Polishing multi-element control method of industrial robot for casting polishing

Abstract

The invention relates to a polishing multi-element control method of an industrial robot for casting polishing. An industrial robot; the visual sensor comprises a laser positioning system and an image processing system; the pressure sensor is used for monitoring the polishing force during polishing; the position sensor is used for providing accurate position information for motion control of the industrial robot by means of a position gyroscope; the controller provides a control decision according to the received information and adjusts motion and operation of the industrial robot; according to the polishing multi-element control method of the industrial robot for casting polishing, a control framework is established by means of information collected by the visual sensor, the pressure sensor and the position sensor, and the control framework fuses the obtained information and then transmits the fused information to the controller; the method has the beneficial effects that each sensor can make a decision independently according to data collected by the sensor, decision results independently generated by different sensors are efficiently fused in a decision layer, and various states in the polishing process are grasped more accurately.

Description

Technical Field

[0001] The invention relates to the technical field of industrial robot application control, and in particular to a multi-element control method for industrial robot grinding for casting grinding. Background Art

[0002] With the rapid development of industrial intelligent manufacturing, industrial robots, as intelligent equipment, have been widely used in industries such as steel, casting, and aviation products. At the same time, the requirements for the control systems of industrial robots are getting higher and higher. At present, the practical application of industrial robot equipment is mainly based on on-site debugging path points, but with the continuous maturity of technology and more precise requirements for industrial robots.

[0003] A Chinese patent with application number CN202310617944.2 discloses a navigation method for an industrial inspection robot based on multi-sensor fusion. The method uses a preset algorithm to control the industrial robot to set its position, which reduces the labor cost of debugging and enables the industrial robot to accurately follow the specified trajectory. However, this method is difficult to develop, and the technical solution causes the robot's response rhythm to be too slow, resulting in low production efficiency in the later stage.

[0004] A Chinese patent with application number CN201810862389.9 discloses a remote-controlled robot polishing control system based on multi-sensor fusion. The system uses an industrial camera to scan and collect images of the workpiece to be processed, and replaces the image features and contour information of the workpiece, thereby reducing the error caused by manual debugging and making the prescribed trajectory of the industrial robot accurate. However, this method is difficult to develop, and the overall integration of the control system described in the patent is not high, resulting in some functional waste and cost waste.

[0005] Therefore, there is an urgent need for a high-precision control method that is simple in control and utilizes the commonly used diversified eigenvalues ​​of industrial robots for coupling to polish the detailed defects of castings in daily production practice. Summary of the invention

[0006] In order to overcome the problems of slow response and complex manual debugging of the industrial robot grinding control method in the prior art, the present invention uses the three-dimensional parameters of vision, force control (pressure sensing), and position control (gyroscope position sensing) for coupling, provides a grinding method and strategy, and improves the grinding accuracy. The technical solution adopted by the present invention is: a multi-element control method for industrial robot grinding for casting grinding, comprising:

[0007] casting;

[0008] Industrial robots;

[0009] The casting is transported to within the operating radius of the industrial robot;

[0010] Visual sensors, including: laser positioning system and image processing system;

[0011] Pressure sensor to monitor grinding force during grinding;

[0012] A position sensor, which provides accurate position information for motion control of the industrial robot by means of a position gyroscope;

[0013] a controller that provides control decisions based on the received information to adjust the movement and operation of the industrial robot;

[0014] The multi-control method for industrial robot polishing of castings is to establish a control framework with the help of information collected by the visual sensor, pressure sensor, and position sensor, and the control framework fuses the acquired information and transmits it to the controller.

[0015] Furthermore,

[0016] The control architecture is hierarchical and includes:

[0017] The data layer performs preliminary fusion of the raw data of the visual sensor, the pressure sensor and the position sensor at the bottom layer and extracts feature values;

[0018] Feature layer, the feature values ​​extracted from the data layer are fused using principal component analysis;

[0019] The decision layer uses the Kalman filter method to obtain a fused estimate based on the information obtained from the feature layer and the specification information of the casting.

[0020] Furthermore,

[0021] Setting a threshold value and prior information of a pass rate in the controller;

[0022] The posterior information is obtained by using the Bayesian inference method according to the information obtained by the decision layer, and the posterior information is compared with the threshold to determine whether it is qualified.

[0023] Furthermore,

[0024] The information collected by the visual sensor, pressure sensor, and position sensor is processed using a smoothing filter method before entering the control architecture for processing.

[0025] Furthermore,

[0026] Before the information collected by the visual sensor, pressure sensor, and position sensor enters the control architecture for processing, the Laida law is used to eliminate abnormal values.

[0027] Furthermore,

[0028] The information collected by the visual sensor, pressure sensor, and position sensor is normalized before entering the control architecture for processing.

[0029] Furthermore,

[0030] There are two industrial robots;

[0031] The two industrial robots are controlled collaboratively and perform grinding task distribution.

[0032] Furthermore,

[0033] The grinding equipment for grinding the casting and the two industrial robots are arranged in a triangular plane;

[0034] Two of the industrial robots grasp one side of the casting, and another industrial robot grasps the polished side and completes the polishing work on the unpolished side.

[0035] Furthermore,

[0036] The polishing task is divided into a plurality of subtasks, and the two industrial robots respectively undertake a plurality of the subtasks, or the distribution of the subtasks is dynamically adjusted according to the specifications of the casting.

[0037] Furthermore,

[0038] The subtasks are divided into a fine grinding task and a rough grinding task, and the two industrial robots process the fine grinding task and the rough grinding task respectively.

[0039] The beneficial effects of the present invention compared with the prior art are as follows: 1) Each sensor can make independent decisions based on the data collected by itself. Within their respective working ranges, they obtain targeted decision results by monitoring and analyzing different parameters. The decision results independently generated by different sensors are efficiently fused at the decision-making layer. This fusion method can combine the advantages of each sensor and more accurately grasp the various states in the polishing process. 2) The system has established a polynomial compensation model, which mainly compensates for the nonlinear data errors that occur in the sensor during data collection and processing. By compensating for nonlinear data errors, the accuracy and reliability of the data can be greatly improved, thereby making the polishing effect of the industrial robot more accurate. 3) The collaborative polishing control method of dual industrial robots is adopted to make the polished workpiece points more comprehensive, and dual industrial robots are used for polishing. In the early stage, the initial points to be polished are determined by manual debugging and calibration, and then the points are offset and adjusted by robot control to ensure that the initial polishing accuracy meets production requirements. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments are described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0041] In order to overcome the problems of slow response and complex manual debugging of industrial robot grinding control methods in the prior art, the present invention uses the three-dimensional parameters of vision, force control (pressure sensing), and position control (gyroscope position sensing) for coupling, provides a grinding method and strategy, and improves the grinding accuracy. The technical solution adopted in this embodiment is: a multi-element control method for industrial robot grinding for casting grinding, comprising:

[0042] Castings; generally refers to castings produced by mold casting.

[0043] Industrial robot; Industrial robot generally refers to a robotic arm with six degrees of freedom. During grinding operations, the workpiece is ground using an industrial robot-based grinding method. According to the grinding process requirements, the robot's motion path is determined, and the robot's trajectory is planned using offline programming software. The trajectory is planned based on the three-dimensional model of the workpiece, and then the program is downloaded to the robot controller to ensure the comprehensiveness and accuracy of the grinding.

[0044] The casting is transported to within the operating radius of the industrial robot;

[0045] Visual sensors, including: laser positioning system and image processing system;

[0046] The visual sensor mainly includes a laser positioning system and an image processing system. The laser positioning system collects the specific position information of the workpiece through the transformation of the distance position collected by the laser, sends it to the robot end, and realizes the edge finding function through the interrupt system. The image processing system extracts the feature information of the workpiece by analyzing and processing the workpiece, and realizes the recognition of the workpiece specifications through feature point extraction and template matching algorithms.

[0047] Pressure sensor to monitor grinding force during grinding;

[0048] The piezoelectric effect of piezoelectric materials is used to generate electric charges when a force is applied. The amount of charge is proportional to the force, and the force information is obtained by measuring the charge. In industrial robot grinding, due to certain errors in the workpiece, if only a fixed trajectory is used for grinding, the scrap rate will be too high. It is necessary to add a pressure sensor to monitor the grinding force in real time. By controlling the magnitude of the grinding force, the stability of the grinding quality can be guaranteed. When the grinding force is too large, it may cause excessive wear or even damage to the surface of the workpiece; if the force is too small, the expected grinding effect may not be achieved. The pressure sensor feeds back the detected force signal to the control system, and the control system offsets the robot's motion trajectory points according to the preset force threshold to achieve precise control of the grinding force.

[0049] A position sensor, which provides accurate position information for motion control of the industrial robot by means of a position gyroscope;

[0050] The principle of conservation of angular momentum is used to measure the robot's posture changes.

[0051] In grinding control, the position gyroscope provides precise position information for the robot's motion control, ensuring that the robot can accurately reach the predetermined grinding position and maintain the correct posture, which is crucial to achieving high-precision grinding operations.

[0052] a controller that provides control decisions based on the received information to adjust the movement and operation of the industrial robot;

[0053] The multi-control method for industrial robot polishing of castings is to establish a control framework with the help of information collected by the visual sensor, pressure sensor, and position sensor, and the control framework fuses the acquired information and transmits it to the controller.

[0054] The characteristic of this technical solution is to couple the multi-parameters of visual sensors, pressure sensors and position sensors to achieve precise control of the grinding work.

[0055] In other embodiments, preferably,

[0056] The control architecture is hierarchical and includes:

[0057] The data layer performs preliminary fusion of the raw data of the visual sensor, the pressure sensor and the position sensor at the bottom layer and extracts feature values;

[0058] Feature layer, the feature values ​​extracted from the data layer are fused using principal component analysis;

[0059] The decision layer uses the Kalman filter method to obtain a fused estimate based on the information obtained from the feature layer and the specification information of the casting.

[0060] Using the Kalman filter-based method, the system state is estimated and fused through two processes: prediction and update. In multi-sensor fusion, the measurement value of each sensor is regarded as an observation of the system state. According to the dynamic model of the system and the observation model of the sensor, the Kalman filter algorithm is used to predict and update the system state, thereby obtaining the fused state estimate. In robot motion control, the position and velocity of the robot are estimated by robot data fusion. Let the state vector of the system be

[0061] x=[x,v] T (1)

[0062] Where x is the position, v is the velocity, and the state equation is

[0063]

[0064] Where (A is the state transfer matrix, B is the control input matrix, u k is the control input, ω k is the system noise), and the observation equation is

[0065]

[0066] Where (H is the observation matrix, v k is the observation noise). Kalman filtering calculates the optimal state estimate by continuously iterating the prediction and update steps. It can effectively process noisy sensor data, make real-time and accurate estimates of the system state, and can adaptively adjust the estimation error. It has good performance in tracking and predicting dynamic systems.

[0067] During the grinding process of industrial robots, it is necessary to accurately obtain a variety of information such as the position of the workpiece, the casting error of the workpiece, and the grinding force in order to achieve precise control. A single sensor is often difficult to meet the needs of comprehensive and accurate information acquisition. Multi-sensor fusion technology can comprehensively utilize the advantages of different sensors and complement information, thereby providing robots with richer and more accurate perception capabilities. Taking the foundry industry as an example, due to certain errors in the workpieces produced by the foundry industry, the original trajectory set by the industrial robot will result in poor grinding effects. The fusion of visual sensors, force sensors, and position sensors can achieve more accurate grinding operations, improve grinding quality and efficiency, and reduce the risk of damage to workpieces and industrial robots.

[0068] Further, preferably,

[0069] Setting a threshold value and prior information of a pass rate in the controller;

[0070] The posterior information is obtained by using the Bayesian inference method according to the information obtained by the decision layer, and the posterior information is compared with the threshold to determine whether it is qualified.

[0071] Since it is necessary to judge whether the polishing quality meets the standard after polishing, Bayesian reasoning method is used to verify the algorithm. Using Bayesian theorem, the posterior probability of the event is updated according to the prior knowledge and the evidence provided by the sensor, so as to make the best decision. Let event A be the situation that needs to be judged (such as whether the workpiece is qualified), and the evidence provided by sensor i is B i The prior probability P(A) represents the probability estimate of event A when there is no sensor information, and the conditional probability P(B i |A) means that sensor i observes evidence B when event A occurs. iThe probability of , Bayes' theorem is Based on the evidence from multiple sensors, the posterior probability P(A|B1,B2...B n ), when the posterior probability exceeds the set threshold, a corresponding decision is made. That is, for the judgment of the grinding quality of the workpiece, the posterior probability of the workpiece being qualified is calculated using Bayesian reasoning based on evidence such as whether the grinding force detected by the force sensor is within a reasonable range and the surface roughness of the workpiece observed by the visual sensor, thereby determining whether the workpiece is qualified.

[0072] In some embodiments, preferably,

[0073] The information collected by the visual sensor, pressure sensor, and position sensor is processed using a smoothing filter method before entering the control architecture for processing.

[0074] Since the data collected by the polishing robot's state perception system may have certain deviations, the data obtained by its original sensors must be preprocessed to ensure its accuracy and reliability.

[0075] During the operation of the polishing robot, in order to effectively filter out disordered random abnormal data, a simple moving average smoothing filtering method is adopted, which can better process the original data and thus improve the performance and stability of the system, as shown in the following formula.

[0076]

[0077] Among them, pt is the filtering result at time point t, and xt-i is the observation value at ti.

[0078] In some embodiments, preferably,

[0079] Before the information collected by the visual sensor, pressure sensor, and position sensor enters the control architecture for processing, the Laida law is used to eliminate abnormal values.

[0080] In order to ensure the normal operation of the grinding robot equipment, a reliable threshold needs to be determined. If this threshold is exceeded, it can be identified as abnormal data. Next, use Laida's law to eliminate these outliers, as shown in the following formula.

[0081]

[0082] In some embodiments, it is preferred

[0083] The information collected by the visual sensor, pressure sensor, and position sensor is normalized before entering the control architecture for processing.

[0084] By normalizing the original state data of the polishing robot and reducing its numerical space to the range of [0,1], the model can be effectively converged and the accuracy of the algorithm can be greatly improved, as shown in the following formula.

[0085]

[0086] Among them: Xmax is the maximum value of all monitored values, and Xmin is the minimum value.

[0087] In other embodiments, specifically,

[0088] There are two industrial robots;

[0089] The two industrial robots are controlled collaboratively and perform grinding task distribution.

[0090] Preferably,

[0091] The grinding equipment for grinding the casting and the two industrial robots are arranged in a triangular plane;

[0092] Two of the industrial robots grasp one side of the casting, and another industrial robot grasps the polished side and completes the polishing work on the unpolished side.

[0093] Preferably,

[0094] The grinding task is divided into several subtasks, and the two industrial robots respectively undertake several subtasks, or dynamically adjust the subtask distribution according to the casting specifications. The subtasks are divided into fine grinding tasks and rough grinding tasks, and the two industrial robots respectively handle the fine grinding tasks and rough grinding tasks.

[0095] Specifically, during operation: please refer to a pipe grinding system (application number: 202321663700.X, publication number: CN220389003U) disclosed an industrial robot application grinding technology solution. When two industrial robots work together to grind, one assists the other in grinding and grasping. For the parts that cannot be polished on the grasping side, the grinding method / strategy described in this embodiment is not used to achieve control of the grinding quality, and realize multi-dimensional control of industrial robot grinding for casting grinding.

[0096] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-element control method for industrial robot grinding of castings, characterized in that: include: casting; Industrial robots; The casting is transported to within the operating radius of the industrial robot; Visual sensors, including: laser positioning system and image processing system; Pressure sensor to monitor the grinding force during grinding; A position sensor, which provides accurate position information for motion control of the industrial robot by means of a position gyroscope; a controller that provides control decisions based on the received information to adjust the movement and operation of the industrial robot; The multi-control method for industrial robot polishing of castings is to establish a control framework with the help of information collected by the visual sensor, pressure sensor, and position sensor, and the control framework fuses the acquired information and transmits it to the controller.

2. The multi-element control method for industrial robot grinding of castings according to claim 1 is characterized in that: The control architecture is hierarchical and includes: The data layer performs preliminary fusion of the raw data of the visual sensor, the pressure sensor and the position sensor at the bottom layer and extracts feature values; Feature layer, the feature values ​​extracted from the data layer are fused using principal component analysis; The decision layer uses the Kalman filter method to obtain a fused estimate based on the information obtained from the feature layer and the specification information of the casting.

3. The multi-element control method for industrial robot grinding of castings according to claim 2 is characterized in that: Setting a threshold value and prior information of a pass rate in the controller; The posterior information is obtained by using the Bayesian inference method according to the information obtained by the decision layer, and the posterior information is compared with the threshold to determine whether it is qualified.

4. The multi-element control method for industrial robot grinding of castings according to claim 1 is characterized in that: The information collected by the visual sensor, pressure sensor, and position sensor is processed using a smoothing filter method before entering the control architecture for processing.

5. The multi-element control method for industrial robot grinding of castings according to claim 1, characterized in that: Before the information collected by the visual sensor, pressure sensor, and position sensor enters the control architecture for processing, the Laida law is used to eliminate abnormal values.

6. The multi-element control method for industrial robot grinding of castings according to claim 1, characterized in that: The information collected by the visual sensor, pressure sensor, and position sensor is normalized before entering the control architecture for processing.

7. The multi-element control method for industrial robot grinding of castings according to claim 1, characterized in that: There are two industrial robots; The two industrial robots are controlled collaboratively and perform grinding task distribution.

8. The multi-element control method for industrial robot grinding of castings according to claim 7, characterized in that: The grinding equipment for grinding the casting and the two industrial robots are arranged in a triangular plane; Two of the industrial robots grasp one side of the casting, and another industrial robot grasps the polished side and completes the polishing work on the unpolished side.

9. The multi-element control method for industrial robot grinding of castings according to claim 7, characterized in that: The polishing task is divided into a plurality of subtasks, and the two industrial robots respectively undertake a plurality of the subtasks, or the distribution of the subtasks is dynamically adjusted according to the specifications of the casting.

10. The multi-element control method for industrial robot grinding of castings according to claim 9, characterized in that: The subtasks are divided into a fine grinding task and a rough grinding task, and the two industrial robots process the fine grinding task and the rough grinding task respectively.

Images