Machine tool assembly step closed-loop control method based on intelligent vision

Abstract

The invention relates to a machine tool assembly step closed-loop control method based on intelligent vision. The problem that in the prior art, an automatic assembly mode is low in assembly detection precision, and consequently dynamic assembly errors are large is solved. The method comprises the following steps: S1, initializing a system and inputting an area; s2, data acquisition and processing; s3, determining and controlling a working area; s4, verifying and outputting a result; s5, testing and operating the system; s6, ending and feeding back; and S7, terminating the system. The device has the advantages that the reliability of assembly quality is improved, the accuracy of assembly operation is ensured, and the accuracy and stability of machine tool assembly are integrally improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent mechanical control technology, specifically to a closed-loop control method for machine tool assembly steps based on intelligent vision. Background Technology

[0002] Intelligent production line automation technology is an important indicator of a country's manufacturing and technological level. It integrates high technologies such as mechanical engineering, electronics, automation, information technology, the Internet of Things, and artificial intelligence. It typically involves the assembly of multiple components. The various stages of intelligent manufacturing production are highly integrated, achieved through the high degree of automation of subsystems such as production planning, manufacturing assembly, and quality assurance. However, in many existing intelligent production lines, some steps still require manual operation, or the overall production line allocation requires manual scheduling, or some automated production processes are complex, resulting in low workpiece processing and storage efficiency, leading to high production costs and low efficiency. Furthermore, existing automated assembly methods suffer from low assembly and inspection accuracy, resulting in large dynamic assembly errors and impacting automated assembly efficiency.

[0003] To address the shortcomings of existing technologies, people have conducted long-term explorations and proposed various solutions. For example, Chinese patent literature discloses an automated production system for CNC machine tool processing [CN202310754387.9], which includes a first conveyor belt and a second conveyor belt. Its features include a CNC machining unit, a first robotic arm, a second robotic arm, a central control unit, and an assembly unit. The central control unit is signal-connected to the CNC machining unit, the first robotic arm, the second robotic arm, and the assembly unit for interactive control.

[0004] The above solution has solved to some extent the problem of complex operation processes and the need for manual scheduling in the existing intelligent assembly production. However, the solution still has many shortcomings, such as low assembly inspection accuracy, which leads to large dynamic assembly errors and affects the efficiency of automated assembly. Summary of the Invention

[0005] The purpose of this invention is to address the above-mentioned problems by providing a closed-loop control method for machine tool assembly steps based on intelligent vision.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a closed-loop control method for machine tool assembly steps based on intelligent vision, comprising the following steps:

[0007] S1, System Initialization and Input Area;

[0008] S2, Data Acquisition and Processing;

[0009] S3. Work area determination and control;

[0010] S4. Verification and result output;

[0011] S5, System Testing and Operation;

[0012] S6. Conclusion and Feedback;

[0013] S7, System Termination.

[0014] In step S1, the control system is powered on and initialized, and enters the data acquisition stage. Before data acquisition, the area and location to be acquired are transmitted. In step S2, during the data acquisition stage, cleaning data is acquired, and component data of the current working area is obtained through sensing devices. Dynamic features of the components are extracted, the acquired data is compared and corrected, and the correction error is checked.

[0015] In the aforementioned closed-loop control method for machine tool assembly steps based on intelligent vision, the sensing devices include sensors, industrial cameras, and piezoelectric ceramic actuators; the component data includes the component's shape, orientation, XYZ coordinates, and reflection intensity; and the types of error correction include lead screw linear guide error and spindle runout error.

[0016] In the aforementioned closed-loop control method for machine tool assembly steps based on intelligent vision, the extraction of dynamic features of components includes the following steps:

[0017] S21. Data Acquisition: Capture the shape, orientation, XYZ coordinates, and reflection intensity of parts during the assembly process using sensing devices, industrial cameras, and piezoelectric ceramic actuators.

[0018] S22, Feature Fusion: Generate a textured 3D model that combines visual and geometric information to provide a foundation for the next step of feature extraction;

[0019] S23. Error Judgment: When the data error of a component exceeds the limit, the control system adjusts through a dynamic compensation mechanism and monitors the assembly process in real time. If the current step fails, the PLC will prevent the start of the device at the next station to ensure the accuracy and quality of the entire assembly process.

[0020] S23. Comparison and Decision-Making: Conduct compliance checks on each step to ensure that each assembly step meets the process requirements.

[0021] In the aforementioned closed-loop control method for machine tool assembly steps based on intelligent vision, the data error exceeding the limit judgment in step S23 includes the following steps:

[0022] S23-1, Comparison and Decision Algorithm: By comparing the model with the current data, determine whether each assembly step meets the process requirements;

[0023] S23-2. Dynamic judgment: If the pose error of the component is less than the set threshold and the thread length meets the requirements, it is judged as qualified; otherwise, an error is prompted and correction is performed.

[0024] S23-2. Joint determination of torque and vision:

[0025] Condition 1: The pose detected by vision is correct (angle error less than 2°) and the torque value is within the process requirements, it is judged as qualified.

[0026] Condition 2: The vision inspection is qualified, but the torque is abnormal, it is judged as loose assembly.

[0027] Condition 3: The torque is qualified, but the vision inspection is not completely passed, it is judged as incorrect installation.

[0028] In the above closed-loop control method for machine tool assembly steps based on intelligent vision, the algorithm of the dynamic compensation mechanism in step S23 is as follows:

[0029]

[0030] In the above closed-loop control method for machine tool assembly steps based on intelligent vision, the errors of the lead screw and linear guide include straightness error (ΔL) and parallelism error (ΔP). Among them, the calculation formula for the straightness error (ΔL) is:

[0031]

[0032] Among them, y(x) is the deviation at the current measurement position, y ideal (x) is the ideal straight trajectory, and L is the total length of the lead screw.

[0033] The calculation formula for the parallelism error (ΔP) is:

[0034]

[0035] Among them, z i is the height of the i-th measurement point, is the average height of all measurement points, and N is the total number of measurement points.

[0036] The final working area is determined by the corrected position deviation. If there is a comparison difference, it is adjusted to the correct area through comparison and the next process continues. After ensuring that the position correction is completed, the control system proceeds to the next step.

[0037] In the above closed-loop control method for machine tool assembly steps based on intelligent vision, the calculation of the spindle runout error includes axial runout calculation (ΔA) and radial runout calculation (ΔR). Among them, the calculation formula for the axial runout calculation (ΔA) is:

[0038]

[0039] Where S1 and S2 are the readings of the two eddy current sensors, and θ is the angle between the two sensors, which is usually 180°.

[0040] The formula for calculating radial runout (ΔR) is as follows:

[0041] ΔR = max(S3) - min(S3),

[0042] S3 is the reading of the radial sensor.

[0043] In the above-mentioned closed-loop control method for machine tool assembly steps based on intelligent vision, step S3 includes the following steps:

[0044] S31. Determine the final working area by correcting the positional deviation;

[0045] S32. If there are discrepancies, adjust to the correct area through comparison and continue to the next process;

[0046] S33. After ensuring that the position correction is completed, the control system will perform further processing;

[0047] Step S4 includes the following steps:

[0048] S41. The control system initiates data verification to confirm the accuracy of the target position.

[0049] S42. After successful confirmation, the result is output through the control system.

[0050] S43. If the correction is successful, the system will perform a cleanup and output the final adjustment result.

[0051] In step S5, after completing the verification and data output in step S3, the control system enters the actual operation stage. In this stage, all actuators complete the debugging work and ensure that the data is completely transmitted to the control system. During the operation, the system monitors in real time to ensure that the operation is accurate. In step S6, after the control system completes its work, it returns to the merging stage and clears the data cache to ensure that the data is clear and accurate. The control system finally confirms the output and marks the process as complete. In step S7, after the control system completes all processes, it automatically shuts down and clears all operating parameters, ending the current workflow.

[0052] Compared with existing technologies, the advantages of this invention are as follows: by combining intelligent vision with multiple sensing devices, comprehensive data collection of components is achieved. Furthermore, relying on precise error judgment, namely lead screw linear guide, spindle runout and dynamic compensation mechanism, the assembly accuracy of the machine tool is guaranteed. Secondly, the assembly angle and posture of each component are determined by torque and vision in combination, and multi-dimensional verification improves the reliability of assembly quality. Combined with real-time monitoring and standardized start-stop cleaning, the accuracy of assembly operation is ensured, thus improving the overall accuracy and stability of machine tool assembly. Attached Figure Description

[0053] Figure 1 This is a flowchart of the method of the present invention; Detailed Implementation

[0054] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0055] like Figure 1 As shown, a closed-loop control method for machine tool assembly steps based on intelligent vision includes the following steps:

[0056] S1, System Initialization and Input Area;

[0057] S2, Data Acquisition and Processing;

[0058] S3. Work area determination and control;

[0059] S4. Verification and result output;

[0060] S5, System Testing and Operation;

[0061] S6. Conclusion and Feedback;

[0062] S7, System Termination.

[0063] In step S1, the control system is powered on and initialized, and enters the data acquisition stage. Before data acquisition, the area and location to be acquired are transmitted. In step S2, during the data acquisition stage, cleaning data is acquired, and component data of the current working area is obtained through sensing devices. Dynamic features of the components are extracted, the acquired data is compared and corrected, and the correction error is checked.

[0064] By extracting dynamic features of components and combining visual information with point cloud data, the ability to detect and correct errors in the assembly process is improved. It can adapt to complex assembly environments, improve assembly efficiency, and reduce the risk of misassembly and omission.

[0065] The sensing equipment includes sensors, industrial cameras, and piezoelectric ceramic actuators; the component data includes the component's shape, orientation, XYZ coordinates, and reflection intensity; the types of error corrections include lead screw linear guide error and spindle runout error.

[0066] Furthermore, the dynamic feature extraction of components includes the following steps:

[0067] S21. Data Acquisition: Capture the shape, orientation, XYZ coordinates, and reflection intensity of parts during the assembly process using sensing devices, industrial cameras, and piezoelectric ceramic actuators.

[0068] S22, Feature Fusion: Generate a textured 3D model that combines visual and geometric information to provide a foundation for the next step of feature extraction;

[0069] Features include:

[0070] ① Bolt head orientation: The orientation of the bolt is described using Euler angles;

[0071] ② Gasket stacking order: The correct order of the gaskets is checked by the height difference along the Z-axis;

[0072] ③ Sealing ring compression: The degree of compression of the sealing ring is detected by the volume change rate.

[0073] S23. Error Judgment: When the data error of a component exceeds the limit, the control system makes adjustments through a dynamic compensation mechanism, such as adjusting the linear guide base and monitoring the assembly process in real time. If the current step fails, the PLC will prohibit the start of the device at the next station to ensure the accuracy and quality of the entire assembly process.

[0074] S23. Comparison and Decision-Making: Conduct compliance checks on each step to ensure that each assembly step meets the process requirements.

[0075] For example, bolt installation inspection:

[0076] Extract point cloud data of the bolt area and calculate the pose deviation. Calculate the visible length of the thread (to determine if it is tightened) and perform a comprehensive judgment. The judgment criteria include pose error and thread length; if a set threshold is met, it is judged as qualified.

[0077] In detail, the data error exceeding the limit judgment in step S23 includes the following steps:

[0078] S23-1, Comparison and Decision Algorithm: By comparing the model with the current data, determine whether each assembly step meets the process requirements; for example: bolt installation detection: extract point cloud data of the bolt area, calculate the pose deviation, and determine the visible length of the thread (to determine whether it is tightened).

[0079] S23-2, Dynamic Judgment: If the positional error of the component is less than the set threshold and the thread length meets the requirements, it is judged as qualified; otherwise, an error is prompted and correction is performed.

[0080] S23-2, Joint determination of torque and vision:

[0081] Condition 1: If the visual inspection pose is correct (angle error less than 2°) and the torque value is within the process requirements (such as 50±5 Nm), it is determined to be qualified.

[0082] Condition 2: If the visual inspection is qualified but the torque is abnormal (such as the torque fails to meet the standard due to thread slipping), it is determined to be an assembly loose connection.

[0083] Condition 3: If the torque is qualified but the visual inspection is not fully passed, it is determined to be an incorrect installation (such as forcibly tightening with misaligned threads).

[0084] The algorithm of the dynamic compensation mechanism in step S23 is as follows:

[0085]

[0086] Where, K p is the proportionality coefficient, T i is the integral time constant, T d is the derivative time constant, and e(t) is the deviation signal.

[0087] Code example:

[0088]

[0089] Preferably, the errors of the lead screw linear guide include the straightness error (ΔL) and the parallelism error (ΔP). Among them, the calculation formula for the straightness error (ΔL) is:

[0090]

[0091] Where, y(x) is the deviation at the current measurement position, y ideal (x) is the ideal straight-line trajectory, and L is the total length of the lead screw;

[0092] The calculation formula for the parallelism error (ΔP) is:

[0093]

[0094] Where, z i is the height of the i-th measurement point, is the average height of all measurement points, and N is the total number of measurement points.

[0095] Determine the final working area through the corrected position deviation. If there is a comparison difference, adjust to the correct area through comparison and continue with the next process. After ensuring that the position correction is completed, the control system proceeds to the next step.

[0096] Specifically, the calculation of spindle runout error includes axial runout calculation (ΔA) and radial runout calculation (ΔR). The formula for calculating axial runout (ΔA) is as follows:

[0097]

[0098] Where S1 and S2 are the readings of the two eddy current sensors, and θ is the angle between the two sensors, which is usually 180°.

[0099] The formula for calculating radial runout (ΔR) is as follows:

[0100] ΔR = max(S3) - min(S3),

[0101] S3 is the reading of the radial sensor.

[0102] Step S3 includes the following steps:

[0103] S31. Determine the final working area by correcting the positional deviation;

[0104] S32. If there are discrepancies, adjust to the correct area through comparison and continue to the next process;

[0105] S33. After ensuring that the position correction is completed, the control system will perform further processing;

[0106] Step S4 includes the following steps:

[0107] S41. The control system initiates data verification to confirm the accuracy of the target position.

[0108] S42. After successful confirmation, the result is output through the control system.

[0109] S43. If the correction is successful, the system will perform a cleanup and output the final adjustment result.

[0110] In step S5, after completing the verification and data output in step S3, the control system enters the actual operation stage. In this stage, all actuators complete the debugging work and ensure that the data is completely transmitted to the control system. During the operation, the system monitors in real time to ensure that the operation is accurate. In step S6, after the control system completes its work, it returns to the merging stage and clears the data cache to ensure that the data is clear and accurate. The control system finally confirms the output and marks the process as complete. In step S7, after the control system completes all processes, it automatically shuts down and clears all operating parameters, ending the current workflow.

[0111] In summary, the principle of this embodiment is as follows: This invention combines a multi-dimensional, multi-sensor dual error prevention mechanism with data compensation to achieve efficient error detection and correction, ensuring the accuracy and reliability of the assembly process and reducing assembly errors caused by misjudgment of a single sensor or data interference.

[0112] The dual error prevention mechanism includes the following:

[0113] 1. Multi-sensor data fusion

[0114] Visual error prevention: When relying solely on a vision system, errors may occur due to external factors such as lighting and angle, especially when detecting the installation position of bolts. The vision system may misjudge the thread angle or position, leading to incorrect assembly decisions.

[0115] Torque sensor error prevention: Relying solely on a torque sensor cannot detect some common assembly problems, such as bolts not being fully inserted or loose connections (i.e., the torque is correct but the bolt is not fully in place). In this case, the torque value may meet the standard, but the assembly has not fully achieved the expected requirements.

[0116] By combining visual recognition with torque monitoring data, the system can perform dual verification in the following ways:

[0117] Visual-torque joint judgment: When there is an inconsistency between visual detection and torque data, the system will conduct a comprehensive analysis and provide an error message.

[0118] 2. Redundant sensor settings

[0119] Multiple sensor placement: To improve accuracy and reliability, the system places multiple sensors at critical locations such as spindle runout detection and bolt tightening processes. For example, axial runout detection uses two symmetrically arranged eddy current sensors to eliminate the effects of eccentric installation.

[0120] Redundant data verification: By employing redundant designs with multiple sensors, false alarms can be eliminated by comparing data from different sensors when anomalies occur. For example, in vibration detection, if the measurement results from two sensors differ significantly, the problem can be identified as a sensor issue, thereby improving data reliability.

[0121] 3. Comprehensive analysis of error types

[0122] Multimodal data processing: Combining data from vision, torque, pose, geometry, and other aspects for comprehensive judgment. This allows errors that a single sensor cannot detect to be supplemented and verified by other sensors, thereby improving the system's anti-interference capability.

[0123] Anomaly Feedback: When an error is detected, the system will handle the anomaly through alarms, prompts, or repair guidance. For example, during bolt installation, if the sequence is incorrect or the torque difference is too large, the system will trigger a visual alarm or torque balancing alarm to remind the operator to make adjustments.

[0124] 4. Technical advantages of dual error prevention

[0125] Improve assembly accuracy: The dual error prevention mechanism reduces the risk of misjudgment by a single sensor by combining multiple sensors and multiple data sources, thereby improving the overall assembly accuracy.

[0126] Reduce misassembly and omissions: Through dual verification of visual and physical parameters (such as torque), the system can promptly detect and correct common problems such as omissions and loose connections, significantly reducing the assembly error rate.

[0127] Preventing interference and misjudgments: In complex assembly environments, a single sensor may misjudge due to interference from external factors. A dual error-proofing mechanism ensures accurate execution of each assembly step through cross-validation of multiple data sources.

[0128] 5. Feedback and Correction Mechanism

[0129] The system automatically feeds back errors in the assembly process to the control system through real-time data acquisition and feedback, and adjusts the assembly strategy in real time to ensure that each step meets the process requirements.

[0130] If an anomaly is detected, the system will automatically trigger corrective steps and notify the operator through the alarm system to intervene, thereby preventing the error from spreading to the next process.

[0131] Data processing includes the following:

[0132] 1. Compliance assessment of procedures

[0133] Comparison and Decision Algorithm: By comparing the model with the current data, determine whether each assembly step meets the process requirements.

[0134] Dynamic judgment: If the positional error is less than the set threshold and the thread length meets the requirements, it is judged as qualified; otherwise, an error is prompted and correction is performed.

[0135] Joint determination of torque and vision: pose and torque are detected by vision.

[0136] 2. Data feedback and closed-loop control

[0137] Process step control: The detection results of each step are fed back to the control system to determine whether to proceed to the next process. For example, if the current step fails, the control system will prevent the equipment at the next station from starting.

[0138] Anomaly Handling and Correction: If a step fails a check, the control system will guide the operator to make adjustments through alarms and correction prompts. If the alarm persists for more than 5 minutes, the control system will automatically generate a maintenance work order and report the error information to the MES system.

[0139] 3. Dynamic compensation mechanism

[0140] Error compensation: When the control system detects that certain errors exceed the limits (such as linear guide straightness error or excessive spindle runout), the control system will automatically trigger the compensation mechanism to dynamically adjust the base through fine-tuning devices such as the piezoelectric ceramic platform to ensure assembly accuracy.

[0141] 4. Multi-sensor collaboration and data fusion

[0142] Vision + Torque + Geometric Data Fusion: The control system integrates data from multiple sensors (such as vision sensors, torque sensors, and geometric measurement sensors) to make multi-dimensional judgments and corrections. This enables the control system to more comprehensively detect and correct various errors in the assembly process.

[0143] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A closed-loop control method for machine tool assembly steps based on intelligent vision, characterized in that, It includes the following steps: S1. System initialization and input area; S2. Data acquisition and processing; S3. Working area determination and control; S4. Verification and result output; S5. System testing and operation; S6. End and feedback; S7. System termination.

2. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 1, characterized in that, In step S1, the control system powers on and initializes and enters the data acquisition stage. Before data acquisition, the area and position to be collected are conveyed. In step S2, during the data acquisition stage, the cleaning data is acquired, and the component data of the current working area is obtained through the sensing device, the dynamic characteristics of the components are extracted, the collected data is compared and corrected, and the correction error is checked.

3. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 2, characterized in that, The said sensing device includes sensors, industrial cameras and piezoelectric ceramic exciters; the said component data includes the shape, standing position, XYZ coordinates and reflection intensity of the components; the types of the said correction error include lead screw linear guide error and spindle runout error.

4. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 3, characterized in that, The extraction of the said component dynamic characteristics includes the following steps: S21. Data acquisition: The shape, standing position, XYZ coordinates and reflection intensity of the parts during the assembly process are captured through the sensing device, industrial camera and piezoelectric ceramic exciter; S22. Feature fusion: A textured 3D model is generated, combining visual and geometric information, providing a basis for the next step of feature extraction; S23. Error determination: When the data error of the components exceeds the limit, the control system adjusts through the dynamic compensation mechanism and monitors the assembly process in real time. If the current step fails, the PLC will prohibit the start of the device at the next station, ensuring the accuracy and quality of the entire assembly process; S23. Comparison and decision-making: The compliance determination of the steps is carried out to make each assembly step meet the process requirements.

5. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 4, characterized in that, The judgment of the data error exceeding the limit in step S23 includes the following steps: S23-1. Comparison and decision-making algorithm: By comparing the model and the current data, it is judged whether each assembly step meets the process requirements; S23-2. Dynamic judgment: If the pose error of the components is less than the set threshold and the thread length meets the requirements, it is judged as qualified, otherwise an error is prompted and corrected; S23-2. Joint judgment of torque and vision: Condition 1: The visual inspection pose is correct (angle error less than 2°) and the torque value is within the process requirements, judged as qualified, Condition 2: The visual inspection is qualified but the torque is abnormal, judged as loose connection in assembly, Condition 3: The torque is qualified but the visual inspection is not fully passed, judged as incorrect installation.

6. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 4, characterized in that, The algorithm of the dynamic compensation mechanism in step S23 is:

7. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 3, characterized in that, The said lead screw linear guide error includes linearity error (ΔL) and parallelism error (ΔP), where the calculation formula of the linearity error (ΔL) is: Where y(x) is the deviation of the current measurement position, y ideal (x) is the ideal straight line trajectory, and L is the total length of the lead screw; The calculation formula of the said parallelism error (ΔP) is: Among them, Z i It is the height of the i-th measurement point. is the average height of all measurement points, and N is the total number of measurement points. The final working area is determined through the corrected position deviation. If there is a comparison difference, it is adjusted to the correct area through comparison and the next process continues. After ensuring that the position correction is completed, the control system proceeds to the next processing.

8. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 7, characterized in that, The calculation of the said spindle runout error includes axial runout calculation (ΔA) and radial runout calculation (ΔP), where the calculation formula of the axial runout calculation (ΔA) is: Where S1 and S2 are the readings of the two eddy current sensors, and θ is the angle between the two sensors, which is usually 180°. The formula for calculating the radial runout (ΔR) is as follows: ΔR = max(S3) - min(S3), S3 is the reading of the radial sensor.

9. The closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 1, characterized in that, Step S3 includes the following steps: S31. Determine the final working area by correcting the positional deviation; S32. If there are discrepancies, adjust to the correct area through comparison and continue to the next process; S33. After ensuring that the position correction is completed, the control system will perform further processing; Step S4 includes the following steps: S41. The control system initiates data verification to confirm the accuracy of the target position. S42. After successful confirmation, the result is output through the control system. S43. If the correction is successful, the system will perform a cleanup and output the final adjustment result.

10. A closed-loop control method for machine tool assembly steps based on intelligent vision according to claim 9, characterized in that, In step S5, after the verification and data output in step S3 are completed, the control system enters the actual operation stage. In this stage, all actuators complete the debugging work and ensure that the data is completely transmitted to the control system. During the operation, the system monitors in real time to ensure that the operation is accurate. In step S6, after the control system completes its work, it returns to the merging stage and clears the data cache to ensure that the data is clear and accurate. The control system then confirms the output and marks the process as complete. In step S7, after the control system completes all processes, it automatically shuts down and clears all operating parameters, ending the current workflow.

Images