Open field vegetable unmanned tractor and multi-machine tool coordinated operation hitching control system

Abstract

The application discloses an open field vegetable unmanned tractor and multi-machine tool cooperative operation hitching and control system and belongs to the technical field of agricultural intelligent equipment. In order to solve the problems that the traditional agricultural machine automation system is difficult to expand and has poor reusability due to tight coupling design, and it is difficult to meet the demand of one machine and multiple tools cooperative operation, the system adopts a distributed architecture of three-end cooperation of equipment end-cloud end-display end, and a machine tool hitching and matching adaptation unit and a cooperative control unit are built in a vehicle-mounted intelligent agricultural machine operation system. Through automatic identification of machine tool information, differential control of suspension / traction dual mode, and operation mode dynamic reconstruction mechanism, and in cooperation with three types of standardized data exchange interfaces, the system realizes plug and play of multiple types of operation machine tools and tractor-machine tool cooperative control. The application improves the multi-machine tool adaptation efficiency and operation precision, and the system has excellent openness and operation reliability, and can support the whole process unmanned operation of open field vegetable cultivation, planting, management and harvesting.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment technology, and in particular to a control system for unmanned tractors and multi-implementation collaborative operation of open-field vegetables. Background Technology

[0002] Open-field vegetables are an important category in my country's vegetable production system. Their production process covers multiple stages, including tillage, ridging, sowing, transplanting, plant protection, and harvesting. The agricultural machinery used in each stage differs significantly in structure, operating parameters, control logic, and dynamic characteristics. With the rapid development of agricultural production towards unmanned and intelligent operations, achieving full automation of vegetable production through unmanned agricultural machinery has become a core development trend in the industry. This also places higher demands on the multi-implementation adaptability, composite operation efficiency, and system scalability of unmanned agricultural machinery.

[0003] Currently, most existing agricultural machinery automation control systems are customized for single-operation implements, resulting in a tightly coupled and bound structure between the control system hardware and software and the specific implement. This tight coupling design directly leads to poor system scalability and low functional reusability. When it is necessary to change the implement to switch operation stages, it is often necessary to redevelop the adapted control logic and manually debug and configure a large number of operation parameters by professionals. The integration cycle is long and the adaptation cost is high, making it difficult to achieve rapid composite operation of "one machine with multiple implements" and failing to meet the actual production needs of continuous multi-stage operations in open field vegetables.

[0004] In existing technical solutions that have achieved preliminary compatibility with multiple implements, a unified control strategy is generally adopted, without differentiated design for the dynamic characteristics of implements with different connection methods such as three-point suspension and towing. For three-point suspension implements, conventional control methods often use a single depth feedback adjustment, failing to incorporate tractor operating speed into the control linkage system. This fails to compensate for the inherent response lag of the electro-hydraulic suspension mechanism, easily leading to significant deviations in tillage and sowing depth when operating speed fluctuates, resulting in poor consistency in work quality. For towing implements, existing solutions lack a coordinated control mechanism for traction force and travel speed, and also lack effective yaw compensation logic. During operation, problems such as traction overload damaging the implement and implement lateral swaying off the course are prone to occur, posing both equipment safety hazards and failing to guarantee the straightness and width accuracy of the working area.

[0005] Furthermore, existing multi-tool control systems lack a robust dynamic reconfiguration and safety verification mechanism for control links during operation mode switching. During mode switching, the loading and switching of control algorithms and the transmission of operation parameters lack a response and confirmation process. The execution layer may enter the operation state before completing algorithm binding, easily leading to parameter mismatches, chaotic control logic, and other problems. This poses a safety hazard of uncontrolled operation and makes it difficult to guarantee the safety and parameter consistency during the switching process between different operation modes, thus limiting the reliability of practical applications of multi-tool composite operations.

[0006] In terms of system interfaces and compatibility, existing agricultural machinery automation systems mostly adopt proprietary communication protocols and customized data interfaces. The data interaction standards between different brands of tractors and implements are not uniform, making cross-brand equipment compatibility and adaptation difficult and costly to modify. At the same time, the overall system architecture is relatively closed. The access of new agricultural implements requires targeted modifications to the upper-level application code, and it is also difficult to achieve standardized data interoperability with third-party agricultural management platforms and agricultural machinery monitoring systems. The system's openness and scalability are insufficient, which restricts the large-scale promotion and application of the industry.

[0007] In practical field applications, existing unmanned agricultural machinery systems generally suffer from shortcomings such as insufficient communication reliability, limited positioning accuracy, and poor long-term operational stability. The complex network environment in the field can easily lead to data packet loss and communication interruptions, affecting the real-time nature of remote monitoring and control commands. Single satellite positioning methods are easily obstructed by trees, overpasses, etc., resulting in positioning jumps, which cannot meet the centimeter-level navigation requirements of precision operations for open-field vegetables. At the same time, most systems have a high degree of software architecture coupling, and functional modules cannot be upgraded or replaced independently. Long-term continuous operation can easily lead to problems such as system resource anomalies, operational lag, and even program crashes, making it difficult to support continuous unmanned operation of the entire process of cultivating, planting, managing, and harvesting open-field vegetables.

[0008] Therefore, this application proposes a control system for unmanned tractors and multi-implementation collaborative operation of open-field vegetables. Summary of the Invention

[0009] One objective of this invention is to propose a control system for unmanned tractors and multi-implementation collaborative operation of open-field vegetables. This invention can achieve plug-and-play and fully unmanned collaborative operation of various types of machinery for open-field vegetable production through a distributed architecture design that coordinates the equipment end, cloud end, and display end. It can also integrate automatic identification and adaptation of implements, dual-mode differentiated collaborative control, dynamic reconstruction of control links, and a standardized interface system.

[0010] The unmanned tractor and multi-implementation collaborative operation control system for open-field vegetables according to an embodiment of the present invention includes an on-board control unit mounted on the tractor, multiple types of implements that can be attached, and a remote monitoring terminal. The system is characterized by adopting a distributed architecture with three terminals—equipment end, cloud end, and display end—collaborating. The device uses an on-board computing unit as its core hardware carrier. The on-board computing unit runs an intelligent agricultural machinery operating system. The intelligent agricultural machinery operating system has a built-in communication framework, a multi-task scheduling module, and a task logic control module. It also connects the drive module, positioning module, perception module, decision-making module, planning module, and control module through a standardized data receiving and control interface. Each implement is equipped with an independent implement controller and establishes data communication with the on-board computing unit through a bus. The cloud service module deployed in the cloud is bidirectionally connected to the device end and the display end through a wireless communication link, and undertakes the functions of multi-agricultural machinery data aggregation, message forwarding, anomaly monitoring and unified task scheduling. The display terminal includes two types of human-computer interaction carriers: vehicle-mounted display and interactive terminal and mobile client, which are used for visual display of work status, setting of work parameters and remote monitoring and intervention. The intelligent agricultural machinery operating system has a built-in implement matching and adaptation unit and a collaborative control unit. The implement matching and adaptation unit is used to automatically identify the implement type and match the corresponding control parameter template after the implement is attached. The collaborative control unit is used to call the corresponding control algorithm according to the implement type and dynamically reconstruct the control link when the operation mode is switched, so as to realize the coordinated control of tractor operating parameters and implement operation parameters.

[0011] Furthermore, the implement controller has pre-stored and solidified implement description information, which includes implement model, connection type, geometric parameters, and rated operating parameter range. After the implement is attached, the implement controller automatically reports the implement description information to the vehicle computing unit via the ISOBUS / CAN bus. After the implement attachment adaptation unit parses and obtains the implement type, it retrieves the corresponding control algorithm module from the pre-stored implement parameter template library, automatically completes parameter adaptation, and realizes plug-and-play functionality for the implement.

[0012] Furthermore, the collaborative control unit categorizes the working implements into two types based on their connection method: three-point suspension type and towing type, and binds differentiated control algorithms to each type. For three-point suspension implements, the suspension height is the core control variable. A feedforward-feedback composite control strategy based on the linkage between tractor speed and tillage / sowing depth is adopted. The working height of the three-point suspension mechanism is adjusted in real time through the electro-hydraulic suspension controller. For towed implements, the magnitude of traction force, yaw compensation, and implement attitude are the control objects. A speed-traction force coordinated control algorithm is adopted to suppress traction overload and implement swaying.

[0013] Furthermore, when the operation mode is switched, the collaborative control unit sends a control algorithm switching instruction to the execution control layer and synchronously transmits the target operation quality index parameters, which include sowing depth, plant spacing and operation width. Only after receiving a response signal indicating that the new algorithm module has been successfully bound can the execution control layer initiate the control logic for the corresponding work mode, ensuring the safety and parameter consistency of the mode switching process.

[0014] Furthermore, the standardized data receiving and control interface defines three types of standardized data exchange interfaces. All three types of interfaces use JSON format for data serialization and implement data interaction through a topic subscription-publishing transmission mechanism. The first type is the job task data interface, which includes fields such as task ID, job type, job location, job parameters, task status and progress percentage. It is used to unify the description format of different types of job tasks and support the seamless scheduling of multiple types of job tasks. The second category is agricultural machinery status data interface, which includes fields for tractor real-time speed, engine speed, fault codes, implement suspension height and working depth, used to support remote vehicle diagnosis and status warning; The third category is the environmental perception data interface, which includes stratified soil temperature and humidity, meteorological data and air pressure fields, providing a standardized environmental data foundation for precision agriculture decision-making.

[0015] Furthermore, the vehicle-mounted computing unit adopts a layered software architecture, consisting of a UI interface layer, a local service layer, a core functional module group, a support module, and an agricultural implement controller side unit, from top to bottom. The UI interface layer corresponds to the human-computer interaction layer of the vehicle-mounted display touch terminal, and is used for visualizing the work status, setting parameters, and switching work modes. The local service layer is divided into internal local services and external local services, which are responsible for internal data bus communication between in-vehicle functional modules and network data interaction with external cloud and mobile terminals, respectively. Both rely on local databases to achieve data persistence. The core functional module group includes an agricultural machinery positioning module, a drive module, a fusion positioning module, an AI tillage module, an automatic ridging module, an automatic tillage module, a remote driving module, and a tracking operation module. The support modules include operation and maintenance management, log management, system monitoring, anomaly handling, remote update and remote monitoring modules, which provide basic guarantees for system operation; The implement controller side unit is divided into three categories according to the type of implement: tillage implements, transplanting implements and other implements. Each type of implement is equipped with an implement control module, a sensor driver and protocol transmission module, and completes data interaction with the vehicle system through a preset communication protocol.

[0016] Furthermore, the system adopts a self-developed message communication library as the communication middleware throughout the entire link. The message communication library encapsulates underlying network connection management, message serialization, heartbeat keep-alive, disconnection reconnection and traffic peak shaping logic, provides a simple topic subscription-publishing interface to the outside world, and the system command response latency is no higher than 50ms, and the reporting cycle of vehicle status data is no greater than 500ms.

[0017] Furthermore, the vehicle-mounted display and interaction terminal is a ruggedized industrial tablet computer with a 10.1-inch touchscreen, equipped with a customized Linux and Android dual system, and features 4G LTE, Wi-Fi, CAN, and RS485 communication interfaces. Its protection level is no less than IP54, and its operating temperature range is -20℃ to +70℃. Its display interface is divided into four functional areas. The top status bar is used to display meteorological and environmental data in real time. The left agricultural machinery status area uses a tractor 3D model as the core to display steering angle, vehicle speed, operation mode, PTO status and suspension height parameters. The right soil moisture area uses a layered bar chart to display the temperature and humidity data of the upper, middle and lower soil layers. The lower right operation information area is used to display and support editing of operation width, operation efficiency and operation depth parameters.

[0018] Furthermore, the mobile client is an application developed based on the Android system, adopting the MVVM architecture pattern and integrating a map SDK and a local persistent database; The mobile client supports simultaneous access for multiple agricultural machines and features functions such as operation monitoring and remote scheduling, real-time monitoring and threshold warning of agricultural machinery and implements status, display of soil moisture and meteorological data trends, visual playback of operation path trajectories, abnormal alarm push notifications, and maintenance record management.

[0019] Furthermore, the positioning module adopts a fusion positioning algorithm combining RTK+GNSS dual-frequency differential positioning and vehicle-mounted inertial navigation. The actual measured track deviation in the field is better than ±2.5cm, and the system supports OTA remote updates of the vehicle system software and control algorithm module.

[0020] The beneficial effects of this invention are: This invention embeds machine description information into the controller of each machine and automatically reports and identifies it after connection via ISOBUS / CAN bus. It also calls the machine parameter template library to automatically match the corresponding control algorithm module. Machine adaptation can be completed without manual parameter configuration. This breaks through the technical bottleneck of traditional agricultural machinery systems that require customized development of control logic for single types of machines and re-debugging and configuration when changing machines. It significantly reduces the integration and adaptation cost of multiple machines, significantly improves the versatility and reusability of the system, and truly realizes "ready to use after changing machines" for operating machines. Secondly, this invention differs from the traditional unified implement control method. It designs differentiated collaborative control algorithms for three-point suspended and towed implements. For suspended implements, it adopts a feedforward-feedback composite control strategy that links tractor speed with tillage / sowing depth. By compensating for the response lag of the electro-hydraulic actuator through feedforward and correcting depth deviation in real time through feedback, it can ensure the consistency of working depth under fluctuating working speed conditions. For towed implements, it adopts a scheme that combines speed-traction force collaborative control with yaw compensation. It avoids damage to the implement structure due to traction overload through speed linkage adjustment and suppresses implement lateral sway through steering fine adjustment to ensure working straightness. It specifically solves the problems of collaborative control accuracy and safety of implements with different connection forms. The operation control accuracy and operation stability are significantly better than conventional unified control schemes. However, this invention establishes a handshake verification process for switching work modes by issuing algorithm switching instructions at the collaborative control layer, synchronously transmitting target quality parameters, and responding to confirmation after the execution layer loads and binds the parameters. This dynamic reconstruction mechanism can achieve safe and stable reconstruction of the control link when switching between different work modes, thereby avoiding the risk of parameter mismatch and control failure during the mode switching process and ensuring the safety and consistency of work parameters for multi-tool composite operations. In addition, this invention defines three types of standardized data exchange interfaces: operation tasks, agricultural machinery status, and environmental perception. Combined with a self-developed topic subscription-publishing message communication library, it unifies the data interaction specifications of the entire system, shields the differences between underlying hardware and operating systems, realizes the compatible access of different brands of tractors and supporting implements, and reserves data docking capabilities for third-party agricultural management platforms. It builds an open and scalable system architecture, overcoming the shortcomings of traditional closed agricultural machinery systems, such as insufficient expansion capabilities and high cost of accessing new implements. Finally, this invention adopts a layered and decoupled vehicle-mounted software architecture and a complete closed-loop control link of perception-decision-execution-monitoring. Combined with a positioning algorithm that integrates RTK+GNSS dual-frequency differential and vehicle-mounted inertial navigation, and a self-developed communication middleware with heartbeat keep-alive and disconnection reconnection capabilities, it achieves centimeter-level operational navigation accuracy and millisecond-level control response while ensuring the stability and communication reliability of the system's long-term continuous operation in complex field network environments. It can fully support unmanned operations in multiple stages of open-field vegetable cultivation, such as tillage, ridging, sowing, and transplanting, providing solid technical support for the fully unmanned production of open-field vegetables. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1This is a diagram showing the overall architecture of the "equipment end - cloud end - display end" three-terminal collaboration of the unmanned tractor and multi-implementation collaborative operation control system for open-field vegetables proposed in this invention; Figure 2 This is a diagram of the onboard computing unit software architecture of the unmanned tractor and multi-implementation collaborative operation control system for open-field vegetables proposed in this invention. Detailed Implementation

[0022] To make the technical means and objectives and effects of the present invention easier to understand, the embodiments of the present invention will be described in detail below with reference to specific illustrations.

[0023] like Figure 1-2 As shown, the present invention discloses an unmanned tractor and multi-implementation collaborative operation control system for open-field vegetables, including an on-board control unit mounted on the tractor, multiple types of implements that can be attached, and a remote monitoring terminal. The system is characterized by adopting a distributed architecture with three terminals: the device end, the cloud end, and the display end.

[0024] The core hardware carrier of the device is the on-board computing unit mounted in the tractor cab, which is the local perception, decision-making and execution control core of the entire system.

[0025] Specifically, at the hardware level, the on-board computing unit is based on a high-performance computing unit and a high-speed networking module, and is connected to external peripherals such as a general controller, RTK+GNSS dual-frequency positioning equipment, display and interaction equipment, electric steering wheel, and perception sensor group. The perception sensor group includes millimeter-wave radar and industrial vision camera, which are used for obstacle distance detection and visual environment recognition, respectively. It can cover the detection and recognition needs of static and dynamic obstacles in the field, and provide data input for the perception obstacle avoidance function.

[0026] On the software side, the onboard computing unit runs a customized Linux operating system, relying on the real-time kernel of Linux to ensure low-latency scheduling of control commands. It deploys a self-developed intelligent agricultural machinery operating system, which has a built-in communication framework, a multi-task scheduling module, and a task logic control module as basic operational support. The communication framework supports both inter-process communication and network communication, enabling high-speed data interaction between various functional modules in the vehicle, as well as bidirectional network communication with external cloud and display terminals. The multi-task scheduling module is responsible for prioritizing multi-threaded tasks such as positioning, perception, decision-making, and control, ensuring that control tasks are executed first to meet real-time requirements.

[0027] Among them, the intelligent agricultural machinery operating system organically connects functional modules such as drive module, positioning module, perception module, decision-making module, planning module, and control module through standardized data receiving and control interfaces. Each module has cohesive functions and unified interfaces, supporting independent upgrades and replacements. Iterative optimization of individual functional modules can be completed without changing the overall architecture.

[0028] Secondly, each attachable implement is equipped with an independent implement controller. The controller has a built-in processing unit, storage unit and bus communication interface. It establishes bidirectional data communication with the on-board computing unit through ISOBUS / CAN bus to realize real-time interaction and coordinated control of tractor host operating parameters and implement operation parameters.

[0029] It should be noted that the cloud deployment is based on a cloud service server cluster, which can support the concurrent access of dozens or even hundreds of agricultural machines. Cloud service modules are deployed on it, including message forwarding units, data management units, anomaly monitoring units, and task scheduling units.

[0030] The cloud establishes bidirectional connections with the device and display terminals via 4G / 5G wireless communication links. The message forwarding unit is responsible for the transparent transmission and routing of instructions and status data between the device and display terminals. The data management unit uniformly stores and manages the operation data, status data, and environmental data reported by all agricultural machinery. The anomaly monitoring unit monitors the operating status and fault codes of all agricultural machinery in real time and automatically pushes early warning information to the corresponding display terminal when an anomaly is triggered. The task scheduling unit can uniformly allocate work plots and work tasks to multiple agricultural machines, realizing farm-level agricultural machinery cluster scheduling and management.

[0031] The demonstration platform includes two types of human-computer interaction carriers: vehicle-mounted display and interactive terminals and mobile clients, which are adapted to the usage needs of field operators and remote managers, respectively.

[0032] Specifically, the vehicle-mounted display and interaction terminal is installed in the tractor cab, providing local operation and status viewing access for on-site operators, and supporting operations such as setting operation parameters, switching operation modes, and switching between manual and automatic modes; the mobile client is an application installed on Android mobile devices, targeting farm managers and remote operators, providing functions such as remote monitoring, task assignment, and anomaly viewing. Together, the two constitute a dual-channel human-machine interaction system, realizing human-machine collaboration and remote monitoring.

[0033] The intelligent agricultural machinery operating system organically integrates implement attachment and adaptation units and collaborative control units, which are the core functional units for realizing plug-and-play and collaborative operation of multiple implements.

[0034] The implement matching unit is responsible for automatic implement identification and parameter matching. Each implement controller has a built-in non-volatile storage unit, which is pre-programmed with complete description information of the corresponding implement, including key attribute parameters such as the implement's unique identification ID, model name, mechanical connection type (three-point suspension / traction type), implement working width, suspension point geometric coordinate parameters, rated working depth range, rated working speed range, maximum traction force threshold, and sensor configuration information.

[0035] Once the implement and tractor are physically locked together, the implement controller automatically establishes a physical communication link with the on-board computing unit via the ISOBUS / CAN bus. After the link is established, the implement controller actively sends an implement identification data frame to the on-board system. The implement matching unit receives and parses the implement ID and type fields in the data frame, searches the pre-stored implement parameter template library for the corresponding control algorithm configuration file and default operating parameter set, automatically loads the corresponding control algorithm module into the running memory, and completes parameter initialization. After initialization, the implement matching unit sends an adaptation confirmation frame to the implement controller and reads the current status data of the implement. After verifying that the parameters match correctly, the entire adaptation process is completed. The entire process does not require the operator to manually select the implement model or enter configuration parameters. The complete adaptation time for a single implement is no more than 2 seconds, truly realizing "ready to use after implement change".

[0036] The collaborative control unit matches differentiated control strategies to work implements with different connection methods. According to the mechanical connection method of the implements, all work implements are divided into two categories: three-point suspension type and towing type, and corresponding control algorithm modules are bound to them respectively.

[0037] It should be noted that for three-point suspension implements, including common implements such as plows, rotary tillers, ridging machines, and seeders, the system uses suspension height as the core control variable and adopts a feedforward-feedback composite control strategy based on the linkage between tractor speed and tillage / sowing depth. The feedforward control channel collects the tractor's travel speed in real time and calculates the feedforward adjustment amount of the suspension height based on the pre-calibrated speed-work depth mapping model. When the work speed changes, the height of the three-point suspension mechanism is adjusted in advance to counteract the response lag characteristics of the electro-hydraulic actuator. The feedback control channel collects real-time actual working depth data from depth sensors installed on the machine's working components. After subtracting the actual depth from the preset target depth, a feedback correction amount is generated through a PID control algorithm. The feedback correction amount is superimposed with the feedforward adjustment amount and output to the electro-hydraulic suspension controller to drive the three-point suspension mechanism to complete precise height adjustment.

[0038] This composite control strategy can stabilize the tillage / sowing depth control deviation within ±1cm under working conditions where the operating speed fluctuates by ±20%, ensuring consistent operating depth throughout the entire process.

[0039] For towed implements, the system aims to ensure traction safety, straightness of operation, and stability of implement posture. It employs a speed-traction coordinated control algorithm, which collects tension data from the tension sensor on the tractor's traction mechanism in real time. When the traction force reaches 85% of the implement's rated maximum traction force, the system automatically reduces the tractor's speed to prevent traction overload from damaging the implement's structure or overloading the power system. Simultaneously, it calculates the lateral deviation between the actual working path and the preset planned path using RTK positioning data. Combined with yaw angle data collected by the implement's posture sensor, it generates steering correction commands. The system then fine-tunes the tractor's direction via the electronically controlled steering wheel to compensate for lateral sway during implementation, ensuring consistent working width and straight-line operation accuracy. The deviation of the straight-line operation track can be controlled within ±2.5cm.

[0040] When the operation mode changes, such as from tillage to sowing or from rotary tillage to ridging, the collaborative control unit can dynamically reconfigure the control link. The specific execution process is as follows: The collaborative control unit first issues a work stop command to the execution control layer, controlling the tractor to smoothly decelerate to a stop; Subsequently, a control algorithm switching instruction is issued. The instruction data packet carries the target quality index parameters corresponding to the new operation mode, including constraints such as target operation depth, plant spacing, ridge height, ridge spacing, and operation width. After receiving the switching instruction, the execution control layer first unloads the currently running control algorithm module and releases the corresponding resources, and then loads the control algorithm module corresponding to the new mode from the local algorithm library to complete the module interface mapping and target parameter initialization. After initialization, the execution control layer performs a valid range check on all parameters. Once it confirms that the parameters match the current equipment correctly, it sends an algorithm binding completion response signal back to the collaborative control unit. After receiving the response signal and confirming the parameter matching is valid, the collaborative control unit sends a work start enable command to the execution control layer. Only after receiving the enable command can the execution control layer start the control logic of the new work mode and control the tractor to resume work according to the new parameters.

[0041] This handshake response and parameter verification mechanism can completely avoid the risks of parameter mismatch and control failure during mode switching, ensuring the safety and parameter consistency of equipment switching and mode switching processes.

[0042] The system defines three types of standardized data exchange interfaces as a unified standard for data interaction between the device, the cloud, and the display. All interface data is serialized using JSON format and data transmission is achieved through a topic subscription-publish mechanism of a self-developed message communication library. This shields the differences between the underlying hardware and operating system, ensuring cross-platform availability of upper-layer applications. The specific definitions of the three types of standardized data exchange interfaces are shown in Table 1.

[0043] Table 1 Standardized Data Exchange Interface Definition Table

[0044] The first type is the task data interface, which includes fields such as task ID, task type, boundary coordinates of the work plot, task parameters, task status, and progress percentage. The task type covers all open-field vegetable operation scenarios, including plowing, rotary tillage, ridging, sowing, transplanting, and deep loosening. The task parameter field has an expandable structure, which can carry different parameters such as operation depth, seeding rate, plant spacing, ridge height, and width according to the task type. This interface unifies the description format of different types of task tasks, so that the upper-level scheduling application can issue various task tasks without discrimination, without the need to develop dedicated task scheduling logic for a single type of machinery, which greatly improves the system's versatility and scalability.

[0045] The second category is agricultural machinery status data interfaces, which include fields such as tractor real-time driving speed, engine speed, engine fault code, vehicle battery voltage, implement suspension height percentage, actual working depth, and implement fault code. The cloud and display terminals can subscribe to the data topics corresponding to this interface to obtain the real-time operating status of agricultural machinery and implements, supporting remote vehicle diagnostics and status warning functions. When the fault code field in the interface is non-zero or the value exceeds the preset safety threshold, the system automatically triggers a graded warning push.

[0046] The third category is environmental sensing data interfaces, which include stratified soil temperature and humidity, multi-dimensional meteorological data, and atmospheric pressure fields. Soil temperature and humidity values ​​are collected separately for the upper 0-10cm, middle 10-20cm, and lower 20-30cm layers. Meteorological data includes air temperature, light intensity, and... Parameters such as concentration, wind speed and direction, and cumulative rainfall; This interface provides a standardized environmental data foundation for precision agriculture decision-making. The system can dynamically optimize operational parameters by combining environmental data, such as adjusting sowing depth based on soil moisture and adjusting plant protection spraying parameters based on wind speed.

[0047] The software of the vehicle-mounted computing unit adopts a layered architecture design, consisting of a UI interface layer, a local service layer, a core functional module group, a support module, and an agricultural implement controller side unit from top to bottom. Each layer interacts with the other through standardized interfaces, resulting in a clear hierarchy and low coupling.

[0048] The UI interface layer corresponds to the human-machine interaction layer of the vehicle-mounted display touch terminal. It is the direct entry point for operators to interact with the vehicle-mounted system. It supports operations such as visual display of work status, online configuration of work parameters, switching between manual / automatic / AB line work modes, system calibration, and fault query. The interface adopts a touch interaction design to adapt to the operating habits of field workers.

[0049] The local service layer is divided into two parts: internal local service and external local service. The internal local service corresponds to the inter-process communication bus in the vehicle and is responsible for high-speed, low-latency data interaction between various functional modules in the vehicle, realizing data sharing and instruction transmission between modules. The external local service is responsible for network data interaction with cloud servers and mobile clients, handling the sending, receiving, parsing and protocol conversion of external network data. Both parts of the service rely on the local embedded database to realize persistent storage of operation data, status data and configuration parameters, ensuring that data is not lost and configuration can be recovered after the device is powered off and restarted.

[0050] The core functional modules include an agricultural machinery positioning module, a drive module, a fusion positioning module, an AI tillage module, an automatic ridging module, an automatic tillage module, a remote driving module, and a tracking operation module.

[0051] The agricultural machinery positioning module is responsible for collecting and initially analyzing the raw GNSS positioning data, and outputting basic position and speed information. The drive module integrates a controller unit, a GPS navigation unit, a throttle speed controllable unit, and a differential unit to achieve precise control of the tractor's walking drive, steering drive, and power output. It is the core of vehicle motion control. The fusion positioning module, combined with map management functions, integrates RTK differential positioning data with vehicle-mounted inertial navigation data. In scenarios where satellite signals are briefly blocked, it maintains positioning continuity through inertial navigation data, thereby improving positioning stability in complex field environments. The AI ​​farming module integrates AI line decision-making, planning, and execution sub-modules. It can intelligently generate the optimal operation route based on the perception results of field boundaries and obstacles, realizing intelligent planning of operation paths and automatic execution of operation actions. The automatic ridging module and the automatic tillage module correspond to the dedicated control logic of the ridging and tillage operation, respectively. They have built-in agronomic parameter models for the corresponding operations and can independently complete the full-process automatic control of the corresponding operations. The remote driving module supports remote manual intervention and control, receives remote control commands from the display terminal and converts them into vehicle control signals to cope with special and complex operation scenarios; The tracking operation module supports importing custom operation trajectories, enabling high-precision tracking operations on preset trajectories and meeting the operational needs of irregular plots and customized paths.

[0052] The support modules include operation and maintenance management, log management, system monitoring, anomaly handling, remote update and remote monitoring modules, providing a basic guarantee for the long-term stable operation of the system.

[0053] The operation and maintenance management module supports remote diagnostics and equipment information query, making it easier for operation and maintenance personnel to quickly locate problems; The log management module records the entire system operation log according to a hierarchical strategy, including debug logs, operation logs, and error logs. It supports local storage and cloud reporting for troubleshooting and system optimization. The system monitoring module monitors the usage of system resources such as CPU, memory, and storage in real time, as well as the operating status of each functional module; The anomaly handling module executes tiered handling logic for identified faults and anomalies. Minor anomalies are automatically reset and restored, while severe anomalies are automatically executed with parking safety policies and warnings are reported. The remote update module supports OTA (Over-The-Air) remote upgrades of system software and algorithm modules, and supports differential updates to reduce data consumption. The remote monitoring module is responsible for periodically reporting system status data to the cloud, supporting remote monitoring functions.

[0054] The implement controller side unit is divided into three categories according to the type of implement: tillage implements, transplanting implements, and other implements. Tillage implements include plows, rotary tillers, subsoilers, ridging machines, etc. Transplanting implements correspond to vegetable transplanters, and other implements include seeders, plant protection sprayers, etc.

[0055] Each type of implement controller has a built-in implement control module, sensor drive unit, and protocol transmission module. The implement control module is responsible for receiving control commands from the vehicle system and driving the implement's actuators. The sensor drive unit is responsible for collecting sensor data such as depth, speed, attitude, and pressure on the implement. The protocol transmission module packages the implement status data and sends it to the vehicle system according to the standard format of the "rough tillage controller and stand-alone machine PDA communication protocol", realizing two-way interaction and coordinated linkage between the tractor's main control parameters and the implement's operating parameters.

[0056] The system uses a self-developed message communication library as a unified communication middleware across the entire chain. This communication library encapsulates complex underlying logic such as network connection management, message serialization, heartbeat keep-alive, disconnection reconnection, and traffic shaping, and provides a simple topic subscription-publishing interface to the outside world.

[0057] Various terminals and modules connected to the system only need to subscribe to data topics of interest, such as "real-time status of agricultural machinery", "current operation tasks", "soil moisture" and "abnormal alarms", and they can automatically receive push messages for the corresponding data without having to worry about the underlying transmission protocols, connection management and other technical details, which greatly reduces the complexity of system integration.

[0058] This communication library has been specifically optimized for weak network environments in the fields. The heartbeat keep-alive mechanism can periodically check the link connectivity, the disconnection reconnection mechanism adopts an exponential backoff strategy, automatically rebuilds the connection and resends cached data after the network recovers, and the traffic peak shaping mechanism can cache and smoothly send sudden high-frequency data to avoid network congestion.

[0059] Tests have shown that the end-to-end response latency of system commands based on this communication library is no higher than 50ms, which can meet the stringent requirements of unmanned field operations for control timeliness; the reporting cycle of vehicle-mounted core status data is configurable, with a default value of no more than 500ms, ensuring that the monitoring terminal can obtain the real-time operating status of agricultural machinery.

[0060] The vehicle-mounted display and interaction terminal uses a ruggedized industrial tablet computer, designed specifically for complex field working environments. The screen is a 10.1-inch high-definition touch screen that supports touch operation while wearing gloves. The terminal is equipped with a customized dual system of Linux and Android. The Linux system is used to run real-time control-related programs, while the Android system is used to run interactive interface applications. The two systems communicate with each other through an internal communication interface. The terminal is equipped with multiple communication interfaces such as 4GLTE, Wi-Fi, CAN, and RS485 to adapt to the connection needs of different peripherals; the whole machine has a protection level of no less than IP54, which can effectively block dust and splashing water vapor in the field, and the working temperature range covers -20℃ to +70℃, which can adapt to the low temperature in the north and the high temperature in summer. At the same time, it has excellent vibration resistance and can withstand the continuous vibration during tractor operation. Its main hardware specifications are shown in Table 2.

[0061] Table 2 Main Specifications of Vehicle-Mounted Display and Interactive Terminal

[0062] Its display interface adopts an all-blue, tech-style design, with the overall layout divided into four functional areas: The top status bar displays real-time meteorological and environmental data, including wind direction, relative humidity, atmospheric pressure, light intensity, minute rainfall, and current weather status icons, helping operators to keep abreast of changes in the field environment and predict the impact of the environment on the quality of operations. The agricultural machinery status area on the left uses a 3D model of a tractor as the visual core. The model can move in sync with the vehicle's steering and suspension lifting. Around the model, the steering angle value, current driving speed, current working mode, PTO power output status, suspension height percentage, and the pressure and working status of the three sets of oil circuits are displayed in a clear and intuitive way, presenting the vehicle's core operating parameters. The soil moisture zone on the right uses a layered bar chart to visualize the temperature and humidity data of the upper, middle and lower soil layers. The data is collected by soil moisture sensors deployed in the field and reported through the environmental sensing data interface. The default reporting cycle is once every 10 minutes. Operators can intuitively view the soil moisture conditions of different soil layers. The lower right operation information area displays key agronomic parameters such as the current operation width, operation efficiency, and actual operation depth in real time. It also supports operators to edit and modify parameters such as the target operation depth online. After the modification is confirmed, the parameters are sent to the control module in real time to realize the dynamic on-site control of operation parameters.

[0063] The mobile client is a native application developed based on the Android system. The software architecture adopts the MVVM architecture pattern, uses the AndroidJetpack component library to achieve responsive UI and lifecycle management, the core development language is Kotlin, local persistent storage uses the ObjectBox high-performance database, and integrates a professional map SDK to realize map visualization function, with a smooth interactive experience and stable running performance.

[0064] This client supports simultaneous management of multiple agricultural machines, and its core functions are divided into five modules: The first is the operation monitoring and remote dispatch module, which uses a map view as the core display interface. It can simultaneously display the current location, operation status, operation plot boundary and task progress percentage of multiple agricultural machines, and support managers to remotely issue operation task instructions, adjust operation parameters and control vehicle start and stop. Second, the agricultural machinery and implement status monitoring module receives real-time agricultural machinery operation data and implement status data reported by the vehicle system. It can view detailed parameters such as speed, rotation speed, depth, and fault codes, and supports custom parameter thresholds. When the data exceeds the preset threshold, it will automatically push a warning notification. Thirdly, the soil moisture and meteorological data display module receives field sensor data through the environmental perception data interface and displays the historical trends of soil temperature and humidity and meteorological environment in the form of line graphs, supporting the viewing of statistical data by daily, weekly and monthly dimensions. Fourth is the path trajectory visualization module, which uses Canvas to draw the agricultural machinery operation path trajectory. The trajectory color can distinguish the operation status, supports historical trajectory playback and operation deviation analysis, and can count the operation area and operation mileage. Fifth is the abnormal alarm and maintenance record module, which centrally displays all alarm information and processing status, automatically records the maintenance history of agricultural machinery, supports maintenance reminder settings, and assists in the health management of agricultural machinery throughout its entire life cycle.

[0065] The system's positioning module adopts a fusion positioning algorithm that combines RTK+GNSS dual-frequency differential positioning with vehicle-mounted inertial navigation. It eliminates the system error of satellite positioning through RTK differential correction and performs fusion calculation by combining high-frequency attitude and acceleration data from the inertial navigation system to output high-frequency, high-precision position and attitude information.

[0066] In open field environments with good satellite signals, the system's positioning accuracy can reach the centimeter level, and the measured track deviation in the field is better than ±2.5cm, which can fully meet the requirements of navigation accuracy for precision operations of open-field vegetables. In scenarios where satellite signals are temporarily blocked by trees, overpasses, etc., the inertial navigation fusion algorithm can maintain positioning accuracy for a short period of time and avoid positioning jumps affecting the quality of operations.

[0067] The system supports OTA (Over-The-Air) download technology, which can remotely update the vehicle system software, various control algorithm modules, and terminal applications. It supports differential upgrade packages to reduce traffic consumption. The upgrade process adopts an A / B partition scheme, and can automatically roll back if the upgrade fails. The system upgrade and function iteration can be completed without returning the agricultural machinery to the factory, which greatly reduces equipment maintenance costs and upgrade cycles.

[0068] This system was designed with openness and scalability in mind, and has multi-dimensional compatibility.

[0069] In terms of agricultural machinery compatibility, through the ISOBUS / CAN standard protocol adaptation layer, the system can be compatible with tractors and supporting implements of mainstream brands on the market, without relying on specific hardware manufacturers, and without needing to redevelop the core control logic for different brands. At the level of new machinery access, new agricultural machinery only needs to provide standardized machinery description files according to the machinery information model defined by the system, specifying geometric features, dynamic characteristics, working capacity parameters, etc., and system integration can be completed without modifying the upper-level application code, thus enabling the rapid access of new machinery. At the platform integration level, the cloud interface has reserved the ability to connect with third-party agricultural management platforms, which can support data exchange with third-party systems such as the Ministry of Agriculture and Rural Affairs' agricultural mechanization management system and farm management platform, so as to realize the unified collection and analysis of operational data. In terms of operation and maintenance expansion, in addition to OTA remote updates, the system also supports remote operation and maintenance, remote log capture, and remote parameter debugging capabilities, which can significantly reduce field operation and maintenance costs.

[0070] Example 1

[0071] To verify the actual performance and reliability of the system, a comprehensive performance test was conducted on typical open-field vegetable planting plots. The test was divided into two dimensions: comprehensive field test and communication layer test.

[0072] The field comprehensive test selected three typical composite operation conditions: tillage, ridging, and sowing. The test indicators included navigation and positioning accuracy, path planning capability, obstacle avoidance response capability, machine attachment and switching latency, and interactive system response speed. The communication layer-specific test builds a CS architecture stress test script to conduct high-concurrency request tests, weak network environment signal mutation tests, and abnormal disconnection and intermittent recovery tests on wireless communication links.

[0073] This test was repeated 10 times in parallel, and the test indicators are summarized in Table 3.

[0074] Table 3 Summary of various test indicators

[0075] After testing and verification, this system meets the technical requirements for unmanned operation of open-field vegetables in terms of core indicators such as communication reliability, system stability, real-time response, and positioning accuracy.

[0076] The data packet loss rate was 0% during a 20-hour continuous stress test, proving that the heartbeat keep-alive and disconnection reconnection mechanisms of the self-developed message communication library can effectively ensure the integrity of data transmission in complex field network environments. The system ran at full load for 20 consecutive hours without program crashes, indicating that the system software architecture is stable, resource scheduling is reasonable, and it has the robustness for long-term continuous operation. The centimeter-level positioning accuracy and millisecond-level command response can support the needs of precision operations for open-field vegetables, ensuring the quality and efficiency of operations.

[0077] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control system for unmanned tractors and multi-implant collaborative operation of open-field vegetables, characterized in that: It includes an on-board control unit mounted on a tractor, various types of work implements that can be attached, and a remote monitoring terminal. Its characteristic is that the system adopts a distributed architecture that coordinates the equipment end, the cloud end, and the display end. The device uses an on-board computing unit as its core hardware carrier. The on-board computing unit runs an intelligent agricultural machinery operating system. The intelligent agricultural machinery operating system has a built-in communication framework, a multi-task scheduling module, and a task logic control module. It also connects the drive module, positioning module, perception module, decision-making module, planning module, and control module through a standardized data receiving and control interface. Each implement is equipped with an independent implement controller and establishes data communication with the on-board computing unit through a bus. The cloud service module deployed in the cloud is bidirectionally connected to the device end and the display end through a wireless communication link, and undertakes the functions of multi-agricultural machinery data aggregation, message forwarding, anomaly monitoring and unified task scheduling. The display terminal includes two types of human-computer interaction carriers: vehicle-mounted display and interactive terminal and mobile client, which are used for visual display of work status, setting of work parameters and remote monitoring and intervention. The intelligent agricultural machinery operating system has a built-in implement matching and adaptation unit and a collaborative control unit. The implement matching and adaptation unit is used to automatically identify the implement type and match the corresponding control parameter template after the implement is attached. The collaborative control unit is used to call the corresponding control algorithm according to the implement type and dynamically reconstruct the control link when the operation mode is switched, so as to realize the coordinated control of tractor operating parameters and implement operation parameters.

2. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The implement controller has pre-stored and solidified implement description information, which includes implement model, connection type, geometric parameters, and rated operating parameter range. After the implement is attached, the implement controller automatically reports the implement description information to the vehicle computing unit via the ISOBUS / CAN bus. After the implement attachment adaptation unit parses and obtains the implement type, it retrieves the corresponding control algorithm module from the pre-stored implement parameter template library, automatically completes parameter adaptation, and realizes plug-and-play functionality for the implement.

3. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The collaborative control unit classifies the working implements into two categories based on their connection method: three-point suspension type and towing type, and binds differentiated control algorithms to each category. For three-point suspension implements, the suspension height is the core control variable. A feedforward-feedback composite control strategy based on the linkage between tractor speed and tillage / sowing depth is adopted. The working height of the three-point suspension mechanism is adjusted in real time through the electro-hydraulic suspension controller. For towed implements, the magnitude of traction force, yaw compensation, and implement attitude are the control objects. A speed-traction force coordinated control algorithm is adopted to suppress traction overload and implement swaying.

4. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, When the operation mode is switched, the collaborative control unit sends a control algorithm switching instruction to the execution control layer and synchronously transmits the target operation quality index parameters, including sowing depth, plant spacing and operation width. Only after receiving a response signal indicating that the new algorithm module has been successfully bound can the execution control layer initiate the control logic for the corresponding work mode, ensuring the safety and parameter consistency of the mode switching process.

5. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The standardized data receiving and control interface defines three types of standardized data exchange interfaces. All three interfaces use JSON format for data serialization and implement data interaction through a topic subscription-publishing transmission mechanism. The first type is the job task data interface, which includes fields such as task ID, job type, job location, job parameters, task status and progress percentage. It is used to unify the description format of different types of job tasks and support the seamless scheduling of multiple types of job tasks. The second category is agricultural machinery status data interface, which includes fields for tractor real-time speed, engine speed, fault codes, implement suspension height and working depth, used to support remote vehicle diagnosis and status warning; The third category is the environmental perception data interface, which includes stratified soil temperature and humidity, meteorological data and air pressure fields, providing a standardized environmental data foundation for precision agriculture decision-making.

6. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The vehicle-mounted computing unit adopts a layered software architecture, which consists of a UI interface layer, a local service layer, a core functional module group, a support module, and an agricultural implement controller side unit from top to bottom. The UI interface layer corresponds to the human-computer interaction layer of the vehicle-mounted display touch terminal, and is used for visualizing the work status, setting parameters, and switching work modes. The local service layer is divided into internal local services and external local services, which are responsible for internal data bus communication between in-vehicle functional modules and network data interaction with external cloud and mobile terminals, respectively. Both rely on local databases to achieve data persistence. The core functional module group includes an agricultural machinery positioning module, a drive module, a fusion positioning module, an AI tillage module, an automatic ridging module, an automatic tillage module, a remote driving module, and a tracking operation module. The support modules include operation and maintenance management, log management, system monitoring, anomaly handling, remote update and remote monitoring modules, which provide basic guarantees for system operation; The implement controller side unit is divided into three categories according to the type of implement: tillage implements, transplanting implements and other implements. Each type of implement is equipped with an implement control module, a sensor driver and protocol transmission module, and completes data interaction with the vehicle system through a preset communication protocol.

7. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The system uses a self-developed message communication library as the communication middleware throughout the entire link. The message communication library encapsulates underlying network connection management, message serialization, heartbeat keep-alive, disconnection reconnection and traffic peak shaping logic. It provides a simple topic subscription-publishing interface to the outside world, and the system command response latency is no higher than 50ms, and the reporting cycle of vehicle status data is no greater than 500ms.

8. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The vehicle-mounted display and interaction terminal is a ruggedized industrial tablet computer with a 10.1-inch touchscreen. It is equipped with a customized Linux and Android dual system, and features 4G LTE, Wi-Fi, CAN, and RS485 communication interfaces. Its protection level is no less than IP54, and its operating temperature range is -20℃ to +70℃. Its display interface is divided into four functional areas. The top status bar is used to display meteorological and environmental data in real time. The left agricultural machinery status area uses a tractor 3D model as the core to display steering angle, vehicle speed, operation mode, PTO status and suspension height parameters. The right soil moisture area uses a layered bar chart to display the temperature and humidity data of the upper, middle and lower soil layers. The lower right operation information area is used to display and support editing of operation width, operation efficiency and operation depth parameters.

9. The unmanned tractor and multi-implementation collaborative control system for open-field vegetables according to claim 1, characterized in that, The mobile client is an application developed based on the Android system, adopting the MVVM architecture pattern and integrating a map SDK and a local persistent database. The mobile client supports simultaneous access for multiple agricultural machines and features functions such as operation monitoring and remote scheduling, real-time monitoring and threshold warning of agricultural machinery and implements status, display of soil moisture and meteorological data trends, visual playback of operation path trajectories, abnormal alarm push notifications, and maintenance record management.

10. The unmanned tractor and multi-implementation collaborative operation control system for open-field vegetables according to claim 1, characterized in that, The positioning module adopts a fusion positioning algorithm that combines RTK+GNSS dual-frequency differential positioning with vehicle-mounted inertial navigation. The actual measured track deviation in the field is better than ±2.5cm, and the system supports OTA remote updates of the vehicle system software and control algorithm module.

Images