A real-time adjusting control system and method for operation response parameters of a corn detassling machine

Abstract

The present application relates to the technical field of agricultural machinery automation control, and particularly relates to a corn detasseling machine operation response parameter real-time adjustment control system and method, comprising: a dynamic sensing module, a plurality of multi-axis mechanical arms and end effectors thereof, and a predictive central processing and control module. The dynamic sensing module is used to obtain the real-time kinematic state of the top growth point of the target plant in a disturbed environment; the predictive central processing and control module is configured to: based on the real-time kinematic state, predict the motion range of the top growth point of the target plant in a future preset time, and based on the predicted motion range, compensatively control the action of the multi-axis mechanical arm to perform a grabbing and removing operation on the top growth point. The system changes from "passive following" to "active prediction", can effectively overcome system delay, realize accurate positioning of a high-speed swinging target, and improve work efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of agricultural machinery automation control, and in particular to a system and method capable of adapting to disturbance environments such as strong winds, achieving precise and efficient removal of corn tassels through dynamic prediction and adaptive control. BACKGROUND

[0002] Corn detasseling is a key link in corn hybrid seed production, and its purpose is to remove the tassels of female plants to prevent self-pollination and ensure seed purity. In actual production, corn detasseling mainly relies on manual or traditional mechanical detasseling machines.

[0003] In recent years, with the development of robotics and artificial intelligence, agricultural robots that use visual perception for precise identification and detasseling have emerged. However, these robots are mostly applied in static or quasi-static environments, and their operating objects are usually relatively stationary. Since the corn tassel is a lightweight, flexible rod structure, it will produce high-frequency, irregular and large-amplitude swings in natural wind fields and other disturbance environments. The existing control strategies for agricultural robots are mostly based on real-time error feedback, and the system response speed lags far behind the rapid swings of the target, resulting in positioning failure and missed grabbing, and making it impossible to effectively work in real and variable field environments.

[0004] Therefore, how to solve the problem of precise identification, stable tracking and reliable operation of flexible and dynamic targets in a strong disturbance environment is a technical bottleneck that needs to be solved to realize automatic and intelligent corn detasseling. SUMMARY

[0005] The present application aims to solve the problems in the background art and provides a corn detasseling machine operation response parameter real-time adjustment control system and method. The system can significantly improve the success rate and efficiency of automatic removal of the top growth point of corn in a disturbance environment such as a wind field through dynamic prediction and adaptive control.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] A corn detasseling machine operation response parameter real-time adjustment control system, comprising a dynamic perception module, a plurality of multi-axis robot arms and their end effectors, and a predictive central processing and control module. The dynamic perception module is used to obtain the real-time kinematic state of the target plant top growth point in a disturbance environment; the predictive central processing and control module is configured to predict the motion range of the target plant top growth point within a predetermined time in the future based on the real-time kinematic state, and to compensatorily control the action of the multi-axis robot arm based on the predicted motion range to perform grabbing and removal operations on the top growth point.

[0008] Further, the dynamic perception module comprises a three-dimensional vision sensor and a three-dimensional wind speed and direction meter.

[0009] Further, the predictive central processing and control module extracts the motion trajectory, speed, acceleration, dominant vibration frequency and amplitude of the target by calculating and spectrum analyzing the continuous multi-frame pose information collected by the three-dimensional vision sensor.

[0010] Further, the predictive central processing and control module establishes a dynamic prediction model for describing the motion of the target, and uses a state estimation algorithm such as Kalman filtering to fuse model prediction and sensor observation to generate an optimized prediction of the future motion range.

[0011] Further, the predictive central processing and control module uses a predictive servo control strategy to take a specific point in the optimized predicted motion range as the target instruction point of the motion of the mechanical arm.

[0012] Further, the end effector is a composite structure comprising a guide cover, an inflatable air bag and a negative pressure generation and control unit, and the target is adsorbed and actively stabilized by negative pressure airflow.

[0013] Further, the predictive central processing and control module controls the size of the negative pressure airflow in a closed loop according to the range value of the predicted motion range.

[0014] Further, the multi-axis mechanical arm integrates a force sensor for force feedback monitoring during adsorption to realize safety decision-making.

[0015] The present application also provides a control method corresponding to the above-mentioned system, comprising the steps of dynamic state acquisition, motion range prediction, mechanical arm compensatory control, active stabilization and adsorption, and removal operation.

[0016] The present application has the advantages that by introducing a compensatory control strategy based on dynamic prediction, the system changes from "passive following" to "active prediction", which can effectively overcome system delay and realize accurate positioning of high-speed swinging targets; by designing a unique active stabilization end effector, the system can actively suppress target swing before grabbing through negative pressure, creating favorable conditions for reliable operation; through multi-sensor fusion and adaptive adjustment, the system can adapt to changing wind field environment in real time, significantly improving all-weather operation capability, successful de-male rate and operation efficiency, while reducing damage to plants. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows.

[0018] Figure 1 The overall structure schematic diagram provided for the embodiment 1 of the present application.

[0019] Figure 2 The control method flow chart diagram provided for the embodiment 2 of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application.

[0021] Embodiment one

[0022] Referring to Figure 1 The corn detasseling machine operation response parameter real-time adjustment control system provided in the embodiment is arranged on a four-wheel driven high wheelbase mobile platform 1, and the platform can adapt to the field topography in the late growth period of corn. The platform is also equipped with a laser radar and a GPS autonomous navigation system.

[0023] Referring to Figure 1 The present application comprises a dynamic perception module 2, one or more multi-axis mechanical arms 3, an end effector 4 connected to the end of the mechanical arm, and a predictive central processing and control module 5.

[0024] The dynamic perception module 2 comprises:

[0025] At least one three-dimensional vision sensor 21: for example, an Intel RealSense D435i depth camera, working at 60 fps or higher frame rate, installed at a high position of the mobile platform 1, used to obtain the three-dimensional space coordinates (X, Y, Z) and attitude (pitch, yaw, roll) of the target corn top growth point (i.e. the core male ear) in real time.

[0026] An environmental sensor 22: at least one three-dimensional ultrasonic anemometer, installed on the top of the mobile platform 1, used to measure the instantaneous wind speed and direction of the operation point in real time.

[0027] The end effector 4 is a composite structure, comprising:

[0028] A guide cover 41: a downwardly open horn-shaped cover made of light high molecular material, with smooth inner wall, used to preliminarily physically guide and correct the swaying corn tip when the mechanical arm approaches.

[0029] An inflatable air bag 42: an annular air bag made of silica gel is arranged around the inside of the guide cover 41. The air bag is connected to a micro air pump.

[0030] The negative pressure generation and control unit 43 controls a high-flow turbine fan that communicates with the central region of the guide hood 41 through a pipe. A high-speed electromagnetic valve is provided on the pipe. The edge of the central opening of the guide hood, i.e., the sealing opening, is provided with a flexible sealing lip.

[0031] The working logic of the predictive central processing and control module 5 is as follows:

[0032] First step: dynamic state acquisition and calculation

[0033] The controller receives continuous image frames from the three-dimensional visual sensor 21 in real time. The corn tip is identified by running an AI model, and the pose information of each frame is calculated in combination with the depth map. By performing a difference operation on the pose data of consecutive frames, the real-time motion trajectory, velocity, and acceleration vector of the target point are obtained. At the same time, the motion range, dominant vibration frequency, and maximum amplitude of the target in the current wind field are extracted by performing a fast Fourier transform on the pose time series of this period.

[0034] Second step: prediction and decision-making

[0035] A dynamic prediction model is established inside the controller, for example, the corn plant is regarded as a second-order damped vibration system driven by wind force (measured by an anemometer) and having a specific natural frequency. The specific natural frequency is obtained by fast Fourier transform analysis.

[0036] Further, an extended Kalman filter algorithm is used to fuse the prediction results of the dynamic model with the next frame of actual observation results of the visual sensor. This process aims to filter out sensor noise and output a more smooth and accurate optimized prediction of the target motion range within the next 100-200 milliseconds.

[0037] Preferably, the controller adopts a predictive servo control strategy. The controller directly commands the robotic arm to track the "predicted position after 150 milliseconds". The specific target point can be set as the point in the predicted motion range closest to the end effector.

[0038] Third step: adaptive negative pressure adjustment

[0039] When the controller predicts that the robotic arm is about to reach the target position, the range value (i.e., the amplitude) of the predicted motion range is evaluated.

[0040] If the amplitude is less than a preset threshold, it indicates that the wind disturbance is small, and the predicted motion range is small, so only a small negative pressure is needed to capture the core male ear. At this time, the controller will control the turbine fan of the negative pressure generation unit 43 to reduce the output power. It should be noted that the output power of the turbine fan is always greater than the output power of the turbine fan required to pull the core male ear off the main stem.

[0041] If the amplitude is greater than the preset threshold, it indicates that the wind disturbance is severe, the predicted movement range is large, and a larger negative pressure is needed to capture the core tassel. At this time, the controller will control the turbine fan of the negative pressure generation unit 43 to increase the output power.

[0042] When the guide cover 41 is located at the predicted movement range position, the high-speed valve is opened instantaneously, generating a powerful negative pressure airflow that "sucks" the swinging corn tip to and stabilizes it in the central area of the guide cover. The size of the negative pressure is determined by the evaluation value of the movement range, the larger the amplitude, the higher the turbine fan output power, and the larger the negative pressure generated to ensure effective and rapid grabbing of the corn tip.

[0043] At the same time, the micro air pump slightly inflates the inflatable air bag 42, causing it to expand and fill the partial gap between the inner wall of the guide cover and the corn tip. This is not for clamping, but to narrow the air leakage channel, greatly enhancing the sealing of the central area, thereby significantly improving the adsorption and stabilization effect of the negative pressure. The inflatable air bag 42 is connected to a solenoid valve for air exhaust of the inflatable air bag 42 after the removal action is completed.

[0044] Step 4: Force feedback

[0045] The multi-axis robot arm 3 integrates a force sensor. During the negative pressure adsorption process, the controller monitors the force sensor readings in real time.

[0046] In the initial stage, the force reading is small.

[0047] When the corn tip is successfully adsorbed and forms a good fit with the sealing port, the force sensor will detect a significant, suction-induced pull force rise.

[0048] The controller sets a preset threshold. When the force feedback data exceeds the threshold, the controller confirms that the adsorption is successful and can trigger the next removal operation.

[0049] If the force feedback does not reach the threshold for a long time, or the force feedback is abnormally large, the system determines that this attempt has failed, releases the negative pressure, and controls the robot arm to retreat and find the next target, avoiding dead loops and damage to the equipment.

[0050] Example Two

[0051] This embodiment provides a real-time adjustment control method for corn detasseling machine operation response parameters based on the system described in Example One. Referring to Figure 2 , the method comprises the following steps:

[0052] Step S101: Dynamic state acquisition. After the system is started, the mobile platform 1 travels in the field, and the dynamic perception module 2 continuously scans and acquires the real-time kinematic state of the target corn top growth point in front.

[0053] Step S201: Motion Range Prediction. The central processing and control module 5 processes the acquired state data and predicts the motion range of the target in a short time in the future, as described in Embodiment One.

[0054] Step S301: Compensatory Control of the Manipulator. Based on the prediction result, the controller controls the manipulator 3 with the end effector 4 to approach the target in a high speed in an early predicted manner, using a predictive servo strategy.

[0055] Step S401: Active Stabilization and Adsorption. After approaching the target, according to the predicted motion range uncertainty, it is decided to start a negative pressure airflow and airbag gap adjustment of how much intensity to actively suppress and adsorb the target.

[0056] Step S501: Removal Operation. After confirming the successful adsorption through the force sensor, the removal action is performed. This action can be: a) further increasing the negative pressure pulse to directly pull the tassel into the suction; b) while maintaining the adsorption, performing a fast, spiral upward pulling action by the manipulator.

[0057] The above description is only a preferred embodiment of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A real-time adjustment control system for operating response parameters of a corn detassling machine, characterized by, The application relates to a system for automatically picking and removing the top growing point of a target plant in a disturbed environment, comprising: a dynamic sensing module for acquiring the real-time kinematic state of the top growing point of the target plant in the disturbed environment; one or more multi-axis robot arms, the ends of which are connected with end effectors; and a predictive central processing and control module, which is configured to: predict the motion range of the top growing point of the target plant in a preset time in the future based on the real-time kinematic state; and compensatively control the action of the multi-axis robot arm based on the predicted motion range, so as to perform picking and removing operations on the top growing point in the disturbed environment; a three-dimensional vision sensor for acquiring real-time pose information of the top growing point of the target plant in three-dimensional space, wherein the pose information comprises position coordinates and attitude angles; and an environment sensor for acquiring physical parameters of the surrounding environment during system operation; the environment sensor is a three-dimensional wind speed and direction instrument for measuring wind speed and direction parameters in real time; the predictive central processing and control module is configured to: calculate the real-time kinematic parameters of the top growing point of the target plant based on the multi-frame pose information collected by the three-dimensional vision sensor at continuous time points, wherein the kinematic parameters comprise a motion trajectory, a speed and / or an acceleration; perform frequency spectrum analysis on the time sequence of the pose information, and extract the motion range, a dominant vibration frequency and an amplitude of the top growing point of the target plant in the current wind field environment; the predictive central processing and control module is further configured to: establish a dynamic prediction model for describing the motion of the top growing point of the target plant; and use a state estimation algorithm to fuse the predicted state of the dynamic prediction model and the actual observation state of the three-dimensional vision sensor, so as to generate an optimized prediction of the motion range of the top growing point of the target plant in a preset time period in the future; when controlling the multi-axis robot arm, the predictive central processing and control module uses a predictive servo control strategy, takes the position point closest to the end effector in the optimized predicted motion range as a target instruction point of the motion of the multi-axis robot arm, and compensates for the response delay of the system; a guide cover for guiding and correcting the target attitude; one or more inflatable air bags made of flexible material arranged in the guide cover, the air bags realize surrounding gap adjustment on the top growing point of the target plant through inflation, so as to improve the adsorption force of negative pressure on the target plant; and a negative pressure generation and control unit coaxially arranged with the guide cover, the negative pressure generation and control unit is configured to: when the guide cover approaches a preset negative pressure adsorption range of the top growing point of the target plant, generate a controlled negative pressure airflow to adsorb and stabilize the top growing point of the target plant in the central region of the guide cover, so as to inhibit the swing of the top growing point of the target plant caused by external airflow disturbance; the predictive central processing and control module is further configured to: evaluate the range value of the optimized predicted motion range in real time; and control the generation of an adaptive negative pressure airflow that can adsorb the top growing point of the target plant at any position point in the optimized predicted motion range.

2. The system of claim 1, wherein, The state estimation algorithm is a Kalman filtering algorithm or an extended form thereof.

3. The system of claim 1, wherein, The multi-axis robot arm is integrated with a force sensor, and the predictive central processing and control module is further configured to: acquire force feedback data of the force sensor in real time during the process that the end effector adsorbs the target plant top growing point; and adjust the multi-axis robot arm with the end effector to approach the next target plant top growing point in real time when the force feedback data value exceeds a preset threshold.

4. A method for real-time adjustment control of operation response parameters of a corn detassling machine, which employs the real-time adjustment control system of operation response parameters of a corn detassling machine according to claim 1, characterized in that, The method comprises the following steps: Step S101: dynamic state acquisition step: acquiring kinematic state of the target plant top growing point in a disturbed environment in real time through a dynamic perception module deployed on a mobile platform; Step S201: motion range prediction step: predicting motion range of the target plant top growing point in a preset time in the future based on the real-time kinematic state by using a predictive central processing and control module; Step S301: robot compensatory control step: compensatory controlling action of one or more multi-axis robot arms based on the predicted motion range, so that an end effector connected to the multi-axis robot arm approaches the motion range; Step S401: active stabilization and adsorption step: after the end effector approaches the motion range, generating a controlled negative pressure airflow by the end effector to adsorb the top growing point of the target plant and inhibit swing of the top growing point caused by external disturbance; Step S501: removal operation step: after the adsorption is completed, removing the top growing point from the plant main stem through subsequent action of the end effector and / or the multi-axis robot arm.

Images