Robot polishing contact state phased identification and adaptive control method and system

Abstract

This application relates to the field of robot automatic control technology, and provides a method and system for phased identification and adaptive control of robot grinding contact state. The method acquires real-time force signals from the robot's end effector, processes these signals to obtain compensation force signals, and extracts the amplitude characteristics of these compensation force signals. Based on these amplitude characteristics, the contact state between the robot and the workpiece is determined. When the amplitude characteristic is less than a first preset threshold, it is determined to be in a non-contact stage; when the amplitude characteristic is greater than or equal to the first preset threshold and less than a second preset threshold, it is determined to be in a contact transition stage; and when the amplitude characteristic is greater than or equal to the second preset threshold, it is determined to be in a stable contact stage. In the non-contact stage, a position control strategy is executed to move the robot's end effector towards the workpiece. In the contact transition stage, a graded force control strategy is executed. In the stable contact stage, a constant force control strategy is executed to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation.

Description

Technical Field

[0001] This application relates to the field of robot automatic control technology, and in particular to a method and system for phased recognition and adaptive control of robot grinding contact state. Background Technology

[0002] Currently, in robotic grinding operations, contact state recognition mostly uses a single force threshold judgment method, which can only distinguish between contact and non-contact states, and cannot identify the contact transition process. Existing control strategies mostly use a direct switching method, which can easily generate a large impact force at the moment of contact, resulting in damage to the grinding head or scratches on the workpiece surface. Summary of the Invention

[0003] This application provides a method and system for phased identification and adaptive control of robot grinding contact state, which aims to solve the problems of existing technologies that use position-force hybrid control, which do not dynamically adjust the control weight according to different stages of the contact process, resulting in low constant force control accuracy, poor grinding consistency, and inability to meet the needs of high-precision grinding operations.

[0004] In a first aspect, embodiments of this application provide a method for phased identification and adaptive control of robot grinding contact states, the method comprising: The real-time force signal of the robot's end effector is acquired, the real-time force signal is processed to obtain the compensation force signal, and the amplitude characteristics of the compensation force signal are extracted. The contact state between the robot and the workpiece is determined based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. In the non-contact phase, a position control strategy is implemented to move the robot's end effector towards the workpiece. In the contact transition phase, a graded force control strategy is implemented. In the stable contact phase, a constant force control strategy is implemented to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set in both the contact transition and stable contact phases. When the actual contact force exceeds this threshold, an abnormal contact state is identified, and the end effector stops its feed motion. During both phases, a differential tracker tracks the compensated force signal in real time, outputting the tracked value and its differential value. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the differential value remains consistently below the stability threshold, the current contact state is considered stable. In the stable contact phase, the continuous time the contact force remains within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is considered complete.

[0005] Secondly, this application provides a robot grinding contact state phased recognition and adaptive control system, including: The signal acquisition unit is used to acquire the real-time force signal of the robot's end effector, process the real-time force signal to obtain the compensation force signal, and extract the amplitude characteristics of the compensation force signal. The state determination unit is used to determine the contact state between the robot and the workpiece based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. The grinding completion unit is used to execute a position control strategy to control the robot's end effector to move towards the workpiece during the non-contact phase, execute a graded force control strategy during the contact transition phase, and execute a constant force control strategy during the stable contact phase to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set during the contact transition and stable contact phases. When the actual contact force exceeds the force safety upper limit threshold, an abnormal contact state is determined, and the end effector is controlled to stop its feed motion. During the contact transition and stable contact phases, a differential tracker performs real-time tracking processing on the compensated force signal, outputting the tracking value and its differential value of the force signal. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the force differential value is continuously lower than the stability threshold, the current contact state is determined to have reached stability. During the stable contact phase, the continuous time for the contact force to remain within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is determined to be complete.

[0006] This application uses the dual characteristics of force signal amplitude and rate of change to divide the contact state into three stages: no contact, contact transition, and stable contact. This results in high accuracy in contact state identification and avoids misjudgment. By automatically switching the corresponding control strategy at different contact stages, the position and force control weights are gradually adjusted during the contact transition stage to achieve a smooth transition without impact. During the stable contact stage, a constant force control strategy is implemented to effectively ensure the stability of the grinding contact force, improve grinding quality and work consistency, and extend the service life of the grinding head.

[0007] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0008] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic flowchart illustrating the steps of a robot grinding contact state phased identification and adaptive control method according to an embodiment of this application; Figure 2 This is a schematic diagram illustrating the principle of a robot grinding contact state phased identification and adaptive control method provided in an embodiment of this application; Figure 3 This is a schematic block diagram of a robot grinding contact state phased recognition and adaptive control system provided in one embodiment of this application; Figure 4 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.

[0010] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0013] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0014] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0015] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0016] Currently, in robotic grinding operations, contact state recognition mostly relies on a single force threshold judgment method, which can only distinguish between contact and non-contact states, failing to identify the contact transition process. Existing control strategies often employ direct switching, which can easily generate significant impact forces at the moment of contact, leading to damage to the grinding head or scratches on the workpiece surface. Some technologies using position-force hybrid control do not dynamically adjust control weights according to different stages of the contact process, resulting in low constant force control accuracy, poor grinding consistency, and an inability to meet the demands of high-precision grinding operations.

[0017] Please refer to Figure 1 and Figure 2 This application provides a method for phased recognition and adaptive control of robot grinding contact states, applied to computer equipment. The computer equipment can be deployed on a single server or server cluster. It can also be deployed on handheld terminals, laptops, wearable devices, or robots, etc.

[0018] This embodiment provides a staged identification and adaptive control method for robot grinding contact states, applicable to surface grinding, deburring, and polishing operations on metal and non-metal workpieces by various industrial robots. Before implementing this method, the calibration and gravity compensation of the six-dimensional force sensor must be completed in advance to eliminate the influence of sensor zero-point drift and end-load gravity on the force signal measurement. The force signal after compensation processing is the compensated force signal, ensuring that the force signal can accurately reflect the contact force between the robot end effector and the workpiece.

[0019] The provided method for phased identification and adaptive control of robot grinding contact states includes steps S101 to S103. Details are as follows: Step S101. Obtain the real-time force signal of the robot end effector, process the real-time force signal to obtain the compensation force signal, and extract the amplitude characteristics of the compensation force signal.

[0020] Specifically, force signal acquisition is achieved by a six-dimensional force sensor installed between the robot end effector flange and the grinding head, which collects force signals (Fx, Fy, Fz) and torque signals (Mx, My, Mz) in three orthogonal directions of the robot end effector in the sensor coordinate system.

[0021] In grinding operations, the normal contact force perpendicular to the workpiece surface is the core component determining the grinding quality and contact state. Therefore, subsequent signal processing and state determination in this method primarily focus on the normal force signal; the tangential force signal serves only as an auxiliary verification and attitude adjustment basis and does not participate in the determination of the core contact stage. The raw force signals collected by the sensors are transmitted to the robot's main controller in real time via an industrial Ethernet bus, CAN bus, or EtherCAT bus, with a transmission delay of no more than 1 millisecond to ensure the real-time performance of the force signals.

[0022] The robot's main controller samples the acquired raw force signals at preset fixed time intervals. The sampling frequency is set according to the precision requirements of the grinding operation, the robot's control response speed, and the controller's computing power, with a selectable range from 100 Hz to 1000 Hz. A higher sampling frequency results in higher temporal resolution of the force signal and better real-time contact state recognition, but it will correspondingly increase the computational burden and data storage pressure on the main controller. Therefore, a balance must be struck based on the actual working conditions.

[0023] In this embodiment, the sampling frequency is preferably set to 500 Hz, that is, the force signal acquisition and reading of one sampling point is completed every 2 milliseconds. The sampling data is temporarily stored in the ring buffer of the controller for subsequent compensation processing and feature extraction.

[0024] The original force signal obtained from sampling undergoes multi-dimensional compensation processing to eliminate interference from sensor zero-point drift, end-effector load gravity, robot motion inertia, and high-frequency noise, resulting in a compensated force signal that accurately reflects the contact force. The compensation processing specifically includes the following steps: Based on the pre-completed static calibration results of the six-dimensional force sensor, the raw voltage signal output by the sensor is converted into actual force and torque values ​​through the force-voltage mapping matrix obtained from the calibration, eliminating the sensor's nonlinearity error and inter-dimensional coupling error. During calibration, at least 12 sets of standard forces and standard torques of different directions and magnitudes are applied within the sensor's range using a standard force loading device, and the corresponding output voltages are collected. A 6×6 inter-dimensional decoupling matrix is ​​obtained by fitting using the least squares method, completing the calibration compensation.

[0025] Gravity compensation is based on the pre-calibrated end-effector load gravity vector and center of gravity position. Combined with the robot's current joint angles, the spatial attitude of the sensor coordinate system is obtained through forward kinematics calculation. The projection components of the end-effector load gravity on each sensor axis under the current attitude are calculated. This gravity component is subtracted from the original force signal to obtain a force signal that eliminates the influence of gravity. During gravity compensation parameter calibration, the robot is controlled in at least five significantly different poses. Sensor outputs are collected in a stationary state, and the total magnitude of the load gravity and the three-dimensional coordinates of the center of gravity are obtained by solving a system of linear equations to establish a gravity compensation model.

[0026] Inertial compensation is based on the robot's rigid body dynamics model. Based on the angle, angular velocity, and angular acceleration of each joint at the current moment, it calculates the components of the inertial force and inertial torque generated by the end effector's motion in the sensor coordinate system, cancels out the inertial component from the force signal, and eliminates the interference of the robot's acceleration and deceleration motion on the force signal measurement.

[0027] Noise reduction through filtering involves low-pass filtering of the compensated force signal to remove high-frequency noise introduced during the grinding process, such as grinding head vibration and motor electromagnetic interference. This embodiment employs a fourth-order Butterworth low-pass filter with a cutoff frequency of 50 Hz. The filtering process uses a forward-backward filtering method to avoid phase shift and ensure the accuracy of the force signal's amplitude and phase.

[0028] Amplitude features are extracted from the compensated and filtered normal force signal. The amplitude feature is the absolute value of the compensated force signal corresponding to the current sampling point, directly reflecting the magnitude of the normal contact force between the robot's end effector and the workpiece. For determining the direction of the normal force, the direction of the force applied to the workpiece surface by the grinding head is taken as the positive direction, and the amplitude feature is taken as the absolute value of the force signal to eliminate the influence of the direction sign on the state determination.

[0029] The extracted amplitude features are synchronously output to the contact state determination module and the control strategy execution module for subsequent state identification and control quantity calculation.

[0030] Step S102. Determine the contact state between the robot and the workpiece based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage.

[0031] Specifically, a first preset threshold and a second preset threshold are set in advance. The values ​​of the two preset thresholds need to be determined comprehensively based on the specific process requirements of the grinding operation, the characteristics of the workpiece material, the material and hardness of the grinding head, the measurement accuracy of the sensor, and the zero drift range.

[0032] The first preset threshold is the amplitude boundary between the non-contact stage and the contact transition stage. It is set as the upper limit of the zero-drift range of the compensated sensor and is used to determine the critical state at which the end effector and the workpiece make initial contact. This threshold is obtained through a sensor no-load static test: the compensated force signal is continuously collected for more than 10 seconds in a non-contact state, the maximum fluctuation amplitude of the signal is counted, and 1.2 times the maximum fluctuation amplitude is set as the first preset threshold to ensure that the non-contact state is not misjudged as contact due to zero-drift fluctuation.

[0033] The second preset threshold is the amplitude boundary between the contact transition stage and the stable contact stage, set to 80% to 90% of the preset target contact force, used to determine the critical state when the contact force reaches the stable grinding requirement. In this embodiment, the second preset threshold is preferably 85% of the target contact force, which ensures that the contact force is close to the target value when entering the stable contact stage, and also reserves sufficient transition space to avoid frequent state jumps.

[0034] The real-time extracted amplitude features are compared point by point with the two preset thresholds mentioned above, and the current contact state is determined according to the preset judgment logic. The specific judgment rule is as follows: If the amplitude characteristic of the current sampling point is less than the first preset threshold, it is determined that the current stage is non-contact. At this time, the grinding head and the workpiece surface do not make effective contact, and the contact force is only the sensor noise and zero drift. If the amplitude characteristic of the current sampling point is greater than or equal to the first preset threshold and less than the second preset threshold, it is determined that the current stage is the contact transition stage. At this time, the grinding head and the workpiece surface have made initial contact, but the contact force has not yet reached the target level of stable grinding and is in the transition process of gradually establishing the contact force. If the amplitude characteristic of the current sampling point is greater than or equal to the second preset threshold, it is determined that the current contact stage is stable. At this time, the contact force has approached or reached the preset target value, and the constant force control state of stable grinding can be entered.

[0035] To avoid misjudgment and frequent switching of the state caused by instantaneous fluctuations in the force signal, a state anti-jitter mechanism is added to this step: the state switch is officially confirmed only when the amplitude characteristics of N consecutive sampling points meet the judgment conditions of the corresponding stage; where N is the preset number of anti-jitter sampling points, and in this embodiment N is set to 5, corresponding to a duration of 10 milliseconds, which can effectively filter out instantaneous interference while ensuring real-time performance.

[0036] The determined contact state result is output to the control strategy execution module in real time to trigger the control strategy for the corresponding stage. The contact state determination process is executed synchronously with the force signal sampling process. That is, after the force signal acquisition and feature extraction of each sampling point are completed, a contact state determination is performed synchronously to ensure the real-time and synchronous nature of contact state recognition.

[0037] Step S103. In the non-contact stage, a position control strategy is executed to control the robot end effector to move towards the workpiece. In the contact transition stage, a graded force control strategy is executed. In the stable contact stage, a constant force control strategy is executed to maintain a constant contact force between the robot end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set in the contact transition and stable contact stages. When the actual contact force exceeds the force safety upper limit threshold, an abnormal contact state is determined, and the end effector is controlled to stop its feed motion. In the contact transition and stable contact stages, a differential tracker performs real-time tracking processing on the compensated force signal, outputting the tracking value and its differential value of the force signal. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the force differential value is continuously lower than the stability threshold, the current contact state is determined to be stable. In the stable contact stage, the continuous time for the contact force to remain within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is determined to be complete.

[0038] Specifically, in the non-contact stage, there is no effective contact between the robot end effector and the workpiece. At this time, there is no need for force control, only position control strategy is executed to control the robot end effector to move quickly toward the workpiece according to the preset grinding path, so as to improve the overall efficiency of the grinding operation.

[0039] The position control strategy adopts the conventional Cartesian space trajectory planning method for robots. Based on the preset grinding path and moving speed generated by offline programming, the motion parameters of each joint of the robot are obtained through inverse kinematics calculation. The robot's end effector is controlled to perform linear interpolation motion along the preset path. The motion process adopts S-shaped acceleration and deceleration planning to avoid impact during start-up and stopping.

[0040] During the movement of the end effector, the system continuously executes the contact state judgment step. Once the contact transition stage is determined, the position control strategy is immediately stopped and the system switches to the graded force control strategy without delay, achieving seamless connection of control modes.

[0041] In this embodiment, the preset moving speed during the non-contact phase is set to 100 mm / s, which can be adjusted within the range of 20 mm / s to 200 mm / s depending on the workpiece material and grinding precision requirements.

[0042] During the initial contact phase, when the robot's end effector first makes contact with the workpiece, applying a full target contact force at this point would result in a significant impact due to the robot's inertia and positional errors. This could lead to scratches on the workpiece surface, accelerated wear of the grinding head, or even breakage of the grinding head or displacement of the workpiece. Therefore, a graded force control strategy is employed in this phase. By gradually increasing the target contact force, soft contact is achieved, effectively reducing the impact force at the moment of contact.

[0043] The graded force control uses the preset target contact force as a benchmark. During the transition phase, the scaled contact force is used as the control target. As the contact state progresses, the actual expected force is gradually increased until it returns to the full target contact force after entering the stable contact phase.

[0044] Once the system determines that it has entered the contact transition stage, it multiplies the preset target contact force by a first proportional coefficient to obtain the actual expected force for the current stage. The first proportional coefficient is a positive number less than 1; in this embodiment, it is set to 0.3 to 0.5 to ensure that the initial contact force during the transition stage is much smaller than the target contact force, thus suppressing contact impact at its source. Based on this actual expected force, the system adjusts the feed position of the end effector in real time using a force control algorithm, controlling the end effector to gradually approach the workpiece surface with a lower contact force, allowing the contact force to rise smoothly. When the system determines that it has entered the stable contact stage, it immediately restores the actual expected force to the original target contact force, completing the control switch for the contact transition stage and formally entering the constant force grinding state.

[0045] During the stable contact phase, a stable contact has been established between the robot's end effector and the workpiece. At this point, a constant force control strategy is implemented to maintain the contact force stably at the preset target contact force, thereby ensuring the consistency of the amount of material removed during grinding and the uniformity of the surface quality.

[0046] This embodiment uses an admittance control algorithm to achieve constant force control. The admittance control algorithm simulates the interaction process between the robot and the environment as a second-order mass-damped-spring system. Based on the deviation between the real-time measured compensation force signal and the target contact force, the position correction amount of the robot end effector is calculated. By adjusting the normal feed position of the robot end effector, the deviation of the contact force is compensated, so that the contact force remains constant.

[0047] The specific calculation process for admittance control is as follows: Obtain the pre-set target contact force Fd, and read the compensation force signal Fc at the current sampling time; Calculate the contact force deviation e = Fd - Fc; Based on the mass-damped-spring model, the end-effector acceleration correction is calculated from the force deviation: a = (eB) vK x) / M, where M is the virtual mass parameter, B is the virtual damping parameter, K is the virtual stiffness parameter, v is the velocity correction amount at the previous moment, and x is the position correction amount at the previous moment. By performing a numerical integration on the acceleration correction a, the velocity correction v at the current moment is obtained; By performing a numerical integration on the velocity correction amount v, the position correction amount x at the current moment is obtained; The position correction value x is superimposed on the normal coordinates of the robot's original grinding path to correct the feed position of the end effector, thereby causing the contact force to converge toward the target value.

[0048] In this embodiment, the virtual mass parameter M is set to 1 kg, the virtual damping parameter B is set to 1000 N / m, and the virtual stiffness parameter K is set to 0 N / m. Setting the stiffness parameter to 0 indicates that the admittance controller is a purely damped system, which can achieve better force control stability and avoid system oscillation.

[0049] During the polishing process, the system monitors the changes in contact force in real time and continuously corrects the end position through the closed-loop control process described above, maintaining the contact force within the preset allowable fluctuation range.

[0050] During grinding operations, abnormal contact situations such as collisions, workpiece detachment, and grinding head breakage may occur. These situations can lead to a sudden and significant increase in contact force, potentially damaging the robot, grinding head, or workpiece. Therefore, this embodiment adds an abnormal contact protection function to the core method. By pre-setting a force safety upper limit threshold, which is higher than the preset target contact force (in this embodiment, it is set to 150% of the target contact force), the main controller monitors the actual amplitude of the compensation force signal in real time throughout the grinding process, from the contact transition phase to the stable contact phase, and compares the actual contact force with the force safety upper limit threshold. When the actual contact force exceeds the force safety upper limit threshold, the main controller determines that an abnormal contact state is in place and immediately sends a stop-feed command to the robot's motion control module to prevent the contact force from continuing to rise and causing damage to the equipment and workpiece. Through the abnormal contact protection function, equipment risks under abnormal operating conditions can be effectively avoided, improving the safety of grinding operations.

[0051] This embodiment supplements the contact state stability verification mechanism to assist in verifying the reliability of the contact state and avoid misjudgment of the state due to instantaneous fluctuations in the force signal. During the contact transition phase and the stable contact phase, the main controller calls the differential tracker to track and process the real-time compensation force signal. The differential tracker can extract the differential information of the force signal while suppressing noise, outputting a smooth force signal tracking value and its corresponding differential value. A stability threshold is preset, which is set according to the fluctuation range of the normal contact force under grinding conditions. The main controller acquires the absolute value of the force signal differential value in real time and counts the duration it remains below the stability threshold. When the duration reaches the preset verification duration (set to 100 milliseconds in this embodiment), it is determined that the current contact state has reached stability, which can be used as an auxiliary verification basis for contact phase switching, improving the anti-interference capability of state recognition.

[0052] In traditional robotic grinding operations, a single grinding duration is typically used to determine whether grinding is complete. This method does not consider the actual stable state of the contact force, easily leading to insufficient or excessive grinding. Therefore, this embodiment adds a dual-condition automatic grinding completion judgment function to the core method. During the stable contact phase, the main controller simultaneously starts two timing channels: the first channel counts the continuous time the contact force remains within a preset allowable range, which is set to ±10% of the target contact force; the second channel counts the dwell time of the end effector at the current grinding position, i.e., the cumulative time after the end effector enters the current grinding position. A contact force attainment time threshold and a dwell time threshold are preset, respectively. These two thresholds are preset based on the workpiece material, grinding requirements, and grinding head performance. In this embodiment, the contact force attainment time threshold is set to 4 seconds, and the dwell time threshold is set to 5 seconds. When the duration of both timing channels reaches their respective preset thresholds, the main controller determines that the operation at the current grinding position is complete. The dual-condition judgment mechanism can take into account both the polishing time and the stability of the contact force, and more accurately judge the polishing completion, thereby improving the consistency of polishing quality.

[0053] In some embodiments, the method further includes: when it is determined that the work at the current grinding position is completed, controlling the robot end effector to move to the next grinding position to continue executing the contact state judgment and control strategy.

[0054] This embodiment describes the cyclic operation process of multi-station grinding. When the main controller determines that the current grinding position has completed its task, it first exits the constant force control mode and switches to the position control mode, controlling the robot's end effector to move to the next grinding position's contact point along a preset path. Upon reaching the next contact point, the system restarts the contact state judgment process, sequentially executing the complete process of position control in the non-contact stage, graded force control in the contact transition stage, and constant force control in the stable contact stage, completing the task at the next grinding position until all preset grinding positions have been completed. By cyclically executing the state recognition and control process, the automated operation of the entire workpiece grinding process is achieved.

[0055] In some embodiments, the step of executing a graded force control strategy during the contact transition phase includes: during the contact transition phase, scaling the preset target contact force according to a first proportional coefficient as the actual expected force for the current phase, and controlling the end effector to approach the workpiece surface with a force less than the full target contact force; when it is determined that the stable contact phase has been entered, restoring the target contact force to its original value as the actual expected force, and causing the end effector to perform constant force grinding on the workpiece with the full target contact force.

[0056] This embodiment details the process of executing the graded force control strategy in step S103 during the contact transition phase. When the main controller receives the instruction from the contact state judgment module to enter the contact transition phase, it immediately activates the graded force control mode. In this mode, the main controller multiplies the preset target contact force by a first proportional coefficient to obtain the actual expected force for the current phase. The first proportional coefficient is a positive number less than 1; in this embodiment, it is set to 0.3-0.5 to ensure that the initial contact force during the transition phase is much smaller than the target contact force, thus avoiding contact impact. Based on the actual expected force, the main controller adjusts the feed position of the end effector using a force control algorithm, controlling the end effector to gradually approach the workpiece surface with a lower contact force. When the contact state judgment module determines that a stable contact phase has been entered, the main controller restores the actual expected force to the original target contact force, completing the control switch for the contact transition phase and entering the constant force grinding state. Through the graded force control method, the impact force at the moment of contact can be effectively reduced, achieving impact-free soft contact and protecting the grinding head and the workpiece surface.

[0057] In some embodiments, the step of implementing a constant force control strategy during the stable contact phase to maintain a constant contact force between the robot end effector and the workpiece includes: acquiring a preset target contact force during the stable contact phase, calculating the deviation between the real-time compensation force signal and the target contact force, adjusting the feed position of the robot end effector according to the calculated deviation, and maintaining the contact force within a preset allowable range during the grinding process.

[0058] This embodiment details the process of executing the constant force control strategy in step S103 during the stable contact phase. When the main controller receives the command from the contact state judgment module to enter the stable contact phase, it immediately initiates the constant force control mode. This embodiment uses an admittance control algorithm to achieve constant force control. The parameters of the admittance controller are set according to the workpiece material, grinding head type, and grinding requirements. In this embodiment, the mass parameter is set to 1 kg, the damping parameter is set to 1000 N / m, and the stiffness parameter is set to 0 N / m. Setting the stiffness parameter to 0 indicates that the admittance controller is a pure damping system, which can achieve better force control stability. The specific process of constant force control is as follows: The main controller first acquires the pre-set target contact force. Then, it calculates the deviation between the real-time compensation force signal and the target contact force. Next, based on the mass damping spring model, it calculates the acceleration correction of the robot's end effector using the contact force deviation. The acceleration correction is integrated once to obtain the velocity correction, and then the velocity correction is integrated again to obtain the position correction. Finally, the position correction is superimposed on the normal coordinates of the robot's original grinding path, and the feed position of the robot's end effector is adjusted so that the contact force is kept within a preset allowable range near the target contact force.

[0059] In some embodiments, acquiring the real-time force signal of the robot end effector includes: acquiring force signals in multiple directions using force sensors installed on the robot end effector, sampling the acquired force signals at preset fixed time intervals, and transmitting the sampled force signals to the robot's main controller.

[0060] This embodiment details the process of acquiring the real-time force signal of the robot's end effector in step S101. Before executing this method, the calibration and gravity compensation operations of the six-dimensional force sensor are first completed. The sensor calibration operation applies a force and torque of known magnitude and direction to the six-dimensional force sensor using a standard force loading device, establishing a mapping relationship between the sensor's output voltage and the actual force, thus eliminating nonlinear and coupling errors in the sensor. The gravity compensation operation acquires the sensor's output signal under different robot poses, calculates the gravity vector and center of gravity position of the end effector load, and establishes a gravity compensation model. This allows the robot to subtract the load gravity component from the sensor's output signal to obtain the true contact force signal in any pose. After completing the sensor calibration and gravity compensation, the six-dimensional force sensor is activated to begin acquiring the force signal from the robot's end effector. The six-dimensional force sensor is installed between the flange of the robot's end effector and the grinding head, enabling direct measurement of the contact force between the grinding head and the workpiece. The raw force signal acquired by the sensor is transmitted to the robot's main controller via an industrial Ethernet or CAN bus. The main controller samples the raw force signal at preset fixed time intervals. In this embodiment, the sampling frequency is set to 500 Hz, meaning that a force signal at one sampling point is acquired every 2 milliseconds. The main controller stores the sampled force signals in its internal memory for subsequent compensation processing and feature extraction.

[0061] In some embodiments, the step of executing the position control strategy to control the robot end effector to move toward the workpiece during the non-contact phase includes: controlling the robot end effector to move toward the workpiece at a preset speed along a preset grinding path during the non-contact phase, continuously executing the contact state judgment step during the movement, and stopping the execution of the position control strategy when it is determined that the contact transition phase has been entered.

[0062] This embodiment details the process in step S103 where the position control strategy is executed during the non-contact phase to control the robot's end effector to move towards the workpiece. During the non-contact phase, the main controller executes the position control strategy, controlling the robot's end effector to move towards the workpiece along a preset grinding path at a preset speed. The preset grinding path is generated in advance using offline programming software and stored in the robot's main controller's internal memory. The preset speed is set according to the efficiency requirements of the grinding operation and the robot's motion performance; in this embodiment, it is set to 100 millimeters per second. During the movement of the robot's end effector, the main controller performs a contact state judgment each time it collects a force signal from a sampling point. When the contact state judgment module determines that the robot has entered the contact transition phase, the main controller sends a stop position control command to the robot's motion control module and simultaneously sends a command to start graded force control, achieving seamless switching of the control strategy.

[0063] In some embodiments, a parameter self-tuning database is established to pre-store initial parameter sets corresponding to different workpiece materials and grinding head types, including a first preset threshold, a second preset threshold, position control speed, graded force proportional coefficient, admittance controller parameters, etc. When the system starts, the corresponding initial parameter set is loaded according to the current task.

[0064] After the non-contact phase ends, record the maximum impact force and the duration of force oscillation during the contact transition phase. If the maximum impact force exceeds the preset safe impact force limit, reduce the preset speed of the next non-contact phase by 5% to 10%; if the duration of force oscillation exceeds the preset oscillation time limit, reduce the first proportional coefficient of the next contact transition phase by 10% to 15% to reduce the initial contact force of the transition phase and suppress oscillation.

[0065] After the contact transition phase ends, record the root mean square error of the force signal during this transition phase. If the root mean square error exceeds the preset transition accuracy threshold, adjust the difference between the first preset threshold and the second preset threshold, increasing the difference by 5% to 8% to increase the buffer space of the transition phase.

[0066] During the stable contact phase, at each preset parameter tuning period (set to 10 seconds in this embodiment), the standard deviation and average deviation of the compensation force signal within that period are calculated. If the standard deviation exceeds the preset stability threshold, the damping parameter of the admittance controller is increased by 5% to 10%; if the average deviation continues to exceed the preset allowable deviation range, the second preset threshold is finely adjusted to optimize the judgment boundary of the stable contact phase.

[0067] All parameter adjustments are limited to preset safety ranges to avoid system instability caused by excessive parameter adjustments. After each parameter adjustment, the adjusted parameters and corresponding operating condition information are stored in the parameter self-tuning database for parameter initialization in subsequent operations.

[0068] In some embodiments, a master-slave multi-robot control system is constructed, designating one robot as the master robot and the rest as slave robots. The master robot is responsible for global job scheduling and contact state benchmark determination, while the slave robots receive synchronization instructions from the master robot and execute corresponding control strategies.

[0069] All robots independently execute the force signal acquisition, compensation processing, and feature extraction steps of step S101, and transmit the real-time extracted amplitude features to the main control robot via industrial Ethernet.

[0070] The main control robot establishes a global workpiece deformation compensation model and calculates the elastic deformation of the workpiece at each grinding position based on the real-time contact force data of all robots. The measured force signal of each robot is subtracted from the force component caused by the workpiece deformation at the corresponding position to obtain the corrected true compensation force signal.

[0071] Based on the corrected actual compensation force signal, the master robot executes the contact state judgment in step S102 and generates a global contact state reference command. When the master robot determines that it has entered a certain contact stage, it immediately sends a synchronization switching command to all slave robots.

[0072] Upon receiving the synchronization switching command, the slave robot pauses its own contact state judgment logic and immediately switches to the control strategy corresponding to the command. Simultaneously, the slave robot continuously feeds back its force signal data to the master robot to update the workpiece's global deformation compensation model.

[0073] Once all robots have entered the stable contact phase, the main control robot dynamically allocates subsequent grinding areas based on the grinding progress of each robot, ensuring a balanced workload for each robot and avoiding over- or under-grinding in some areas.

[0074] In some embodiments, during the system initialization phase, the reference feed position and reference force signal characteristics of the new grinding head under standard working conditions (preset target contact force, preset feed speed) are recorded and stored in the wear compensation database.

[0075] During the stable contact phase, after completing the work at each grinding position, record the actual feed position and average contact force at that position. Calculate the difference between the current actual feed position and the reference feed position, and the difference between the current average contact force and the reference contact force, under the same grinding path.

[0076] The wear of the grinding head is calculated based on the difference in feed position and the difference in contact force. The formula for calculating the wear is: Wear = Feed position difference × Contact force correction coefficient, where the contact force correction coefficient is preset according to the workpiece material and the type of grinding head. In this embodiment, the contact force correction coefficient is set to 0.8 for metal workpieces and resin grinding heads.

[0077] When the cumulative wear reaches the preset light wear threshold (set to 5% of the grinding head diameter in this embodiment), the system automatically adds a compensation amount equal to the wear amount to the normal coordinates of the subsequent grinding path to ensure that the grinding depth remains unchanged. At the same time, the target contact force is increased by 3% to 5% to compensate for the reduction in contact area caused by the wear of the grinding head.

[0078] When the cumulative wear reaches the preset severe wear threshold (set to 20% of the grinding head diameter in this embodiment), the system issues a grinding head replacement prompt. If a sudden drop in contact force is detected exceeding the preset sudden change threshold (set to 30% of the target contact force in this embodiment), it is determined that the grinding head is broken, and the abnormal contact protection procedure is immediately executed.

[0079] In some embodiments, the surface of the workpiece to be ground is pre-divided into three categories—planar region, convex curved surface region, and concave curved surface region—using 3D modeling software. An independent set of contact state thresholds and control parameters is preset for each type of region. Specifically: for planar regions, a standard parameter set is used; for convex curved surface regions, the second preset threshold is increased by 5% to 10%, and the stiffness parameter of the admittance controller is reduced by 10% to 15% to avoid local overpressure due to small contact area; for concave curved surface regions, the preset speed in the non-contact stage is reduced by 15% to 20%, and the first proportional coefficient in the contact transition stage is reduced by 20% to reduce the initial contact force and prevent the grinding head from colliding with the curved surface sidewall. Before the grinding operation, the divided region information and corresponding parameter sets are imported into the robot's main controller to establish a mapping relationship between the grinding path position and the region parameters.

[0080] During the polishing process, the main controller obtains the current position of the robot's end effector in real time, queries the area type to which the current position belongs based on the mapping relationship, and automatically loads the corresponding parameter set.

[0081] At the transition points between different areas, a gradual parameter switching method is adopted. When the robot enters the transition area, within a preset transition distance (set to 20 mm in this embodiment), the parameters of the previous area are linearly transitioned to the parameters of the next area to avoid contact force impact caused by abrupt parameter changes.

[0082] During the stable contact phase, the tangential component of the force signal is extracted for curved surfaces. When the tangential component exceeds a preset tangential force threshold, the robot end effector's posture is automatically adjusted to ensure the grinding head remains in perpendicular contact with the workpiece surface.

[0083] In some embodiments, a sliding window of length N (set to 20 sampling points in this embodiment) is established to store the amplitude characteristics of the compensation force signal of the most recent N sampling points in real time.

[0084] A linear prediction algorithm is used to predict the amplitude characteristics of the next M sampling points (5 sampling points in this embodiment) based on historical data within a sliding window. The prediction formula is: the feature value of the kth sampling point in the future = the feature value of the current sampling point + k × the average rate of change of the features of the most recent 10 sampling points.

[0085] The predicted feature values ​​of the next M sampling points are compared with a preset threshold to predict whether a contact state transition will occur in the future. If it is predicted that the current state will transition to the next state within the next M sampling points, the pre-control process is initiated in advance.

[0086] The pre-control process is as follows: If it is anticipated that the transition from the non-contact stage to the contact transition stage will be initiated, the moving speed in the non-contact stage will be reduced by 30% in advance, and the graded force control parameters for the contact transition stage will be preloaded. If it is anticipated that the system will switch from the contact transition phase to the stable contact phase, then the actual expected force of the graded force control should be increased in advance to accelerate the transition to the target contact force.

[0087] When the actual contact state judgment result is consistent with the prediction result, the pre-control process continues to be executed; when the actual result is inconsistent with the prediction result, the pre-control process is immediately terminated, the control strategy corresponding to the actual state is switched, the average rate of change calculation parameters of the prediction algorithm are updated, and subsequent prediction results are corrected.

[0088] In some embodiments, by mounting an industrial camera on the robot end effector, with the camera lens facing the contact area between the grinding head and the workpiece, and adjusting the camera's focal length and angle, the contact area is fully presented within the camera's field of view.

[0089] During the system initialization phase, standard images are acquired under three states: no contact, transitional contact, and stable contact. Image features for each state are extracted: No contact state features: There is a significant gap between the grinding head and the workpiece, and the gap width is greater than the preset gap threshold; Transitional contact state features: The grinding head and the workpiece surface are just in contact, and a small light spot appears in the contact area, with the spot area being less than the preset spot area threshold; Stable contact state features: The grinding head and the workpiece surface are in close contact, and a stable light spot is formed in the contact area, with the spot area being greater than or equal to the preset spot area threshold.

[0090] During the polishing process, the industrial camera acquires images of the contact area at a preset sampling frequency (100 Hz in this embodiment), performs preprocessing such as grayscale conversion and edge detection on the images, and extracts the gap width and spot area features of the current image.

[0091] The main controller simultaneously acquires the contact state results determined by the force signal and the contact state results determined by the visual image. If the two results are consistent, the corresponding control strategy is executed; if the two results are inconsistent, a secondary verification process is triggered.

[0092] The secondary verification process is as follows: reduce the movement speed of the robot's end effector by 50% and continuously collect force signals and images for three sampling cycles. If the force signal results in at least two of the three collections are consistent with the visual results, then the consistent results shall prevail; if they are still inconsistent, the sensor is determined to be abnormal, the grinding operation shall be stopped immediately, and an alarm shall be issued.

[0093] In some embodiments, online dynamic optimization of control parameters is achieved by constructing a deep reinforcement learning agent, replacing the fixed-step parameter self-tuning method, and further improving the parameter adaptation capability under complex working conditions.

[0094] Based on the parameter self-tuning mechanism, this embodiment introduces a deep deterministic policy gradient algorithm to construct a reinforcement learning agent, thereby achieving real-time optimization of control parameters. By constructing the state space of the agent, including: the state vector consists of 6-dimensional state variables, namely the current contact force deviation, the differential value of the force signal, the position correction amount, the current workpiece area type identifier, the current admittance damping parameter value, and the current graded force proportional coefficient value, which comprehensively characterize the current running state and parameter configuration of the grinding process.

[0095] Constructing the action space of the intelligent agent: The action vector includes two-dimensional continuous action quantities, including the adjustment amount of the admittance damping parameter and the adjustment amount of the graded force proportional coefficient. The single adjustment amount is limited to ±10% of the current value of the parameter to ensure the smoothness of the parameter adjustment process and avoid system oscillation.

[0096] Design the reward function: The reward function consists of three weighted parts. The first part is the force tracking accuracy reward, where the smaller the contact force deviation, the higher the reward value. The second part is the contact impact penalty, where the larger the maximum impact force during the transition phase, the higher the penalty value. The third part is the grinding efficiency reward, where the shorter the duration of the transition phase, the higher the reward value. The final reward value comprehensively balances grinding quality and grinding efficiency.

[0097] Before system deployment, the grinding conditions of various materials and curved surfaces are simulated in a digital simulation environment to conduct offline pre-training of the reinforcement learning agent. The pre-training steps are no less than 1 million steps, so that the agent can master the basic parameter adjustment strategy. After the pre-training is completed, the trained network model is deployed to the robot's main controller.

[0098] During the actual polishing process, the agent reads the current state value every preset tuning period (set to 5 seconds in this embodiment), outputs the corresponding parameter adjustment action, and executes it. After execution, the reward value is calculated based on the actual polishing effect, and the network model is updated online in small increments to ensure that the control parameters continuously adapt to changes in the current working conditions. Through adaptive optimization using deep reinforcement learning, the control parameters can be automatically optimized for different polishing scenarios, exhibiting stronger generalization ability and better control effect compared to the fixed-step self-tuning method.

[0099] In some embodiments, by introducing acoustic emission sensing signals and force signals for multimodal fusion, dual verification and advanced identification of contact states can be achieved, further improving the accuracy and response speed of state identification.

[0100] An acoustic emission sensor is installed at the junction of the grinding head and the end flange. The sensor's frequency response range covers 10 kHz to 1 MHz. It is used to collect the elastic wave signal generated by the interaction between the grinding head and the workpiece during the grinding contact process. The acoustic emission signal and the six-dimensional force signal are collected synchronously in time. The sampling frequency of the acoustic emission signal is set to 1 MHz.

[0101] The time-domain features and frequency-domain features of the collected acoustic emission signals are extracted respectively. The time-domain features include the signal RMS value, peak value, ring count, and signal energy value; the frequency-domain features include the spectral centroid, peak frequency, and energy proportion of the low, medium, and high frequency bands.

[0102] Multiple sets of acoustic emission feature samples were collected in advance through offline calibration experiments under three states: no contact, contact transition, and stable contact. A contact state classification model was trained based on the support vector machine algorithm to obtain a contact state classifier based on acoustic emission signals.

[0103] In the actual polishing process, the feature vector of the current acoustic emission signal is extracted in real time and input into the trained classification model to obtain the contact state recognition result under the acoustic emission mode. The acoustic emission recognition result and the force signal recognition result are then fused by weighted decision, with the force signal recognition result weighted at 0.6 and the acoustic emission recognition result weighted at 0.4. The final contact state determination result is then output after fusion.

[0104] When the recognition result of one mode changes abnormally, the recognition result of another mode is used as the standard, which can effectively avoid misjudgment of the state caused by single sensor failure or environmental interference. At the same time, the acoustic emission signal is more sensitive to the micro-contact response in the early stage of contact, which can detect the initial contact state in advance and further shorten the response delay of state recognition.

[0105] In some embodiments, by constructing a digital twin of the polishing process, real-time mapping and control strategy pre-simulation optimization of the physical polishing process and the virtual simulation process can be achieved, thereby improving the controllability and yield of the polishing process.

[0106] By using 3D models and physical parameters of the robot, grinding head, and workpiece, a digital twin corresponding to the physical grinding system is constructed in a virtual simulation environment. The twin has built-in robot kinematics model, grinding head dynamics model, workpiece elastic deformation model, and grinding material removal model, which can accurately simulate the changes in contact force and material removal effect during the grinding process.

[0107] During the polishing process, data such as robot joint angles, end effector positions, compensation force signals, and current control parameters of the physical system are synchronized to the digital twin in real time via the industrial bus, driving the virtual model to run synchronously with the physical system and realizing real-time mapping and synchronization of virtual and real states.

[0108] Before each contact state switch, the control parameters to be executed are input into the digital twin, and the subsequent contact process is rehearsed in the virtual environment to predict the magnitude of the contact impact force, force oscillation, and material removal effect. If the rehearsal results show that the impact force exceeds the standard or the force oscillation is too large, the control parameters are iteratively optimized and adjusted in the virtual environment until the rehearsal results meet the process requirements, and then the optimized parameters are sent to the physical robot for execution.

[0109] During the grinding process, the force data from the actual operation of the physical system and the grinding quality inspection data are fed back to the digital twin. Parameter identification algorithms are used to correct parameters such as the grinding coefficient and damping coefficient of the twin model online, continuously improving the simulation accuracy of the twin model and forming a closed-loop optimization mechanism of "virtual pre-simulation - physical execution - feedback correction". Through digital twin pre-simulation control, the effectiveness of control parameters can be verified before execution, avoiding undesirable grinding conditions from the outset and reducing trial-and-error costs. This is particularly suitable for grinding high-value precision workpieces.

[0110] Please see Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of the robot grinding contact state phased identification and adaptive control system provided in this application embodiment. This robot grinding contact state phased identification and adaptive control system is used to execute the steps of the robot grinding contact state phased identification and adaptive control method shown in the above embodiments. The robot grinding contact state phased identification and adaptive control system can be a single server or a server cluster, or the device can be a terminal, such as a handheld terminal, a laptop computer, a wearable device, or a robot.

[0111] like Figure 3 As shown, the robot's phased identification and adaptive control system for grinding contact states includes: The signal acquisition unit 201 is used to acquire the real-time force signal of the robot end effector, process the real-time force signal to obtain the compensation force signal, and extract the amplitude characteristics of the compensation force signal. The state determination unit 202 is used to determine the contact state between the robot and the workpiece based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. The grinding completion unit 203 is used to execute a position control strategy to control the robot end effector to move towards the workpiece during the non-contact stage, execute a graded force control strategy during the contact transition stage, and execute a constant force control strategy during the stable contact stage to maintain a constant contact force between the robot end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set during the contact transition and stable contact stages. When the actual contact force exceeds the force safety upper limit threshold, an abnormal contact state is determined, and the end effector is controlled to stop its feed motion. During the contact transition and stable contact stages, a differential tracker performs real-time tracking processing on the compensated force signal, outputting the tracking value and its differential value of the force signal. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the force differential value is continuously lower than the stability threshold, the current contact state is determined to have reached stability. During the stable contact stage, the continuous time for the contact force to remain within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is determined to be complete.

[0112] In some embodiments, acquiring the real-time force signal of the robot end effector includes: acquiring force signals in multiple directions using force sensors installed on the robot end effector, sampling the acquired force signals at preset fixed time intervals, and transmitting the sampled force signals to the robot's main controller.

[0113] In some embodiments, the step of executing the position control strategy to control the robot end effector to move toward the workpiece during the non-contact phase includes: controlling the robot end effector to move toward the workpiece at a preset speed along a preset grinding path during the non-contact phase, continuously executing the contact state judgment step during the movement, and stopping the execution of the position control strategy when it is determined that the contact transition phase has been entered.

[0114] In some embodiments, the step of executing a graded force control strategy during the contact transition phase includes: during the contact transition phase, scaling the preset target contact force according to a first proportional coefficient as the actual expected force for the current phase, and controlling the end effector to approach the workpiece surface with a force less than the full target contact force; when it is determined that the stable contact phase has been entered, restoring the target contact force to its original value as the actual expected force, and causing the end effector to perform constant force grinding on the workpiece with the full target contact force.

[0115] In some embodiments, the step of implementing a constant force control strategy during the stable contact phase to maintain a constant contact force between the robot end effector and the workpiece includes: acquiring a preset target contact force during the stable contact phase, calculating the deviation between the real-time compensation force signal and the target contact force, adjusting the feed position of the robot end effector according to the calculated deviation, and maintaining the contact force within a preset allowable range during the grinding process.

[0116] In some embodiments, the phased control module is further configured to: when it is determined that the work at the current grinding position is completed, control the robot end effector to move to the next grinding position to continue executing the contact state judgment and control strategy.

[0117] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the robot grinding contact state phased identification and adaptive control system and each module described above can be referred to the corresponding content in the various embodiments of the robot grinding contact state phased identification and adaptive control method, and will not be repeated here.

[0118] The aforementioned method for phased recognition and adaptive control of robot grinding contact states can be implemented as a computer program, which can be used in, for example... Figure 3 It runs on the device shown.

[0119] Please see Figure 4 , Figure 4 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application. The computer device includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0120] The storage medium can store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform any method for phased recognition and adaptive control of the robot's grinding contact state.

[0121] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0122] The internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any kind of robot grinding contact state phased recognition and adaptive control method.

[0123] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the terminal to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0124] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0125] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: The real-time force signal of the robot's end effector is acquired, the real-time force signal is processed to obtain the compensation force signal, and the amplitude characteristics of the compensation force signal are extracted. The contact state between the robot and the workpiece is determined based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. In the non-contact phase, a position control strategy is implemented to move the robot's end effector towards the workpiece. In the contact transition phase, a graded force control strategy is implemented. In the stable contact phase, a constant force control strategy is implemented to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set in both the contact transition and stable contact phases. When the actual contact force exceeds this threshold, an abnormal contact state is identified, and the end effector stops its feed motion. During both phases, a differential tracker tracks the compensated force signal in real time, outputting the tracked value and its differential value. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the differential value remains consistently below the stability threshold, the current contact state is considered stable. In the stable contact phase, the continuous time the contact force remains within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is considered complete.

[0126] In some embodiments, acquiring the real-time force signal of the robot end effector includes: acquiring force signals in multiple directions using force sensors installed on the robot end effector, sampling the acquired force signals at preset fixed time intervals, and transmitting the sampled force signals to the robot's main controller.

[0127] In some embodiments, the step of executing the position control strategy to control the robot end effector to move toward the workpiece during the non-contact phase includes: controlling the robot end effector to move toward the workpiece at a preset speed along a preset grinding path during the non-contact phase, continuously executing the contact state judgment step during the movement, and stopping the execution of the position control strategy when it is determined that the contact transition phase has been entered.

[0128] In some embodiments, the step of executing a graded force control strategy during the contact transition phase includes: during the contact transition phase, scaling the preset target contact force according to a first proportional coefficient as the actual expected force for the current phase, and controlling the end effector to approach the workpiece surface with a force less than the full target contact force; when it is determined that the stable contact phase has been entered, restoring the target contact force to its original value as the actual expected force, and causing the end effector to perform constant force grinding on the workpiece with the full target contact force.

[0129] In some embodiments, the step of implementing a constant force control strategy during the stable contact phase to maintain a constant contact force between the robot end effector and the workpiece includes: acquiring a preset target contact force during the stable contact phase, calculating the deviation between the real-time compensation force signal and the target contact force, adjusting the feed position of the robot end effector according to the calculated deviation, and maintaining the contact force within a preset allowable range during the grinding process.

[0130] In some embodiments, the method further includes: when it is determined that the work at the current grinding position is completed, controlling the robot end effector to move to the next grinding position to continue executing the contact state judgment and control strategy.

[0131] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the steps of the robot grinding contact state phased recognition and adaptive control method provided in any embodiment of this application.

[0132] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMediaCard (SMC), SecureDigital (SD) card, or FlashCard equipped on the computer device.

[0133] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for phased recognition and adaptive control of robot grinding contact states, characterized in that, include: The real-time force signal of the robot's end effector is acquired, the real-time force signal is processed to obtain the compensation force signal, and the amplitude characteristics of the compensation force signal are extracted. The contact state between the robot and the workpiece is determined based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. In the non-contact phase, a position control strategy is implemented to move the robot's end effector towards the workpiece. In the contact transition phase, a graded force control strategy is implemented. In the stable contact phase, a constant force control strategy is implemented to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set in both the contact transition and stable contact phases. When the actual contact force exceeds this threshold, an abnormal contact state is identified, and the end effector stops its feed motion. During both phases, a differential tracker tracks the compensated force signal in real time, outputting the tracked value and its differential value. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the differential value remains consistently below the stability threshold, the current contact state is considered stable. In the stable contact phase, the continuous time the contact force remains within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is considered complete.

2. The method according to claim 1, characterized in that, The acquisition of the real-time force signal from the robot's end effector includes: Force signals from multiple directions are collected by force sensors installed on the robot's end effector. The collected force signals are sampled at preset fixed time intervals and then transmitted to the robot's main controller.

3. The method according to claim 1, characterized in that, The method of executing a position control strategy during the non-contact phase to control the robot's end effector to move toward the workpiece includes: During the non-contact phase, the robot's end effector moves toward the workpiece at a preset speed along a preset grinding path. During the movement, the contact state judgment step is continuously executed. When it is determined that the contact transition phase has been entered, the position control strategy is stopped.

4. The method according to claim 1, characterized in that, The implementation of the graded force control strategy during the contact transition phase includes: During the contact transition phase, the preset target contact force is scaled down by a first proportional coefficient and used as the actual expected force for the current phase. The end effector is controlled to approach the workpiece surface with a force less than the full target contact force. When the stable contact phase is determined, the target contact force is restored to its original value and used as the actual expected force, so that the end effector performs constant force grinding on the workpiece with the full target contact force.

5. The method according to claim 1, characterized in that, The constant force control strategy implemented during the stable contact phase to maintain a constant contact force between the robot's end effector and the workpiece includes: During the stable contact phase, a preset target contact force is obtained, the deviation between the real-time compensation force signal and the target contact force is calculated, and the feed position of the robot end effector is adjusted according to the calculated deviation to maintain the contact force within a preset allowable range during the grinding process.

6. The method according to claim 1, characterized in that, The method further includes: Once the current grinding position is determined to be completed, the robot's end effector is controlled to move to the next grinding position to continue executing the contact state judgment and control strategy.

7. A robot grinding contact state phased recognition and adaptive control system, used to implement the method as described in any one of claims 1-6, characterized in that, include: The signal acquisition unit is used to acquire the real-time force signal of the robot's end effector, process the real-time force signal to obtain the compensation force signal, and extract the amplitude characteristics of the compensation force signal. The state determination unit is used to determine the contact state between the robot and the workpiece based on the amplitude characteristics. When the amplitude characteristics are less than the first preset threshold, it is determined to be a non-contact stage. When the amplitude characteristics are greater than or equal to the first preset threshold and less than the second preset threshold, it is determined to be a contact transition stage. When the amplitude characteristics are greater than or equal to the second preset threshold, it is determined to be a stable contact stage. The grinding completion unit is used to execute a position control strategy to control the robot's end effector to move towards the workpiece during the non-contact phase, execute a graded force control strategy during the contact transition phase, and execute a constant force control strategy during the stable contact phase to maintain a constant contact force between the robot's end effector and the workpiece, thus completing the grinding operation. Specifically, a force safety upper limit threshold is set during the contact transition and stable contact phases. When the actual contact force exceeds the force safety upper limit threshold, an abnormal contact state is determined, and the end effector is controlled to stop its feed motion. During the contact transition and stable contact phases, a differential tracker performs real-time tracking processing on the compensated force signal, outputting the tracking value and its differential value of the force signal. The differential value of the force signal is compared with a preset stability threshold. When the absolute value of the force differential value is continuously lower than the stability threshold, the current contact state is determined to have reached stability. During the stable contact phase, the continuous time for the contact force to remain within a preset allowable range and the dwell time of the end effector at the current position are simultaneously monitored. When both the continuous time and the dwell time reach their respective preset thresholds, the grinding operation at the current position is determined to be complete.

Images