A digital program control system and method for a fruit bag gripping robot

Abstract

The application relates to the field of automatic control, and discloses a digital program control system and method for a fruit bag grabbing manipulator, which comprises a multi-source sensing signal capturing module, a stage identification and parameter generation module, an instruction generation module, a driving interface module and an operation abnormality sensing and warning module. The multi-source sensing signal capturing module obtains and integrates the image, air pressure, displacement and tactile pressure signals of a clamping jaw area; the stage identification and parameter generation module determines the bag taking, bag supporting and bag sleeving stages according to the signal characteristics and generates differentiated motion control parameters; the instruction generation module generates driving instructions, the driving interface module converts the driving instructions into execution signal outputs; and the operation abnormality sensing and warning module generates a warning signal according to preset conditions. The application solves the problems that the manipulator in an unstructured orchard relies on fixed coordinates and preset actions, lacks real-time sensing feedback, causes bag taking failure, bag supporting asymmetry and bag sleeving deviation, and improves operation accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of automation control, specifically to a digital program control system and method for a fruit bag grasping robot. Background Technology

[0002] In existing orchard fruit bagging operations, automated bagging robots have been initially applied. Their typical workflow involves a robotic arm or Cartesian coordinate robot carrying an end effector to grasp fruit bags from the storage area, perform a bag-opening action, and then place the bag over the target fruit, followed by sealing. Currently, most bagging robots employ a program control method based on fixed position coordinates and preset motion sequences. During the robot's installation and debugging phase, the spatial coordinates of the bag-retrieving point, bag-opening point, and bag-applying point are recorded via a teach pendant, and these coordinates are programmed into the robot's control program. In actual operation, the robot sequentially calls pre-set position commands and motion parameters according to the program's prescribed order, driving the movement of each joint or actuator. The advantage of this control method is its simple program structure, facilitating repetitive actions in structured environments. However, orchard environments have significant unstructured characteristics. Fruit tree canopies vary in shape, branches are randomly distributed, and factors such as fruit shading, changes in light intensity, and wind disturbance can cause unpredictable deviations in the actual position and storage posture of the fruit bags. Furthermore, minute dimensional differences during fruit bag manufacturing and slight displacement of the storage device due to mechanical vibration can lead to inconsistent execution results of the same control program at different times. Under the influence of these factors, existing fixed-coordinate-based control methods are prone to problems such as suction failure due to the suction cup not aligning with the center of the fruit bag during bag retrieval, incomplete bag opening due to a mismatch between the opening stroke of the grippers and the actual size of the bag, and misalignment or missing bag during bagging due to the bag opening deviating from the fruit's growth axis. Some improved solutions introduce visual positioning technology to assist in correcting the bag retrieval coordinates; however, these solutions typically only perform one-time position compensation before the bag retrieval action and do not incorporate multi-source sensor signals such as pressure and displacement into real-time closed-loop control. This results in the robotic arm being unable to respond promptly to abnormal states once the action begins, significantly limiting the continuity and success rate of operations. Summary of the Invention

[0003] In view of the shortcomings of the prior art, the purpose of this invention is to provide a digital program control system and method for a fruit bag grasping robot, which solves the problems of high bag grasping failure rate, asymmetrical bag opening, and bag offset caused by the reliance on fixed position coordinates and preset action sequences for operation in unstructured orchard environments.

[0004] This invention acquires image, air pressure, displacement, and tactile pressure signals through a multi-source sensor signal capture module. A stage identification and parameter generation module automatically determines the current stage (bag removal, bag opening, or bagging) based on the temporal and amplitude characteristics of these signals, and generates differentiated motion control parameters for each stage. An instruction generation module generates drive instructions based on the stage status, employing a smooth transition mechanism during stage switching. A drive interface module converts these instructions into drive signals for joints or end effectors. An operation anomaly detection and alarm module analyzes real-time changes in image grayscale, air pressure change rate, and tactile pressure distribution. Upon detecting bag removal failure, adsorption abnormalities, or asymmetrical bag opening, an alarm signal is generated, triggering a retry or parameter adjustment. This ensures the continuity of fruit bagging operations without relying on fixed coordinates.

[0005] This invention provides a digital program control system for a fruit bag gripping robot, comprising:

[0006] The multi-source sensor signal capture module is used to acquire image signals, air pressure signals, displacement signals and tactile pressure signals of the gripper area, and integrate them to form multi-source sensor signals.

[0007] The stage identification and parameter generation module receives multi-source sensor signals and determines the current operation stage of the robot based on the time domain and amplitude characteristics of each signal. The operation stage includes at least the bag picking stage, the bag opening stage, and the bag putting stage. Different motion control parameter signals and stage status signals are generated for different operation stages.

[0008] The instruction generation module receives motion control parameter signals and stage status signals, and generates robot drive instruction signals based on the motion control parameters corresponding to the current operation stage.

[0009] The drive interface module receives drive command signals from the robot arm, converts them into drive signals for each joint or end effector of the robot arm, and outputs them.

[0010] The operation anomaly perception and alarm module receives multi-source sensor signals output by the multi-source sensor signal capture module and robot drive command signals output by the command generation module, and generates operation anomaly alarm signals based on preset anomaly judgment conditions.

[0011] In one embodiment of the present invention, the multi-source sensor signal capture module includes an image acquisition device, a pressure sensor, a displacement sensor, and a tactile sensor array. The image acquisition device is fixed above the robot arm body or the working platform and is used to acquire continuous frame images of the robot arm gripper area before and after the bag-picking action and output image signals. The pressure sensor is installed in the air path connected to the bag-picking suction cup and is used to measure the air pressure value inside the air path at a preset sampling period and output air pressure signals. The displacement sensor is installed on the drive mechanism of the robot arm gripper and is used to detect the opening distance of the gripper during the bag-opening stage and output displacement signals. The tactile sensor array is attached to the inner surface of the robot arm gripper and is used to sense the pressure distribution generated when the gripper contacts the fruit bag and output tactile pressure signals. The multi-source sensor signal capture module filters and performs analog-to-digital conversion on the image signal, pressure signal, displacement signal, and tactile pressure signal respectively, and then encapsulates them into a multi-source sensor signal containing timestamp information and transmits it to the stage identification and parameter generation module.

[0012] In one embodiment of the present invention, after receiving multi-source sensor signals, the stage identification and parameter generation module first extracts the time-domain features of the air pressure signal. When the value of the air pressure signal is consistently lower than a first negative pressure threshold for a first duration, it determines that the robotic arm has entered the bag-retrieving stage and generates a bag-retrieving stage status signal. Simultaneously, it reads a subset of motion control parameters corresponding to the bag-retrieving stage from a preset parameter storage area. The subset of motion control parameters corresponding to the bag-retrieving stage includes the upper limit value of the motion speed and the upper limit value of the acceleration during the bag-retrieving stage. After the bag-retrieving stage ends, the stage identification and parameter generation module extracts the time-domain features of the displacement signal. When the value of the displacement signal reaches a first displacement threshold, it determines that the robotic arm has entered the bag-holding stage. The system generates a bag-holding stage status signal and reads a subset of motion control parameters corresponding to the bag-holding stage from the preset parameter storage area. This subset includes the upper limit of the bag-holding stage motion speed and the upper limit of the bag-holding stage clamping force. After the bag-holding stage ends, the stage identification and parameter generation module extracts the time-domain features of the displacement signal. When the value of the displacement signal reaches the second displacement threshold, the system determines that the robot has entered the bag-making stage and generates a bag-making stage status signal. At the same time, the system reads a subset of motion control parameters corresponding to the bag-making stage from the preset parameter storage area. This subset includes the upper limit of the bag-making stage motion speed and the upper limit of the bag-making stage end effector feed speed.

[0013] In one embodiment of the present invention, during the bag-removing stage, the instruction generation module receives the bag-removing stage status signal and a subset of motion control parameters corresponding to the bag-removing stage, and generates a bag-removing drive instruction signal. The bag-removing drive instruction signal includes a suction cup adsorption start instruction and a first motion trajectory instruction for the robotic arm end effector to move towards the storage bag area. During the bag-opening stage, the instruction generation module receives the bag-opening stage status signal and a subset of motion control parameters corresponding to the bag-opening stage, and generates a bag-opening drive instruction signal. The bag-opening drive instruction signal includes a second motion trajectory instruction to control the two grippers to open outward at a speed not exceeding the upper limit of the movement speed during the bag-opening stage, and a bag-opening deceleration instruction triggered when the tactile pressure signal reaches the upper limit of the clamping force. During the bagging stage, the instruction generation module receives the bagging stage status signal and a subset of motion control parameters corresponding to the bagging stage, and generates a bagging drive instruction signal. The bagging drive instruction signal includes a third motion trajectory instruction to control the robotic arm end effector to move the opened bag opening towards the fruit, and a bagging stop instruction triggered when the tactile pressure signal detects that the fruit bag is in contact with the fruit surface.

[0014] In one embodiment of the present invention, when the instruction generation module receives a stage state signal output by the stage identification and parameter generation module and a switch occurs, it obtains the end motion speed value of the current operation stage as the starting speed value and obtains the preset initial motion speed value of the target operation stage as the ending speed value. Within a preset transition time, it performs linear interpolation on the starting speed value and the ending speed value to generate a transition speed curve. The instruction generation module generates a transition drive instruction signal based on the transition speed curve. The transition drive instruction signal replaces the robot arm drive instruction signal and is transmitted to the drive interface module within the transition time. The value of the transition time is determined based on the difference in motion speed between the current operation stage and the target operation stage. When the transition time ends, the instruction generation module generates a robot arm drive instruction signal according to the subset of motion control parameters corresponding to the target operation stage.

[0015] In one embodiment of the present invention, during the bag-retrieving stage, the operation anomaly perception and alarm module receives the image signal output by the multi-source sensor signal capture module, calculates the grayscale change feature value of the gripper area in the preset frame image after the bag-retrieving action is executed, and if the grayscale change feature value is lower than the preset grayscale change threshold, the bag-retrieving action is determined to have failed and a bag-retrieving failure alarm signal is generated. The bag-retrieving failure alarm signal is transmitted to the instruction generation module, and the instruction generation module generates a bag-retrieving retry drive instruction signal in response to the bag-retrieving failure alarm signal, so that the robot arm re-executes the bag-retrieving action; if the number of consecutive bag-retrieving failure alarm signals reaches the preset retry limit value, the operation anomaly perception and alarm module generates a bag-retrieving failure shutdown alarm signal and suspends the robot arm operation.

[0016] In one embodiment of the present invention, the operation anomaly perception and alarm module calculates the grayscale change feature value as follows: In the first frame image before the bag-removal action is performed, a rectangular preset calibration area of ​​a preset size is defined at the front end of the gripper. The average grayscale value of all pixels within this rectangular preset calibration area is calculated as the first grayscale mean. In the second frame image after the bag-removal action is performed, a rectangular preset calibration area of ​​the same position and size is defined. The average grayscale value of all pixels within this rectangular preset calibration area is calculated as the second grayscale mean. The absolute value of the difference between the first grayscale mean and the second grayscale mean is used as the grayscale change feature value. The formula for calculating the grayscale change feature value is as follows:

[0017] ;

[0018] in, Here, N represents the grayscale variation characteristic value, and N is the total number of pixels in the preset calibration area of ​​the first frame. M represents the grayscale value of the i-th pixel in the first frame, and M represents the total number of pixels in the preset calibration area of ​​the second frame. The grayscale value of the j-th pixel in the second frame. This is the grayscale correction factor. This is the grayscale offset compensation amount.

[0019] In one embodiment of the present invention, during the bag removal stage, the operation anomaly detection and alarm module receives the air pressure signal output by the multi-source sensor signal capture module, continuously acquires multiple air pressure values ​​at a preset sampling period, and calculates the air pressure change rate sequence. Each air pressure change rate in the air pressure change rate sequence is calculated by dividing the difference between the air pressure values ​​at two adjacent sampling times by the preset sampling period. If the air pressure signal value has reached the target negative pressure value, and the absolute value of each air pressure change rate in the air pressure change rate sequence exceeds the preset air pressure change rate threshold in subsequent consecutive sampling periods, the adsorption state is determined to be abnormal and an adsorption anomaly alarm signal is generated. The adsorption anomaly alarm signal is transmitted to the command generation module, and the command generation module generates an adsorption retry drive command signal in response to the adsorption anomaly alarm signal. The calculation formula for each air pressure change rate is as follows:

[0020] ;

[0021] in, Let $k$ be the rate of change of air pressure at the $k$-th sampling time. Let k be the air pressure value at time k. The air pressure value at time k-1. The preset sampling period is α, where α is the gain coefficient for the rate of change of air pressure. β represents the target negative pressure value, and β is the attenuation coefficient of the rate of change of air pressure.

[0022] In one embodiment of the present invention, during the bag-holding stage, the operation anomaly detection and alarm module receives a tactile pressure signal output by the multi-source sensor signal capture module. The tactile pressure signal includes the pressure values ​​of each tactile sensor unit on both sides of the robotic arm gripper. The operation anomaly detection and alarm module calculates the sum of the pressure values ​​of all tactile sensor units on the left gripper as the left pressure sum, calculates the sum of the pressure values ​​of all tactile sensor units on the right gripper as the right pressure sum, and calculates the absolute value of the difference between the left pressure sum and the right pressure sum as the pressure difference. If the pressure difference exceeds a preset difference threshold, the bag-holding is determined to be asymmetrical and a bag-holding asymmetry alarm signal is generated. The bag-holding asymmetry alarm signal is transmitted to the instruction generation module. The instruction generation module adjusts the driving parameters of both grippers in response to the bag-holding asymmetry alarm signal to reduce the pressure difference. The formula for calculating the pressure difference is as follows:

[0023] ;

[0024] in, The pressure difference is represented by A, which is the total number of tactile sensing units in the left gripper. B represents the pressure value of the a-th sensing unit on the left, and B represents the total number of tactile sensing units in the gripper on the right. This represents the pressure value of the b-th sensing unit on the right. This is the pressure normalization coefficient.

[0025] This invention also provides a method for a digital program control system for a fruit bag grasping robot, comprising:

[0026] S1: Acquire image signals, air pressure signals, displacement signals and tactile pressure signals of the gripper area, and integrate them to form multi-source sensing signals;

[0027] S2: Receive multi-source sensor signals, determine the current operation stage of the robot arm based on the time domain and amplitude characteristics of each signal, the operation stage includes at least the bag picking stage, the bag opening stage and the bag putting stage, and generate differentiated motion control parameter signals and stage status signals for different operation stages.

[0028] S3: Receive motion control parameter signals and stage status signals, and generate robot drive command signals according to the motion control parameters corresponding to the current operation stage;

[0029] S4: Receives the robot arm drive command signal, converts it into drive signals for each joint of the robot arm or the end effector, and outputs them;

[0030] S5: Receives multi-source sensor signals and robot drive command signals, and generates an operation abnormality alarm signal based on preset abnormality judgment conditions.

[0031] Beneficial Effects: The present invention provides a digital program control system and method for a fruit bag grasping robot. It acquires image, air pressure, displacement, and tactile pressure signals through a multi-source sensor signal capture module. A stage identification and parameter generation module automatically determines the current stage (bag retrieval, bag opening, or bagging) based on the time-domain and amplitude characteristics of these signals, and generates differentiated motion control parameters for each stage. An instruction generation module generates drive instructions based on the stage state, employing a smooth transition mechanism during stage switching. A drive interface module converts the instructions into drive signals for the joints or end effectors. An operation anomaly perception and alarm module analyzes real-time changes in image grayscale, air pressure change rate, and tactile pressure distribution. When bag retrieval failure, adsorption anomaly, or asymmetrical bag opening is detected, an alarm signal is generated, triggering a retry or parameter adjustment. This ensures the continuity of fruit bagging operations without relying on fixed coordinates. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a system architecture diagram of a digital program control system for a fruit bag grasping robot.

[0034] Figure 2 A schematic diagram illustrating the overall module interaction flow of the system;

[0035] Figure 3 A schematic diagram illustrating the stage recognition and motion control process of the fruit bag grasping robot;

[0036] Figure 4 A schematic diagram illustrating the abnormal operation detection and handling process of the fruit bag grasping robot;

[0037] Figure 5 This is a flowchart of a digital program control method for a fruit bag grasping robot. Detailed Implementation

[0038] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0039] Please see Figures 1-5 The image shows a digital program control system and method for a fruit bag grasping robot according to the present invention. The digital program control system for a fruit bag grasping robot of the present invention includes a multi-source sensor signal capture module for acquiring image signals, air pressure signals, displacement signals, and tactile pressure signals of the gripper area and integrating them to form a multi-source sensor signal; a stage identification and parameter generation module for receiving the multi-source sensor signals and determining the current operation stage of the robot based on the time domain and amplitude characteristics of each signal, wherein the operation stage includes at least a bag-picking stage, a bag-supporting stage, and a bag-attaching stage, and generating differentiated motion control parameter signals and stage status signals for different operation stages; an instruction generation module for receiving the motion control parameter signals and stage status signals, and generating robot drive instruction signals according to the motion control parameters corresponding to the current operation stage; a drive interface module for receiving the robot drive instruction signals, converting them into drive signals for each joint or end effector of the robot, and outputting them; and an operation anomaly perception and alarm module for receiving the multi-source sensor signals output by the multi-source sensor signal capture module and the robot drive instruction signals output by the instruction generation module, and generating an operation anomaly alarm signal based on preset anomaly judgment conditions.

[0040] like Figure 1 As shown, its overall architecture consists of five parts: a multi-source sensor signal capture module, a stage identification and parameter generation module, an instruction generation module, a drive interface module, and an operation anomaly perception and alarm module. These five modules are interconnected with a clear signal flow direction to jointly realize automatic perception, stage identification, action generation, and anomaly handling of the entire process of fruit bag grabbing.

[0041] like Figure 2The diagram shows the overall module interaction flowchart of the digital program control system for the fruit bag grasping robot. Each module works collaboratively to form a closed-loop control system. The multi-source sensor signal acquisition module is the core of the system's perception, responsible for collecting image signals, air pressure signals, displacement signals, and tactile pressure signals from the gripper's working area. It performs signal filtering, analog-to-digital conversion, and timestamp encapsulation processing, providing the system with real-time, accurate, multi-dimensional perception data, which forms the basis for all subsequent control logic. The stage identification and parameter generation module receives multi-source sensor signals and accurately identifies the robot's current working stage by analyzing the signal's time domain and amplitude characteristics. It then generates corresponding motion control parameters and status signals for that stage, enabling differentiated parameter configuration for different work stages and ensuring the robot's movements adapt to the actual needs of fruit bag grasping. The instruction generation module, as the system's control center, receives parameters and status signals from the stage identification and parameter generation module, combines them with work rules to generate standardized robot drive instructions, and simultaneously receives abnormal feedback signals and adjusts the instruction content in real time to ensure the rationality and safety of the control instructions. The drive interface module performs signal conversion, transforming control commands output by the command generation module into drive signals recognizable by the robot's joints and end effectors. This converts electrical signals into mechanical actions, precisely driving the robot to perform actions such as bag picking, bag opening, and bag placement. The robot's joints and end effectors are the system's execution terminals, performing specified mechanical actions based on the drive signals. Simultaneously, they transmit their own operating status and operational feedback information back to the multi-source sensor signal capture module, forming a closed loop of perception and execution. The operation anomaly perception and alarm module synchronously receives multi-source sensor signals and drive command signals. Based on preset judgment rules, it identifies operational anomalies and generates corresponding alarm signals. On one hand, it feeds the anomaly signal back to the command generation module, triggering command adjustments and action retry; on the other hand, it outputs a system alarm, indicating an abnormal operational status. The system alarm module visualizes anomaly information, allowing maintenance personnel to quickly grasp the fault situation. The entire process, through the design of forward control, reverse feedback, and anomaly loop, achieves stable, intelligent, and safe automated operation of the system.

[0042] The multi-source sensor signal acquisition module is the core component of the system for acquiring information about the external environment and the robot's own status. This module integrates four types of sensing elements: an image acquisition device, a pressure sensor, a displacement sensor, and a tactile sensor array. The image acquisition device is fixedly mounted on a bracket above the robot body or the working platform. Its field of view covers the working area of ​​the robot's grippers and the area near the bag outlet of the bag storage device, enabling it to capture local images of the grippers before and after the bag-retrieving action in continuous frames. The pressure sensor is connected in series through an air connector to the air duct between the bag-retrieving suction cup and the vacuum generator, measuring the absolute or relative pressure inside the air duct at a preset fixed sampling period. The displacement sensor is installed on the linear guide or rotary joint that drives the grippers to open and close, continuously outputting pulse signals or analog voltage signals proportional to the displacement during gripper movement. The tactile sensor array consists of multiple independent pressure-sensitive units arranged in rows and columns, attached to the inner surfaces of the grippers on both sides of the robot. Each pressure-sensitive unit can independently sense the magnitude of the normal pressure generated when it comes into contact with the fruit bag. The multi-source sensor signal acquisition module performs signal conditioning on the raw signals output from the four types of sensors. This includes noise reduction and brightness correction for image signals, low-pass filtering of air pressure signals to eliminate airflow pulsation interference, pulse counting and direction determination for displacement signals, and channel scanning and baseline calibration for tactile pressure signals. The conditioned signals are then uniformly converted from analog to digital and timestamp information of the synchronous sampling time is added, encapsulating them into a set of multi-source sensor signals with a unified data format. This multi-source sensor signal is simultaneously transmitted to the stage identification and parameter generation module and the operation anomaly perception and alarm module via the system's internal data bus or shared memory area, providing a unified data foundation for subsequent stage determination and anomaly detection.

[0043] like Figure 3The diagram shows the process flow for stage identification and motion control of the fruit bag grasping robot, including multi-branch decision-making and stage switching transition logic, adapting to the dynamic control requirements of multi-stage robot operations. Multi-source sensor signal input is the initial stage of the process, collecting four types of sensor data: image, air pressure, displacement, and tactile pressure, providing complete data support for stage identification and ensuring the accuracy and real-time nature of stage judgment. The stage identification and parameter generation module extracts and analyzes features from the input signals, accurately dividing the operation into stages based on key indicators such as signal value and duration, which is the core link in achieving staged control. Stage judgment is the core branch node in the process, dividing the operation state into three branches based on signal characteristics: bag retrieval stage, bag opening stage, and bag placement stage. Different branches correspond to independent control logic, meeting the motion characteristics and parameter requirements of each stage. The bag-retrieving stage generates control parameters adapted to the suction cup adsorption and bag gripping, limiting the upper limit of movement speed and acceleration to ensure smooth and efficient bag retrieval. The bag-supporting stage generates control parameters for gripper opening and clamping force, balancing the supporting force and bag protection requirements. The bag-attaching stage generates end effector feed and positioning parameters to ensure the bag opening is accurately attached to the fruit. The parameter generation results of the three stages are uniformly transmitted to the instruction generation module, which converts the parameters into instructions, forming standardized control commands. Stage switching determination is a key transition node in the process, detecting whether a stage state has changed and determining whether a smooth speed transition is required. When a switch is determined to be required, a linear interpolation transition curve generation step is initiated, calculating the transition trajectory based on the current stage speed and the initial speed of the target stage to avoid mechanical shock and bag damage caused by sudden speed changes during stage switching. When no switch is required, the process directly enters the regular drive instruction generation step, outputting the standard control commands for the corresponding stage. The transition curve instructions and regular instructions are finally aggregated at the drive interface output stage, converting the control instructions into mechanical drive signals to complete the motion control output of the entire process. The overall process achieves seamless connection and precise control of the robot's multi-stage operation through the design of branch judgment, transition processing and parameter adaptation.

[0044] After receiving multi-source sensor signals, the stage identification and parameter generation module is responsible for determining the current operation stage of the robotic arm based on sensor data and extracting the corresponding control parameters. This module internally maintains an operation stage state transition table, which defines four basic states: bag retrieval stage, bag support stage, bag placement stage, and reset stage, as well as the sensor signal conditions required for switching between each state. Initially, the robotic arm is in the reset stage, with the end effector in a standby position above the storage area. The stage identification and parameter generation module continuously reads the air pressure signal values ​​from the multi-source sensor signals. When a decrease in air pressure is detected and its value remains below the first negative pressure threshold for multiple consecutive sampling cycles, it determines that the suction cup has successfully adsorbed the fruit bag, switches the current operation stage from the reset stage to the bag retrieval stage, and generates a bag retrieval stage status signal. At this time, the module reads a subset of motion control parameters pre-configured for the bag retrieval stage from its internal parameter storage area. This subset includes at least the upper limit of the robotic arm's end effector speed, the upper limit of acceleration, and the lower limit of the continuous negative pressure required for the suction cup to maintain its adsorption state during the bag retrieval stage. After the bag-removal stage is completed, the stage identification and parameter generation module monitors the displacement signal. When the displacement signal value gradually increases with the opening action of the grippers and reaches the first displacement threshold, it is determined that the bag-opening action has begun. The operation stage is switched to the bag-opening stage, a bag-opening stage status signal is generated, and the corresponding subset of motion control parameters for the bag-opening stage is read. This subset includes the upper limit value of the movement speed and the upper limit value of the clamping force for the bag-opening stage. As the bag-opening action continues, when the displacement signal value further increases and reaches the second displacement threshold, it is determined that the bag opening has been opened to the target size. The operation stage is switched to the bag-fitting stage, a bag-fitting stage status signal is generated, and the corresponding subset of motion control parameters for the bag-fitting stage is read. This subset includes the upper limit value of the movement speed and the upper limit value of the end effector feed speed for the bag-fitting stage. The motion control parameter subsets corresponding to each stage are independent of each other, and their values ​​are different. They are preset according to the different requirements of speed, force, and motion accuracy for the corresponding action stage. The stage identification and parameter generation module combines the current stage status signal and the corresponding subset of motion control parameters read from the parameter storage area into a motion control parameter signal, and transmits it to the instruction generation module.

[0045] The instruction generation module receives motion control parameter signals and stage status signals from the stage identification and parameter generation module. Based on the current operation stage, it calls the corresponding motion planning algorithm to generate specific robotic arm drive instruction signals. When the received stage status signal indicates that the current stage is bag retrieval, the instruction generation module plans a spatial motion trajectory from the current position to the bag retrieval point in the storage area based on the upper limits of the bag retrieval stage's motion speed and acceleration, generating a bag retrieval drive instruction signal that includes the suction cup adsorption start time and spatial coordinate sequence. When the stage state switches to the bag opening stage, the instruction generation module generates a bag opening drive instruction signal that controls the two grippers to open outwards at a specified speed, based on the upper limits of the bag opening stage's motion speed and clamping force. It also sets an upper limit constraint on the clamping force in the motion planning, automatically reducing the opening speed when the clamping force fed back by the tactile pressure signal approaches this upper limit. When the stage state switches to the bagging stage, the instruction generation module generates a bagging drive instruction signal based on the upper limit of the bagging stage movement speed and the upper limit of the end effector feed speed. This signal instructs the end effector to carry the opened fruit bag along the fruit growth direction. Simultaneously, it receives real-time feedback from tactile pressure signals. When contact between the bottom of the fruit bag and the fruit surface is detected, a stop-feed instruction segment is inserted into the drive instruction. Furthermore, the instruction generation module incorporates a smooth transition mechanism for motion parameters between stages. Whenever the stage state signal from the stage identification and parameter generation module changes, the instruction generation module does not immediately jump the motion parameters from the current stage to the next stage. Instead, it uses the end effector movement speed at the last moment of the current stage as the transition start speed and the preset initial movement speed of the next stage as the transition end speed. Within a preset transition time, it linearly interpolates the two speed values ​​to generate a continuously changing speed transition curve. Based on this transition curve, it generates a transition drive instruction signal, which replaces the drive instruction signal generated according to the original stage parameters during the transition period. The transition time is adaptively adjusted based on the speed difference between the two stages to balance operational efficiency and motion smoothness. The instruction generation module ultimately encapsulates all types of drive instruction signals into a unified robotic arm drive instruction signal, which is then sent to the drive interface module and the operation anomaly detection and alarm module, respectively.

[0046] The drive interface module is the interface layer connecting the digital control system and the physical actuators of the robot. After receiving the robot drive command signals from the command generation module, this module decomposes them into position command components, speed command components, and torque command components according to the control interface type of each joint or end effector of the robot. For joints using position control mode, the drive interface module converts the position command components into orthogonal pulse signals or bus-type position command frames; for joints using speed control mode, the drive interface module converts the speed command components into analog voltage signals or digital speed command values; for joints using torque control mode, the drive interface module converts the torque command components into corresponding analog voltage or current command values. The converted drive signals are electrically matched and timing controlled according to the interface protocols of each servo driver of the robot, and output to the corresponding servo drivers via cables or fieldbuses. Ultimately, this drives each joint and end effector of the robot to perform bag picking, bag opening, bag putting, and reset actions according to the planned motion trajectory and action sequence.

[0047] like Figure 4The diagram shows the flowchart for the abnormal perception and handling of the fruit bag grasping robot, including multiple abnormality branches and retry and shutdown loop logic to build a comprehensive operational safety protection system. Multi-source sensor signal acquisition is the foundation of abnormality detection, continuously acquiring real-time data such as images, air pressure, and tactile pressure to provide raw evidence for abnormality identification and ensure that no abnormal signals are missed. Abnormality type detection is a core branch node in the process. Based on signal type and characteristics, abnormalities are classified into three categories: bag-grabbing grayscale abnormalities, suction cup air pressure abnormalities, and bag-supporting pressure abnormalities. These correspond to the fault types of the three key operational stages of the system, achieving accurate classification and identification of abnormalities. The bag-picking grayscale anomaly branch calculates grayscale change characteristics based on image signals to determine whether the bag-picking action was successful, completing the bag-picking failure judgment and accurately identifying bag-picking faults such as empty gripping and missed gripping. The suction cup air pressure anomaly branch analyzes the air pressure change rate sequence to determine whether the suction cup adsorption state is stable, identifying adsorption faults such as air leakage and loose adsorption. The bag-supporting pressure anomaly branch calculates the pressure difference on both sides of the gripper to determine whether the bag-supporting action is symmetrical, identifying bag-supporting faults such as uneven gripper force and bag body displacement. After the bag-picking failure judgment, the retry count verification stage is entered. According to the preset number of retry rules, corresponding processing is performed. If the retry limit is not reached, the bag-picking retry cycle is started, controlling the robot arm to repeatedly perform the bag-picking action to improve the success rate of the operation. If the retry limit is reached, a shutdown alarm is triggered to stop the system operation and avoid invalid operation. After the adsorption anomaly judgment, the adsorption retry stage is directly started to adjust the suction cup air pressure and adsorption position to repair the adsorption anomaly. After the bag-supporting asymmetry judgment, the gripper parameter adjustment stage is started to correct the drive parameters of the grippers on both sides in real time, balance the gripping pressure, and eliminate the asymmetry. After the three processing steps of bag retrieval retry, adsorption retry, and gripper parameter adjustment are completed, the operation status is transmitted back to the multi-source sensor signal acquisition stage, restarting the anomaly detection process and forming a closed-loop cycle for anomaly handling. Upon triggering the shutdown alarm, the system enters a work pause state, terminating mechanical actions and outputting a fault message to ensure equipment and operational safety. The overall process, through its multi-anomaly branching, hierarchical processing, and cyclical detection design, achieves rapid anomaly identification, intelligent processing, and closed-loop protection, comprehensively improving the stability and reliability of the system operation.

[0048] The operation anomaly perception and alarm module continuously receives multi-source sensor signals from the multi-source sensor signal capture module and robotic arm drive command signals from the command generation module throughout the operation. Based on multiple preset anomaly judgment conditions, it performs real-time detection of typical anomalies that may occur during bag retrieval, bag opening, and bagging. During the bag retrieval stage, the module extracts image signals from the multi-source sensor signals, capturing one frame before and one frame after the bag retrieval action. A rectangular preset calibration area covering a local area of ​​the gripper's front end is defined at the same location in these two frames. The arithmetic mean of the grayscale values ​​of all pixels within the two preset calibration areas is calculated, and the absolute value of the difference between the two means is used as the grayscale change feature value. The formula for calculating the grayscale change feature value is as follows:

[0049] ;

[0050] in, Here, N represents the grayscale variation characteristic value, and N is the total number of pixels in the preset calibration area of ​​the first frame. M represents the grayscale value of the i-th pixel in the first frame, and M represents the total number of pixels in the preset calibration area of ​​the second frame. The grayscale value of the j-th pixel in the second frame. This is the grayscale correction factor. This is the grayscale offset compensation amount. The formula integrates six calculation symbols: summation, division, subtraction, absolute value, multiplication, and addition, with far more than 30 characters, making it a complex algorithm. Its core design addresses the characteristics of fruit bags—thin, uniform in color, and susceptible to ambient light interference—by standardizing and quantizing the grayscale information of the images before and after the bag-removal action. First, it calculates the average grayscale value of the preset calibration area in two frames using summation and division operations, eliminating calculation errors caused by single-pixel noise. Then, it extracts the core features of the grayscale difference through subtraction and absolute value operations, avoiding feature distortion caused by the cancellation of positive and negative differences. Subsequently, it introduces a grayscale correction coefficient and a grayscale offset compensation amount to eliminate the influence of external factors such as changes in ambient light intensity, analog-to-digital conversion errors in the image acquisition device, and lens dust obstruction on the detection results. This ensures that the grayscale change feature value accurately reflects whether the gripper successfully grasped the fruit bag. The calculation result of this formula directly serves as the criterion for determining bag-removal failure. When the grayscale change value is below the preset threshold, the system determines that bag retrieval has failed. After the number of consecutive failures reaches the target, a shutdown alarm is triggered. The formula is deeply linked with the image acquisition device and the filtering analog-to-digital conversion unit of the multi-source sensor signal capture module. Combined with the signal timestamp, the system achieves time synchronization, ensuring the real-time performance and accuracy of anomaly detection during the bag retrieval stage. This provides precise data support for the control logic such as bag retrieval retry and shutdown protection of the robotic arm. It is the key algorithm for the system to realize automated and intelligent anomaly handling in the bag retrieval process.

[0051] If the grayscale change characteristic value is lower than the preset grayscale change threshold, it indicates that no significant grayscale change has occurred in the gripper area due to the picking of the bag. Based on this, the operation anomaly perception and alarm module determines that the bag picking has failed and generates a bag picking failure alarm signal. The alarm signal is then transmitted to the command generation module. After receiving the alarm signal, the command generation module generates a bag picking retry drive command signal to control the robot to re-execute the bag picking action sequence. If the number of times the bag picking failure alarm signal is triggered consecutively reaches the preset retry limit, the operation anomaly perception and alarm module generates a bag picking failure shutdown alarm signal and suspends all operations of the robot. After the adsorption action is established, the operation anomaly detection and alarm module extracts continuous sampling values ​​of air pressure signals from multi-source sensor signals. It calculates the air pressure change rate by dividing the air pressure difference between two adjacent sampling times by the sampling period, constructing an air pressure change rate sequence. When the air pressure value has reached the target negative pressure, but the air pressure change rate sequence shows fluctuations exceeding a preset change rate threshold in several consecutive sampling periods, the adsorption state is determined to be abnormal. An adsorption anomaly alarm signal is generated and sent to the command generation module to trigger an adsorption retry. During the bag-holding stage, the operation anomaly detection and alarm module extracts tactile pressure signals from multi-source sensor signals. It sums the pressure values ​​of all tactile sensor units on the left and right grippers, calculating the absolute value of the difference between the total pressure on the left and right sides as the pressure difference. When this pressure difference exceeds a preset difference threshold, an asymmetry in the current bag-holding action is determined, generating a bag-holding asymmetry alarm signal and sending it to the command generation module. The command generation module then independently corrects the drive parameters of the grippers on both sides to gradually eliminate the left-right pressure difference. The operation anomaly detection and alarm module records the type, occurrence time, and corresponding sensor data of all generated alarm signals, forming an alarm information stream that can be read by external monitoring terminals. This enables online monitoring and closed-loop handling of anomalies throughout the entire fruit bag grasping operation. The calculation formulas for each air pressure change rate are as follows:

[0052] ;

[0053] in, Let $k$ be the rate of change of air pressure at the $k$-th sampling time. Let k be the air pressure value at time k. The air pressure value at time k-1. The preset sampling period is α, where α is the gain coefficient for the rate of change of air pressure. Let β be the target negative pressure value and β be the attenuation coefficient of the air pressure change rate. Its initial research and development aimed to solve problems such as air pressure fluctuations in the suction cup air path caused by the breathable material of the fruit bag, the inability of traditional single-difference calculations to accurately identify leaks, and loose adsorption. The formula first obtains the basic air pressure change rate by dividing the air pressure difference between adjacent moments by the sampling period. Then, it combines a gain coefficient to amplify the real abnormal signal. Simultaneously, a target negative pressure deviation correction term is introduced. The deviation between the current air pressure and the target negative pressure is calculated through summation, division, absolute value, and square root operations. The attenuation coefficient is used to balance the dynamic change amplitude, allowing the air pressure change rate to accurately reflect the stability of the adsorption state. The pressure change rate sequence generated by the formula is the core data for determining adsorption anomalies. When the pressure reaches the target negative pressure, if the absolute value of the change rate in the sequence continues to exceed the threshold, the system immediately generates an adsorption anomaly alarm and triggers an adsorption retry command. The formula is adapted to the preset sampling period and gas path structure parameters of the pressure sensor, and works in conjunction with the timestamps and filtering processes of multi-source sensor signals to effectively avoid misjudgment problems caused by transient fluctuations in the gas path and sensor sampling errors. Together with the grayscale change feature value algorithm, it forms a two-dimensional anomaly detection system in the bag removal stage, comprehensively improving the operational reliability of the bag removal process. The formula for calculating the pressure difference is as follows:

[0054] ;

[0055] in, The pressure difference is represented by A, which is the total number of tactile sensing units in the left gripper. B represents the pressure value of the a-th sensing unit on the left, and B represents the total number of tactile sensing units in the gripper on the right. This represents the pressure value of the b-th sensing unit on the right. This is the pressure normalization coefficient. This algorithm is an important supplement to the previous two complex algorithms. Together, they constitute the core algorithm system for multi-source sensor anomaly detection, corresponding to three types of sensor signals: image, air pressure, and touch. They cover the two key operational stages of bag removal and bag opening. The formula first calculates the total pressure of the grippers on both sides through summation, intuitively reflecting the contact force between the grippers and the fruit bag. Then, it extracts the degree of pressure asymmetry between the left and right sides through subtraction and absolute value calculation. Finally, it introduces a pressure normalization coefficient to eliminate calculation errors caused by differences in the range of the touch sensor array, installation position deviations, and different contact areas, allowing the pressure difference to be compared with a unified standard and a preset threshold. When the difference exceeds the preset threshold, the system determines that the bag-holding is asymmetrical. The instruction generation module adjusts the driving parameters of the grippers on both sides in real time to quickly correct the force deviation. This formula is adapted to the signal acquisition, filtering and conversion process of the tactile sensor array and linked with the bag-holding stage judgment logic of the stage identification and parameter generation module. At the same time, it forms a time sequence connection with the grayscale and air pressure algorithms of the bag-removal stage, ensuring that the robot arm can control abnormalities throughout the entire process from bag removal to bag holding. The three algorithm formulas are interconnected and complementary, and the abnormal characteristics are standardized and quantified based on multi-source sensor signals, providing accurate algorithmic support for the digital program control of the system and greatly improving the stability and adaptability of the fruit bag gripping robot arm in agricultural automation operations.

[0056] In one embodiment of the invention, a self-propelled operating platform is equipped with a six-axis articulated robotic arm for fruit bagging. The robotic arm's end effector is equipped with a vacuum suction cup and parallel grippers, and a paper bag measuring 150 mm in length and 120 mm in width is placed inside the bag storage device. After the system is powered on, the multi-source sensor signal acquisition module begins operation: a monochrome camera with a resolution of 128 x 96 pixels captures images of the gripper area at 30 frames per second; a pressure sensor collects the air pressure in the suction cup at a 10-millisecond cycle; an encoder with a resolution of 400 pulses per revolution corresponds to a 0.01 mm displacement of the grippers; and a tactile sensor array includes six pressure-sensitive units on the inner sides of each of the left and right grippers, with a range of 0 to 5 Newtons, and each channel outputs pressure values ​​at a 50-millisecond cycle. These signals are filtered and timestamped before being encapsulated into multi-source sensor signal frames, updated every 30 milliseconds, for use by the stage identification and parameter generation module and the operation anomaly detection and alarm module.

[0057] Initially, the robotic arm is in the reset phase, with its end effector positioned 50 mm above the storage bag area. After the main controller activates the vacuum generator, the air pressure signal drops from 0 kPa to -45 kPa within 60 milliseconds, continuing towards the target value of -70 kPa. The phase identification and parameter generation module detects that the air pressure has been below the first negative pressure threshold of -60 kPa for twelve consecutive sampling cycles, determining that the suction cup has adsorbed the fruit bag, and switches to the bag retrieval phase, generating a bag retrieval phase status signal and reading a subset of bag retrieval phase parameters: upper limit of movement speed 400 mm / s, upper limit of acceleration 1600 mm / s cubic meters, and lower limit of continuous negative pressure -55 kPa. Based on this, the command generation module plans the lifting trajectory and generates a bag retrieval drive command signal. The drive interface module converts the position and speed command components into servo drive signals, and the robotic arm lifts the fruit bag 80 mm above the storage bag area.

[0058] Before and after the bag retrieval action, the operation anomaly detection and alarm module extracts the 50th and 65th frames of the image signal and calculates the average grayscale value within a 40x20 pixel rectangular pre-defined area at the front of the gripper. In one normal cycle, the average value before the action is 158, and the average value after the action is 82, with a grayscale change feature value of 76, exceeding the grayscale change threshold of 35, indicating successful bag retrieval. In another cycle, due to a slight displacement of the caustic bag causing the suction cup to slip, the average grayscale values ​​before and after the action are 161 and 163 respectively, with a feature value of only 2. The module immediately generates a bag retrieval failure alarm signal. The instruction generation module responds to this signal, terminates the current action, and replans a bag retrieval retry trajectory with a fine-tuned lateral offset of 5 millimeters. After the second retry, the feature value rises to 69, and the bag retrieval is successfully completed.

[0059] After successful bag retrieval, the grippers gradually open outward from their initial closed position, increasing the displacement encoder reading. When the displacement reaches the first displacement threshold of 65 mm, the stage identification and parameter generation module switches to the bag-opening stage, reading a subset of parameters for this stage: upper limit of movement speed of 30 mm / s and upper limit of clamping force of 2.5 Newtons. The command generation module generates a bag-opening drive command signal, controlling the grippers to continue opening at a speed of 25 mm / s and monitoring tactile pressure in real time. During the bag-opening process, the operation anomaly perception and alarm module calculates that the total pressure of the six pressure-sensitive units on the left is 1.63 Newtons, while the total pressure on the right is only 1.09 Newtons. The pressure difference of 0.54 Newtons exceeds the difference threshold of 0.4 Newtons, generating an asymmetric bag-opening alarm signal. Based on this, the command generation module independently adjusts the speed on both sides, reducing the speed on the left to 20 mm / s and increasing it on the right to 28 mm / s. After approximately 150 milliseconds, the total pressure on the left decreases to 1.28 Newtons, while the right increases to 1.31 Newtons, reducing the difference to 0.03 Newtons. After returning to normal, both sides resume opening at a uniform speed.

[0060] When the displacement reaches the second displacement threshold of 120 mm, the bag opening is fully opened, and the system switches to the bagging stage. The upper limit of the parameter subset movement speed is 80 mm / s, and the upper limit of the feed speed is 120 mm / s. The instruction generation module plans the trajectory of the opened fruit bag along the fruit growth axis at a speed of 100 mm / s towards the fruit, generating a bagging drive instruction signal. During the feeding process, the tactile pressure signal is continuously scanned. When the pressure value of four or more pressure-sensitive units in the middle and lower part of the gripper exceeds the 0.3 Newton contact threshold, it is considered that the bottom of the fruit bag has contacted the fruit surface. The instruction generation module inserts a bagging stop instruction, the robot arm stops feeding, and then slightly closes the gripper to release the bag opening and reset. During stage switching, the instruction generation module uses a speed smooth transition mechanism. For example, when switching from bag opening to bagging, the end speed of bag opening is 24 mm / s and the initial speed of bagging is 70 mm / s. The speed curve is generated by linear interpolation within a 60-millisecond transition time to eliminate motion abrupt changes. After the entire operation cycle ends, the robot arm returns to the standby position, waiting for the next fruit to be bagged. All alarm signals and sensor data are uploaded to the main control computer via industrial Ethernet for online monitoring and post-event analysis.

[0061] like Figure 5 As shown, this invention provides a digital program control method for a fruit bag grasping robot, comprising: S1: acquiring image signals, air pressure signals, displacement signals, and tactile pressure signals of the gripper area, and integrating them to form multi-source sensor signals; S2: receiving the multi-source sensor signals, determining the current operation stage of the robot based on the time domain and amplitude characteristics of each signal, wherein the operation stage includes at least a bag-picking stage, a bag-supporting stage, and a bag-attaching stage, and generating differentiated motion control parameter signals and stage status signals for different operation stages; S3: receiving the motion control parameter signals and stage status signals, and generating robot drive command signals according to the motion control parameters corresponding to the current operation stage; S4: receiving the robot drive command signals, converting them into drive signals for each joint or end effector of the robot, and outputting them; S5: receiving the multi-source sensor signals and the robot drive command signals, and generating an operation abnormality alarm signal based on preset abnormality judgment conditions.

[0062] This invention discloses a digital program control system and method for a fruit bag grasping robot. A multi-source sensor signal capture module acquires image, air pressure, displacement, and tactile pressure signals. A stage identification and parameter generation module automatically determines the current stage (bag retrieval, bag opening, or bagging) based on the time-domain and amplitude characteristics of these signals, and generates differentiated motion control parameters for each stage. An instruction generation module generates drive instructions based on the stage state, employing a smooth transition mechanism during stage switching. A drive interface module converts the instructions into drive signals for joints or end effectors. An operation anomaly perception and alarm module analyzes real-time changes in image grayscale, air pressure change rate, and tactile pressure distribution. Upon detecting bag retrieval failure, adsorption anomaly, or asymmetrical bag opening, an alarm signal is generated, triggering a retry or parameter adjustment. This ensures the continuity of fruit bagging operations without relying on fixed coordinates.

[0063] Therefore, the present invention provides a digital program control system and method for a fruit bag grasping robot, which solves the problems of high bag grasping failure rate, asymmetrical bag opening, and bag offset caused by the lack of real-time perception and feedback of the actual execution status of existing fruit bag grasping robots in unstructured orchard environments due to their reliance on fixed position coordinates and preset action sequences.

[0064] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A digital program control system for a fruit bag gripping robot, characterized in that, include: The multi-source sensor signal capture module is used to acquire image signals, air pressure signals, displacement signals and tactile pressure signals of the gripper area, and integrate them to form multi-source sensor signals. The stage identification and parameter generation module receives the multi-source sensor signals and determines the current operation stage of the robot arm based on the time domain and amplitude characteristics of each signal. The operation stage includes at least the bag picking stage, the bag opening stage, and the bag putting stage. Differentiated motion control parameter signals and stage status signals are generated for different operation stages. The instruction generation module receives the motion control parameter signal and the stage status signal, and generates a robot arm drive instruction signal according to the motion control parameters corresponding to the current operation stage. The drive interface module receives the drive command signal of the robot arm, converts it into drive signals for each joint or end effector of the robot arm, and outputs them. The operation anomaly perception and alarm module receives multi-source sensor signals output by the multi-source sensor signal capture module and robot arm drive command signals output by the command generation module, and generates operation anomaly alarm signals based on preset anomaly judgment conditions.

2. The digital program control system for a fruit bag gripping robot according to claim 1, characterized in that, The multi-source sensor signal capture module includes an image acquisition device, a pressure sensor, a displacement sensor, and a tactile sensor array. The image acquisition device is fixed above the robot arm body or the working platform and is used to acquire continuous frame images of the robot arm gripper area before and after the bag-picking action and output image signals. The pressure sensor is installed in the air path connected to the bag-picking suction cup and is used to measure the air pressure value inside the air path at a preset sampling period and output air pressure signals. The displacement sensor is installed on the drive mechanism of the robot arm gripper and is used to detect the opening distance of the gripper during the bag-opening stage and output displacement signals. The tactile sensor array is attached to the inner surface of the robot arm gripper and is used to sense the pressure distribution generated when the gripper contacts the fruit bag and output tactile pressure signals. The multi-source sensor signal capture module filters and performs analog-to-digital conversion on the image signal, pressure signal, displacement signal, and tactile pressure signal respectively, and then encapsulates them into a multi-source sensor signal containing timestamp information and transmits it to the stage identification and parameter generation module.

3. The digital program control system for a fruit bag gripping robot according to claim 2, characterized in that, After receiving the multi-source sensor signals, the stage identification and parameter generation module first extracts the time-domain features of the air pressure signal. When the value of the air pressure signal is consistently below a first negative pressure threshold for a first duration, it determines that the robotic arm has entered the bag-retrieving stage and generates a bag-retrieving stage status signal. Simultaneously, it reads a subset of motion control parameters corresponding to the bag-retrieving stage from a preset parameter storage area. This subset includes the upper limit of the motion speed and the upper limit of the acceleration during the bag-retrieving stage. After the bag-retrieving stage ends, the stage identification and parameter generation module extracts the time-domain features of the displacement signal. When the value of the displacement signal reaches a first displacement threshold, it determines that the robotic arm has entered the bag-holding stage and generates a... The system generates a bag-holding stage status signal and simultaneously reads a subset of motion control parameters corresponding to the bag-holding stage from a preset parameter storage area. This subset includes the upper limit of the bag-holding stage's motion speed and the upper limit of the bag-holding stage's clamping force. After the bag-holding stage ends, the stage identification and parameter generation module extracts the time-domain features of the displacement signal. When the value of the displacement signal reaches a second displacement threshold, the system determines that the robot has entered the bag-holding stage and generates a bag-holding stage status signal. Simultaneously, it reads a subset of motion control parameters corresponding to the bag-holding stage from the preset parameter storage area. This subset includes the upper limit of the bag-holding stage's motion speed and the upper limit of the bag-holding stage's end effector feed speed.

4. The digital program control system for a fruit bag gripping robot according to claim 3, characterized in that, During the bag-removal stage, the instruction generation module receives the bag-removal stage status signal and a subset of motion control parameters corresponding to the bag-removal stage, and generates a bag-removal drive instruction signal. The bag-removal drive instruction signal includes a suction cup adsorption start instruction and a first motion trajectory instruction for the robotic arm end effector to move towards the storage bag area. During the bag-opening stage, the instruction generation module receives the bag-opening stage status signal and a subset of motion control parameters corresponding to the bag-opening stage, and generates a bag-opening drive instruction signal. The bag-opening drive instruction signal includes a second motion trajectory instruction to control the two grippers to open outward at a speed not exceeding the upper limit of the bag-opening stage motion speed, and a bag-opening deceleration instruction triggered when the tactile pressure signal reaches the upper limit of the clamping force. During the bagging stage, the instruction generation module receives the bagging stage status signal and a subset of motion control parameters corresponding to the bagging stage, and generates a bagging drive instruction signal. The bagging drive instruction signal includes a third motion trajectory instruction to control the robotic arm end effector to move the opened bag opening towards the fruit, and a bagging stop instruction triggered when the tactile pressure signal detects that the fruit bag is in contact with the fruit surface.

5. The digital program control system for a fruit bag gripping robot according to claim 4, characterized in that, When the instruction generation module receives a stage state signal from the stage identification and parameter generation module indicating a switch, it acquires the end motion speed value of the current operation stage as the starting speed value and the preset initial motion speed value of the target operation stage as the ending speed value. Within a preset transition time, it performs linear interpolation on the starting speed value and the ending speed value to generate a transition speed curve. The instruction generation module generates a transition drive instruction signal based on the transition speed curve. The transition drive instruction signal replaces the robot arm drive instruction signal and is transmitted to the drive interface module within the transition time. The value of the transition time is determined based on the difference in motion speed between the current operation stage and the target operation stage. After the transition time ends, the instruction generation module generates the robot arm drive instruction signal according to the subset of motion control parameters corresponding to the target operation stage.

6. The digital program control system for a fruit bag gripping robot according to claim 3, characterized in that, During the bag-retrieving stage, the operation anomaly perception and alarm module receives the image signal output by the multi-source sensor signal capture module, calculates the grayscale change feature value of the gripper area in the preset frame image after the bag-retrieving action is executed, and if the grayscale change feature value is lower than the preset grayscale change threshold, the bag-retrieving action is determined to have failed and a bag-retrieving failure alarm signal is generated. The bag-retrieving failure alarm signal is transmitted to the instruction generation module, and the instruction generation module generates a bag-retrieving retry drive instruction signal in response to the bag-retrieving failure alarm signal, so that the robot arm re-executes the bag-retrieving action; if the number of consecutive bag-retrieving failure alarm signals reaches the preset retry limit, the operation anomaly perception and alarm module generates a bag-retrieving failure shutdown alarm signal and suspends the robot arm operation.

7. The digital program control system for a fruit bag gripping robot according to claim 6, characterized in that, The operation anomaly perception and alarm module calculates the grayscale change feature value as follows: In the first frame image before the bag-removal action is performed, a rectangular preset calibration area of ​​a preset size is defined at the front end of the gripper. The average grayscale value of all pixels within this rectangular preset calibration area is calculated as the first grayscale mean. In the second frame image after the bag-removal action is performed, a rectangular preset calibration area of ​​the same position and size is defined. The average grayscale value of all pixels within this rectangular preset calibration area is calculated as the second grayscale mean. The absolute value of the difference between the first grayscale mean and the second grayscale mean is used as the grayscale change feature value. The calculation formula for the grayscale change feature value is as follows: ； in, Here, N represents the grayscale variation characteristic value, and N is the total number of pixels in the preset calibration area of ​​the first frame. M represents the grayscale value of the i-th pixel in the first frame, and M represents the total number of pixels in the preset calibration area of ​​the second frame. The grayscale value of the j-th pixel in the second frame. This is the grayscale correction factor. This is the grayscale offset compensation amount.

8. The digital program control system for a fruit bag gripping robot according to claim 3, characterized in that, During the bag-removal stage, the operation anomaly detection and alarm module receives the air pressure signal output by the multi-source sensor signal capture module, continuously acquires multiple air pressure values ​​at a preset sampling period, and calculates a sequence of air pressure change rates. Each air pressure change rate in the sequence is calculated by dividing the difference between the air pressure values ​​at two adjacent sampling times by the preset sampling period. If the air pressure signal value has reached the target negative pressure value, and the absolute value of each air pressure change rate in the sequence exceeds a preset air pressure change rate threshold in subsequent consecutive sampling periods, the adsorption state is determined to be abnormal, and an adsorption anomaly alarm signal is generated. The adsorption anomaly alarm signal is transmitted to the command generation module, which responds to the adsorption anomaly alarm signal by generating an adsorption retry drive command signal. The calculation formula for each air pressure change rate is as follows: ； in, Let $k$ be the rate of change of air pressure at the $k$-th sampling time. Let k be the air pressure value at time k. The air pressure value at time k-1. The preset sampling period is α, where α is the gain coefficient for the rate of change of air pressure. β represents the target negative pressure value, and β is the attenuation coefficient of the rate of change of air pressure.

9. The digital program control system for a fruit bag gripping robot according to claim 3, characterized in that, During the bag-holding stage, the operation anomaly detection and alarm module receives tactile pressure signals output by the multi-source sensor signal capture module. These tactile pressure signals include the pressure values ​​of each tactile sensor unit on both sides of the robotic arm's grippers. The operation anomaly detection and alarm module calculates the sum of the pressure values ​​of all tactile sensor units on the left gripper as the left-side pressure sum, and calculates the sum of the pressure values ​​of all tactile sensor units on the right gripper as the right-side pressure sum. The absolute value of the difference between the left-side pressure sum and the right-side pressure sum is calculated as the pressure difference. If the pressure difference exceeds a preset difference threshold, bag-holding asymmetry is determined, and a bag-holding asymmetry alarm signal is generated. This alarm signal is transmitted to the instruction generation module, which adjusts the drive parameters of both grippers in response to the alarm signal to reduce the pressure difference. The formula for calculating the pressure difference is as follows: ； in, The pressure difference is represented by A, which is the total number of tactile sensing units in the left gripper. B represents the pressure value of the a-th sensing unit on the left, and B represents the total number of tactile sensing units in the gripper on the right. This represents the pressure value of the b-th sensing unit on the right. This is the pressure normalization coefficient.

10. A method for a digital program control system for a fruit bag grasping robot according to any one of claims 1-9, characterized in that, include: S1: Acquire image signals, air pressure signals, displacement signals and tactile pressure signals of the gripper area, and integrate them to form multi-source sensing signals; S2: Receive the multi-source sensor signals, determine the current operation stage of the robot arm based on the time domain and amplitude characteristics of each signal, the operation stage includes at least the bag picking stage, the bag opening stage and the bag putting stage, and generate differentiated motion control parameter signals and stage status signals for different operation stages. S3: Receive the motion control parameter signal and the stage status signal, and generate a robot drive command signal according to the motion control parameters corresponding to the current operation stage; S4: Receive the robot arm drive command signal, convert it into drive signals for each joint of the robot arm or the end effector, and output them; S5: Receive the multi-source sensor signals and the robot arm drive command signals, and generate an operation abnormality alarm signal based on preset abnormality judgment conditions.

Images