126 is used for force-geometry mapping inversion; the real-time excavation resistance in S3.3... These are the original values ​​at the current sampling time, used for load characterization and parameter linkage. Both originate from the same source; the difference lies in... After outlier removal and filtering, This is the original real-time value.

[0069] After obtaining the current ridge surface reference drift through inversion, it is compared with the currently effective ridge surface reference drift. If the absolute value of the difference between the two is... Exceeding the preset update threshold If the difference is positive or negative, the current effective ridge surface reference drift will be updated. If the difference does not exceed the update threshold, the current effective ridge surface reference drift will remain unchanged.

[0070] If continuous No update was triggered in any of the inversion cycles. In this embodiment, the following is taken: If the most recently inverted ridge surface reference drift is used, the current effective value will be forcibly updated to avoid the accumulation of reference drift error. If either the first or second gating fails, the previous effective ridge surface reference drift will be retained, and no inversion calculation will be performed.

[0071] S3.2. Using the drift amount of the ridge surface reference, the expected heading corresponding to the target travel trajectory and the heading deviation calculated by the real-time heading of the harvester as the correction input, the reference translation correction and attitude coupling compensation are performed on the depth reference, and the output is the relative ridge surface operation depth reference after superposition. Specifically, the vehicle controller acquires the currently effective ridge reference drift amount. and the expected heading corresponding to the target trajectory. Real-time heading of the harvester Heading deviation Calculate using the following formula: ; in, This indicates a deviation in heading. Normalize to the interval [-π, π]. and To correct the input, reference translation correction and attitude coupling compensation are performed sequentially. See S3.21 to S3.23 for details.

[0072] S3.21. Perform benchmark translation correction, using the ridge surface benchmark drift as the overall correction amount, and perform a same-direction translation adjustment on the initial depth benchmark to correct the systematic depth deviation introduced by the overall settlement or lifting of the ridge surface. Specifically, the reference translation correction is performed according to the following formula: ; in, This represents the depth reference after translational correction. When... When, it indicates that the ridge surface has risen, and the depth reference increases in the same direction; when When the depth reference decreases, it indicates that the ridge surface has settled and the depth reference decreases in the same direction. The dimension is m.

[0073] S3.22. Perform attitude coupling compensation, using the heading deviation as input, and calculate the compensation amount of the effective digging depth in combination with the cutting geometry of the shovel blade, and correct the instantaneous depth deviation introduced by the tilt of the body when avoiding obstacles. Specifically, when the harvester deflects to avoid obstacles, it tilts laterally, and its roll angle... The attitude coupling compensation amount is acquired by the inertial measurement unit of the RTK positioning unit in S1. Calculate using the following formula: ; in, This represents the attitude coupling compensation amount; This indicates the real-time roll angle of the harvester; Represents the sine function; Represents the cosine function. Roll angle. The direction of the harvester's right side lifting is considered positive, and the direction of the harvester's right side sinking is considered negative. When At this time, the right side of the shovel blade rises relative to the horizontal ground, reducing the actual digging depth. A positive value indicates that a depth command should be added; when At this time, the right side of the shovel blade lowers relative to the horizontal ground, increasing the actual digging depth. A negative value indicates that the depth command should be reduced. The effective component coefficient of the lateral tilt in the direction of excavation depth is: when the heading deviation is zero, the vertical height difference generated by the lateral tilt of the blade acts entirely in the direction of excavation depth, and the compensation is the largest; when the heading deviation increases, the angle between the lateral tilt of the blade and the lateral tilt of the ridge increases, the effective component of the tilt in the direction of excavation depth decreases, and the compensation decreases accordingly. The dimension is m.

[0074] S3.23. The depth reference after the reference translation correction is superimposed with the attitude coupling compensation amount to obtain the final relative ridge surface operation depth reference.

[0075] Specifically, relative to the ridge surface operation depth benchmark Calculate using the following formula: ; in, This represents the reference depth for ridge surface operations. The dimension is m, and it is output to S3.3 as the constraint core for adjusting the mining depth.

[0076] S3.3. Using the relative ridge surface working depth benchmark as the core constraint, the digging load is characterized by real-time digging resistance. The relationship between digging load and conveying capacity is matched, and the digging depth, travel speed and conveying chain speed are adjusted in linkage, and the operation parameter control command is output.

[0077] Specifically, the vehicle controller uses the relative depth of the ridge operation as a benchmark. As the core constraint, real-time resistance mining is performed using standardized multi-source sensing datasets. and conveyor chain load The system matches the relationship between excavation load and conveying capacity, and adjusts the excavation depth, travel speed, and conveyor chain speed in tandem within the adjustable excavation depth range, ultimately generating control commands for operational parameters. See S3.31 to S3.34 for details. S3.31. Determine the depth constraint boundary. Taking the relative ridge surface operation depth benchmark as the core and combining it with the preset operation depth allowable fluctuation threshold, define the adjustable range of the excavation depth. Specifically, the one-sided allowable deviation corresponding to the preset allowable fluctuation threshold of the working depth. Based on the distribution range of yam tuber growth depth, this embodiment takes... Adjustable excavation depth range Defined according to the following formula: ; ; in, This indicates the lower limit of the adjustable range for excavation depth; This indicates the upper limit of the adjustable range for excavation depth; This indicates the preset allowable deviation on one side. The total adjustable range of the excavation depth is... . and All dimensions are in meters (m).

[0078] S3.32 Match the load and conveying relationship, use real-time excavation resistance to represent the excavation load and conveying chain load data to represent the conveying capacity, match the excavation load and conveying capacity relationship, and obtain the coordinated matching range of travel speed and conveying chain speed. Specifically, to excavate resistance in real time Characterizing excavation load to convey chain load Characterizes conveying capacity. Reference conveyor chain load. Baseline driving speed Reference conveyor chain speed Benchmark excavation resistance It has been determined to be 19.59 kN in S3.13.

[0079] The coordinated matching range between travel speed and conveyor chain speed is determined by the following formula: ; ; ; ; in, This indicates the lower limit of the driving speed coordination range; Indicates the upper limit of the driving speed coordination range; This indicates the lower limit of the coordinated matching range for conveyor chain speeds; This indicates the upper limit of the coordinated matching range for conveyor chain speed; This indicates the lower limit of the permissible driving speed; in this embodiment, it is taken as... ; This indicates the maximum permissible speed limit; in this embodiment, it is taken as... ; This indicates the lower limit of the allowable conveyor chain speed; in this embodiment, it is taken as... ; This indicates the upper limit of the allowable conveyor chain speed; in this embodiment, it is taken as... ; This represents the driving speed adjustment coefficient, which is taken in this embodiment. =0.3; This represents the conveyor chain speed adjustment coefficient; in this embodiment, it is taken as... . All dimensions are m / s.

[0080] The above-described matching formula uses a linear form. To simplify the matching relationship, it is suitable for working conditions where the excavation resistance fluctuates slightly around the benchmark value within a stable straight working section. If the resistance deviation is large, a piecewise linear or table lookup method can be used for matching.

[0081] S3.33 Multi-parameter linkage adjustment: Within the adjustable range of digging depth, the digging depth, travel speed and conveyor chain speed are adjusted synchronously in combination with the load and conveying relationship to maintain a dynamic balance between digging load and conveying capacity. Specifically, within the adjustable excavation depth range Within the system, based on the load transmission correspondence, the excavation depth is adjusted synchronously according to the following rules. Driving speed and conveyor chain speed .

[0082] When real-time excavation resistance At that time, it was determined that the excavation load was too high. Travel speed Adjust to the lower limit of the collaborative matching interval Conveyor chain speed Adjust to the upper limit of the collaborative matching interval Digging depth Adjust according to the following formula: ; in, This indicates the adjusted target excavation depth; This represents the excavation depth adjustment coefficient, which is taken in this embodiment. =0.5.

[0083] When real-time excavation resistance At that time, it was determined that the excavation load was too low. Travel speed Adjust to the upper limit of the collaborative matching interval Conveyor chain speed Adjust to the lower limit of the collaborative matching interval Digging depth Adjust according to the following formula: ; When real-time excavation resistance At the same time, maintain the current excavation depth. driving speed Conveyor chain speed .

[0084] During the adjustment process, the load on the conveyor chain is monitored in real time. .like ,in As the upper limit of the conveyor chain load, this embodiment takes This further increases the conveyor chain speed. until or achieve The travel speed is adjusted by 0.05 m / s per step, and the conveyor chain speed is adjusted by 0.1 m / s per step.

[0085] like achieve back Still greater than Then reduce driving speed to and appropriately reduce the excavation depth. To reduce excavation and conveying loads; if this still cannot meet the requirements If the output is slowed down, an alarm will be triggered or a command to stop operation will be issued.

[0086] like ,in As the lower limit of the conveyor chain load, this embodiment takes If necessary, reduce the conveyor chain speed appropriately. In order to save energy.

[0087] S3.34. Generate operation parameter control instructions, integrating the adjusted digging depth, travel speed and conveyor chain speed parameters into operation parameter control instruction output.

[0088] Specifically, the adjusted excavation depth Driving speed and conveyor chain speed These are integrated into operational parameter control commands. The command format includes the target excavation depth. Target driving speed and target conveyor chain speed In this embodiment, The operation parameter control commands are output to the operation execution control unit of S4.

[0089] S4. Coordinated control execution of operation parameters: Taking the trajectory tracking control quantity and operation parameter control command as input, the driving system and operation execution system are synchronously driven through the coordinated control of the whole vehicle; the operation status data after execution is sent back to S1 to participate in the construction and control iteration of the next round of standardized multi-source perception dataset, so as to realize the adaptive closed-loop control of operation parameters under the autonomous obstacle avoidance of the harvester.

[0090] Specifically, the vehicle controller receives the trajectory tracking control input from S2 and the operation parameter control commands from S3. The trajectory tracking control input includes front wheel steering angle and travel speed commands, while the operation parameter control commands include target digging depth, target travel speed, and target conveyor chain speed. The vehicle controller distributes these commands to the driving control unit and the operation execution control unit according to the controlled object, synchronously driving the driving system, digging mechanism, and conveying mechanism. After execution, it collects the actual pose and travel speed of the driving system in real time, as well as the actual digging depth, conveyor chain speed, and real-time digging resistance of the operation execution system, and transmits this data back to S1 to participate in the construction of the next round of standardized multi-source sensing datasets. See S4.1 to S4.4 for the specific process: S4.1 Receive trajectory tracking control quantity and operation parameter control command, parse them and allocate them to the driving control unit and the operation execution control unit respectively, and determine the target control parameters corresponding to each unit; Specifically, the vehicle controller receives the trajectory tracking control quantity output by S2 and the operation parameter control command output by S3 via the bus. The trajectory tracking control quantity includes the... Front wheel steering angle command at each path point With driving speed command The operational parameter control instructions include the target excavation depth. Target driving speed and target conveyor chain speed .

[0091] Furthermore, the vehicle controller parses the above instructions and assigns them according to the controlled object. Front wheel steering angle instruction. Assigned to the driving control unit as the target steering angle for the steering actuator. Driving speed command. relative to target speed Both involve speed control; the vehicle controller takes the smaller of the two values ​​as the final target speed for the driving control unit, denoted as... The purpose of taking a smaller value is to simultaneously meet the curve speed reduction requirements of trajectory tracking and the work load constraints, so as to avoid increased trajectory tracking deviation or excessive excavation load due to excessive speed.

[0092] Target mining depth The digging depth control subunit, assigned to the work execution control unit, serves as the target depth for the digging mechanism's lifting actuator. Target conveyor chain speed. The conveyor chain speed control subunit, assigned to the work execution control unit, serves as the target speed corresponding to the conveyor chain drive motor.

[0093] Finally, the vehicle controller sends the allocated target control parameters to each control unit according to the control cycle. The driving control unit and the operation execution control unit respectively perform amplitude and rate limiting processing on the target control parameters to ensure that the actuators operate within the allowable range.

[0094] S4.2 The control system performs path tracking according to the trajectory tracking control quantity, and synchronously controls the excavating mechanism and the conveying mechanism to perform operation actions according to the operation parameter control instructions, so as to complete the coordinated drive of driving and operation; Specifically, the driving control unit receives the front wheel steering angle command. and the final target speed Then, control is executed as follows: For the front wheel steering angle, the driving control unit will... The control current is converted into that of the electro-hydraulic proportional valve, which drives the steering cylinder to actuate, causing the front wheels to track the steering angle. The steering angle control accuracy is ±0.5°. Regarding driving speed, the driving control unit will... This is converted into displacement control signals for the variable pump and motor in the hydrostatic drive mechanism, allowing the travel speed to track the displacement of the pump and motor. The speed control accuracy is ±0.02m / s.

[0095] The job execution control unit receives the target excavation depth. and target conveyor chain speed Then, control is executed as follows: For excavation depth, the operation execution control unit will... The target displacement of the hydraulic cylinder for adjusting digging depth is converted into a closed-loop control system via feedback from a displacement sensor, driving the digging shovel to rise and fall to the target depth with a depth control accuracy of ±0.005m. For the conveyor chain speed, the operation execution control unit will... The target speed is converted into the speed of the hydraulic motor driving the conveyor chain. The speed sensor provides feedback to form a closed-loop control, driving the conveyor chain to operate at the target chain speed. The chain speed control accuracy is ±0.02m / s.

[0096] The driving control unit and the operation execution control unit achieve synchronous drive through coordinated scheduling by the vehicle controller. The coordinated scheduling method is as follows: the vehicle controller simultaneously sends target control parameters to both control units within the same control cycle, and the two control units execute the parameters separately within the same control cycle. The control cycle is consistent with the sampling cycle of the standardized multi-source sensing dataset in S1, which is 0.05s in this embodiment. If a control unit fails to complete its action within one control cycle due to actuator response delay, the vehicle controller continues to send the same target control parameters in the next control cycle until execution is completed or the preset maximum number of retries is reached, which is 3 in this embodiment.

[0097] It should be noted that during the obstacle avoidance turning section, the driving control unit executes steering according to the front wheel steering angle command output by S2, and the work execution control unit executes the work action according to the target digging depth and target conveyor chain speed output by S3, so as to realize the coordinated drive of driving and work.

[0098] S4.3 Real-time acquisition of the actual position and speed of the driving system, as well as the actual digging depth, conveyor chain speed, and real-time digging resistance of the operation execution system; Specifically, during the execution of S4.2, the vehicle controller collects operational status data in real time through various sensors. The actual pose of the driving system is collected by the RTK positioning unit in S1, including position coordinates, heading angle, pitch angle, and roll angle. The driving speed is collected by the driving speed sensor in S1. The actual digging depth of the operation execution system is collected by a displacement sensor installed on the digging depth adjustment mechanism, with a range of 0–0.6 m, an accuracy of ±0.005 m, and a sampling frequency of 50 Hz. The conveyor chain speed is collected by the conveyor chain speed sensor in S1. The real-time digging resistance is collected by the digging resistance sensor in S1.

[0099] The aforementioned operational status data is processed by the vehicle controller according to the standardized processing flow in S1, including time synchronization, filtering and noise reduction, and dimension normalization, before being incorporated into the standardized multi-source sensing dataset. The acquisition frequency is consistent with the sampling frequency of the standardized multi-source sensing dataset in S1; in this embodiment, it is 20Hz. The acquisition frequencies of the actual digging depth, conveyor chain speed, and real-time digging resistance are consistent with the corresponding sensor sampling frequencies and are aligned to 20Hz after time synchronization.

[0100] S4.4 The collected operation status data is sent back to the multi-source operation perception data acquisition stage of S1 and incorporated into the construction of the standardized multi-source perception dataset to participate in the next round of control parameter iteration, forming a closed loop of full-process adaptive control.

[0101] Specifically, the vehicle controller transmits the operational status data collected in S4.3 back to the multi-source operational perception data acquisition stage in S1. The transmitted data includes the actual pose and speed of the driving system, as well as the actual digging depth, conveyor chain speed, and real-time digging resistance of the operational execution system. After time synchronization, filtering, noise reduction, and dimensional normalization in S1.2, this data is incorporated into a standardized multi-source perception dataset for use in the next round of trajectory planning and operational condition segmentation in S2, adaptive matching decision-making of operational parameters in S3, and collaborative control execution in S4.

[0102] The data feedback participates in the next round of control parameter iteration as follows: S2 replans the target travel trajectory and identifies the working condition segment identifier based on the updated standardized multi-source sensing dataset; S3 re-inverts the ridge surface reference drift amount, corrects the relative ridge surface working depth reference, and adjusts the digging depth, travel speed, and conveyor chain speed in conjunction with the updated standardized multi-source sensing dataset and working condition segment identifier; S4 re-executes the collaborative drive based on the updated trajectory tracking control quantity and working parameter control command. This forms a closed-loop control link of perception, decision-making, execution, and feedback, realizing adaptive closed-loop control of working parameters under the harvester's autonomous obstacle avoidance.

[0103] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0104] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. An adaptive control method for operating parameters of an RTK-navigated autonomous obstacle avoidance yam harvester, characterized in that: Includes the following steps: S1. Multi-source operation perception data acquisition: Taking a yam harvester equipped with an RTK positioning unit, an environmental perception unit and an operation status detection unit as the object, multi-source operation perception data is acquired through synchronous acquisition of multiple units, and a standardized multi-source perception dataset is generated after standardized processing. S2. Autonomous navigation obstacle avoidance trajectory planning and working condition segmentation: Using the standardized multi-source perception dataset as input, combined with the preset working width and obstacle avoidance safety distance constraints, the target driving trajectory and the corresponding trajectory tracking control quantity are generated; at the same time, the working condition segmentation is identified according to the trajectory operation characteristics, and the working condition segment identifiers of stable straight working segment and obstacle avoidance turning segment are output. S3. Adaptive Matching Decision of Operation Parameters: Using the target travel trajectory, working condition segment identifier, and standardized multi-source sensing dataset as input, the ridge surface reference drift is inverted based on the correspondence between digging resistance and shovel geometry parameters within a stable straight working segment; combined with the currently effective ridge surface reference drift and the heading deviation between the target travel trajectory and the real-time heading of the harvester, a relative ridge surface working depth reference is obtained; based on the relative ridge surface working depth reference, the digging depth, travel speed, and conveyor chain speed are coordinated and adjusted to generate operation parameter control commands; S4. Coordinated control execution of operation parameters: Taking the trajectory tracking control quantity and operation parameter control command as input, the driving system and operation execution system are synchronously driven through the coordinated control of the whole vehicle; the operation status data after execution is sent back to S1 to participate in the construction and control iteration of the next round of standardized multi-source perception dataset, so as to realize the adaptive closed-loop control of operation parameters under the autonomous obstacle avoidance of the harvester.

2. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 1, characterized in that, In step S1, multi-source operation sensing data is collected and standardized multi-source sensing dataset is generated, including the following steps: S1.

1. By synchronously collecting data from the RTK positioning unit, environmental perception unit and operation status detection unit mounted on the yam harvester, multi-source operation perception data including the harvester's real-time pose, field obstacle information, digging resistance, travel distance, travel speed, conveyor chain load and conveyor chain speed are obtained. S1.2 Perform time synchronization, filtering and noise reduction, and dimension normalization on the multi-source operation sensing data to generate a standardized multi-source sensing dataset.

3. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 1, characterized in that, In step S2, the target driving trajectory and the corresponding trajectory tracking control quantity are generated, and the working condition segment identification is performed and the working condition segment identifier is output, including the following steps: S2.1 Using the standardized multi-source perception dataset as input, and combining the preset operating width and obstacle avoidance safety distance constraints, plan and generate a target driving trajectory that includes path point coordinates, heading angle and trajectory curvature, and calculate the corresponding trajectory tracking control quantity. S2.2 Extract three types of trajectory operation features from the target driving trajectory: trajectory curvature, heading change rate, and driving speed fluctuation. By setting thresholds, distinguish between stable straight operation segments and obstacle avoidance turning segments. The stable straight operation segment is the effective trigger condition for subsequent ridge surface reference drift inversion, and the obstacle avoidance turning segment is the corresponding condition for subsequent ridge surface reference drift maintenance and attitude coupling compensation. Output the corresponding condition segment identifier.

4. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 1, characterized in that, In step S3, the target driving trajectory, working condition segment identifier, and standardized multi-source perception dataset are used as inputs. Based on the working condition level, benchmark inversion and multi-parameter collaborative matching are triggered to generate operation parameter control instructions, including the following steps: S3.

1. Using the working condition section identifier as the trigger condition, a dual gating mechanism of working condition verification and cause discrimination is adopted. Within the stable straight working section, the excavation resistance data in the standardized multi-source sensing dataset is called, and the force-geometry mapping inversion is performed in combination with the preset shovel geometric parameters to output the ridge surface reference drift. S3.

2. Using the drift amount of the ridge surface reference, the expected heading corresponding to the target travel trajectory and the heading deviation calculated by the real-time heading of the harvester as the correction input, the reference translation correction and attitude coupling compensation are performed on the depth reference, and the output is the relative ridge surface operation depth reference after superposition. S3.

3. Using the relative ridge surface working depth benchmark as the core constraint, the digging load is characterized by real-time digging resistance. The relationship between digging load and conveying capacity is matched, and the digging depth, travel speed and conveying chain speed are adjusted in linkage, and the operation parameter control command is output.

5. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 4, characterized in that, The specific control process of the dual gating mechanism for operating condition verification and cause determination in S3.1 includes the following steps: S3.

11. Receive the working condition segment identifier and perform geometric working condition verification. When the working condition segment identifier is a stable straight working section, pass the first level of gating. S3.

12. Call the excavation resistance data in the standardized multi-source sensing dataset to verify the physical causes, and determine whether the resistance change comes from the ridge elevation drift or the local soil mutation, through the second gating. S3.13 When both the geometric condition verification and the physical cause verification pass, the starting force-geometric mapping inversion calculation is performed in combination with the preset shovel-shaped geometric parameters; if either verification fails, the drift amount of the previous effective ridge surface reference is maintained.

6. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 5, is characterized in that, In S3.12, the physical cause verification is performed based on the dynamic soil-machine coupling stability factor; the dynamic soil-machine coupling stability factor is a quantitative index used to measure the credibility of the current excavation resistance change being dominated by the uniform drift of the ridge elevation. Its value is calculated by fusing three physical dimension components, including the following steps: S3.

121. Call the mining resistance and driving distance data in the standardized multi-source sensing dataset, extract the corresponding data sequence within the short time window, calculate the linear fit goodness of resistance with driving distance, and obtain the linearity component. S3.

122. Extract the resistance gradient values ​​of the current detection segment and the adjacent segments before and after it, calculate the resistance gradient difference between segments, and obtain the spatial continuity component. S3.

123. Call the pose data in the standardized multi-source sensing dataset, extract the heading angle and pitch angle sequences within the short time window, calculate the attitude fluctuation amplitude, and obtain the attitude decoupling components.

7. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 6, is characterized in that, In step S3.12, after calculating the three physical dimension components, the dynamic soil-machine coupling stability factor is obtained through weighted fusion, and the physical cause is determined. After the determination is passed, local outlier removal is performed, and the effective resistance data is output to step S3.13, including the following steps: S3.

124. Perform weighted fusion calculation on the linearity component, spatial continuity component and attitude decoupling component to obtain the comprehensive value of the dynamic soil-machine coupling stability factor; S3.

125. Compare the comprehensive value with the preset calibration trigger threshold range. When the comprehensive value is within the acceptable range, it is determined that the current change in excavation resistance is dominated by the uniform drift of the ridge elevation, and the physical cause is verified. After the physical cause verification in S3.126 is passed, the detected local resistance peaks are eliminated by using amplitude limiting combined with cross-verification of adjacent segments to obtain effective excavation resistance data, which is then output to S3.13 for the inversion calculation of the ridge surface reference drift.

8. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 4, characterized in that, In step S3.2, the depth reference is subjected to reference translation correction and attitude coupling compensation, and the output relative to the ridge surface operation depth reference is superimposed, including the following steps: S3.

21. Perform benchmark translation correction, using the ridge surface benchmark drift as the overall correction amount, and perform a same-direction translation adjustment on the initial depth benchmark to correct the systematic depth deviation introduced by the overall settlement or lifting of the ridge surface. S3.

22. Perform attitude coupling compensation, using the heading deviation as input, and calculate the compensation amount of the effective digging depth in combination with the cutting geometry of the shovel blade, and correct the instantaneous depth deviation introduced by the tilt of the body when avoiding obstacles. S3.

23. The depth reference after the reference translation correction is superimposed with the attitude coupling compensation amount to obtain the final relative ridge surface operation depth reference.

9. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance as described in claim 4, characterized in that, In step S3.3, the process of matching the digging load and conveying capacity with the relative ridge surface operation depth benchmark as a constraint, and adjusting the operation parameters in conjunction with the output control command, includes the following steps: S3.

31. Determine the depth constraint boundary. Taking the relative ridge surface operation depth benchmark as the core and combining it with the preset operation depth allowable fluctuation threshold, define the adjustable range of the excavation depth. S3.32 Match the load and conveying relationship, use real-time excavation resistance to represent the excavation load and conveying chain load data to represent the conveying capacity, match the excavation load and conveying capacity relationship, and obtain the coordinated matching range of travel speed and conveying chain speed. S3.33 Multi-parameter linkage adjustment: Within the adjustable range of digging depth, the digging depth, travel speed and conveyor chain speed are adjusted synchronously in combination with the load and conveying relationship to maintain a dynamic balance between digging load and conveying capacity. S3.

34. Generate operation parameter control instructions, integrating the adjusted digging depth, travel speed and conveyor chain speed parameters into operation parameter control instruction output.

10. The adaptive control method for operating parameters of the yam harvester based on RTK navigation and autonomous obstacle avoidance according to claim 1, characterized in that, In step S4, the coordinated driving of the entire vehicle is executed based on the trajectory tracking control quantity and the operation parameter control command, and the operation status data is sent back to achieve closed-loop iteration, including the following steps: S4.1 Receive trajectory tracking control quantity and operation parameter control command, parse them and allocate them to the driving control unit and the operation execution control unit respectively, and determine the target control parameters corresponding to each unit; S4.2 The control system performs path tracking according to the trajectory tracking control quantity, and synchronously controls the excavating mechanism and the conveying mechanism to perform operation actions according to the operation parameter control instructions, so as to complete the coordinated drive of driving and operation; S4.3 Real-time acquisition of the actual position and speed of the driving system, as well as the actual digging depth, conveyor chain speed, and real-time digging resistance of the operation execution system; S4.4 The collected operation status data is sent back to the multi-source operation perception data acquisition stage of S1 and incorporated into the construction of the standardized multi-source perception dataset to participate in the next round of control parameter iteration, forming a closed loop of full-process adaptive control.