Control method and system and computer readable storage medium

Abstract

The invention relates to a control method and system and a computer readable storage medium, and the control method comprises the steps: obtaining an image of a target body from a visual sensor, extracting the current feature information of the target body based on the image of the target body, modeling an expansion state space based on the current feature information of the target body and the motion information of the visual sensor, and carrying out the modeling of the expansion state space. Discretization model predictive control is adopted, optimal control input is obtained by solving a quadratic programming problem, the optimal control input serves as a motion control instruction to be sent to an action execution body, and the action execution body executes a corresponding action on a target body. The technical problem that most of control systems with industrial mechanical arms at present can only complete specified actions and lack the ability to track a target body, so that if mechanical parts are not located at specified positions, the industrial mechanical arms of the control systems are difficult to complete a target task is solved.

Description

Technical Field

[0001] This invention relates to the field of predictive control technology, and in particular to a control method and system, as well as a computer-readable storage medium. Background Technology

[0002] Robotic arms, especially those used in industry, are closely integrated with workers and mechanical parts in industrial environments. Control systems equipped with industrial robotic arms automatically complete tasks such as assembling mechanical parts on assembly lines once the work parameters are set by the workers. However, most current control systems with industrial robotic arms can only perform designated actions and lack the ability to track target objects. This means that if mechanical parts are not in the designated position, the industrial robotic arm will struggle to complete the target task. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a control method and system, as well as a computer-readable storage medium.

[0004] The technical solution to achieve the purpose of this invention is as follows:

[0005] In a first aspect, embodiments of this application provide a control method, including:

[0006] Acquire images of the target object using a visual sensor;

[0007] Based on the image of the target object, extract the current feature information of the target object;

[0008] Based on the current feature information of the target and the motion information of the vision sensor, an extended state space is modeled, and a discretized model predictive control is adopted. The optimal control input is obtained by solving a quadratic programming problem.

[0009] The optimal control input is sent as a motion control command to the action executor, which then performs the corresponding action on the target body.

[0010] In some embodiments, extracting the current feature information of the target object based on the image of the target object includes:

[0011] A mapping relationship is established between the visual sensor coordinate system and the image coordinate system to determine the correspondence between the target body in the visual sensor coordinate system and the two-dimensional image features in the image coordinate system. The axis perpendicular to the image plane in the visual sensor coordinate system passes through the center of the image, and an interaction matrix is ​​introduced to characterize the influence of the visual sensor motion on the position changes of feature points on the image plane, so as to obtain the current feature information of the target body.

[0012] In some embodiments, the interaction matrix establishes a mathematical relationship between the linear and angular velocities of the visual sensor motion and feature points on the image plane.

[0013] In some embodiments, the interaction matrix projects the motion of the visual sensor in three-dimensional space onto the image plane and expresses it as the rate of change of the coordinates of feature points on the image plane over time.

[0014] In some embodiments, the modeling of the extended state space includes:

[0015] The process of feature changes with the motion of the visual sensor and unknown interference is described by dynamic equations, and the interference is regarded as an augmented variable, thereby constructing an extended state-space model.

[0016] In some embodiments, the use of discretized model predictive control to obtain the optimal control input by solving a quadratic programming problem includes:

[0017] The derivatives describing the state's change over time in the extended state-space model are approximated using the forward finite difference method. After obtaining the approximation of the derivatives, the relationship between the state and the control input in the continuous-time state equation is transformed into a discrete-time state update relationship. In this way, a linear relationship between the predicted output and the control input is constructed. By designing a reference trajectory and a cost function, and introducing a pseudo-inverse method to calculate the optimal control sequence, the model predictive control problem is transformed into a constrained quadratic programming optimization problem for solution, thus obtaining the optimal control input.

[0018] In some embodiments, the action actuator includes a robotic arm.

[0019] Secondly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the control method described in any of the first aspects.

[0020] Thirdly, embodiments of this application provide a control system, including:

[0021] Memory is used to store executable instructions for a computer;

[0022] A processor, when executing computer-executable instructions stored in the memory, implements the control method described in any of the first aspects.

[0023] The significant advantages of this invention compared to existing technologies are:

[0024] (1) By acquiring the target image from the visual sensor in real time and extracting the current feature information of the target based on the image, the present invention can comprehensively grasp the state changes of the target and greatly improve the recognition accuracy of the target position and posture.

[0025] (2) Based on the current feature information of the target body and the motion information of the visual sensor, the present invention models and expands the state space, adopts discretized model predictive control, and obtains the optimal control input by solving the quadratic programming problem, thereby expanding the unknown disturbances and uncertainties in the system into a new state variable. By measuring the system output, the system state and the expanded state variable, i.e. the unknown disturbance, are estimated in real time, which improves the anti-interference ability and robustness of the whole system, eliminates the need for an accurate model, improves control performance and makes it easy to implement.

[0026] (3) The present invention sends the optimal control input as a motion control command to the action executor, and the action executor performs the corresponding action on the target body. Since the motion control command is generated through the optimization of image processing algorithm and control strategy, the tracking speed of the action executor is improved. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the control method of the present invention in one embodiment;

[0028] Figure 2 This is a schematic diagram of the workflow of image-based visual servoing in one embodiment of the present invention;

[0029] Figure 3 This is a flowchart of the MPC process in one embodiment of the present invention;

[0030] Figure 4 This is a schematic diagram of the overall technical route of the control method of the present invention in one embodiment;

[0031] Figure 5 This is a schematic diagram of a three-dimensional coordinate system in one embodiment of the present invention;

[0032] Figure 6 This is an experimental visual tracking pixel error diagram of a robotic arm without using MPC+ESO in one embodiment of the present invention;

[0033] Figure 7 This is an experimental visual tracking pixel error diagram of a robotic arm using MPC+ESO in one embodiment of the present invention. Detailed Implementation

[0034] This disclosure relates to a control method and system, as well as a computer-readable storage medium, which can be used to control a robotic arm to improve the tracking accuracy, tracking speed, and anti-interference capability of the robotic arm.

[0035] like Figure 1 As shown in Figure 7, in some embodiments, a control method is provided, including the following steps:

[0036] Step S10: Acquire an image of the target object from a vision sensor. In some embodiments, the vision sensor includes a camera, which can capture images via a camera. High-resolution, high-frame-rate industrial cameras are preferred as vision sensors to ensure real-time acquisition of clear images of the target object. The acquired images are preprocessed, including noise reduction, contrast enhancement, and color correction, to improve image quality and provide a good foundation for subsequent feature extraction and target recognition. For example, a high-resolution RGB-D camera can be used for image information extraction. RGB-D cameras have many advantages: firstly, they can capture rich color information, presenting vivid and realistic images, which helps to accurately identify object colors and plays a significant role in image recognition, quality inspection, and other fields; secondly, they typically have high resolution, capturing images with rich details, improving the accuracy of object recognition and analysis, and can be magnified and cropped without losing too much detail; thirdly, they are cost-effective, technologically mature, have many market options and relatively low prices, and low maintenance costs.

[0037] Step S20: Extract the current feature information of the target object based on its image;

[0038] like Figure 2 As shown, specifically, the center of the visual sensor coordinate system C is set. The coordinate axes are ,in, The three axis unit vectors (X-axis, Y-axis, Z-axis) of the vision sensor coordinate system are represented by T, which indicates transpose; the coordinates of the image coordinate system L are... ,in Represents the center of the image; two-dimensional pixel coordinates on the image plane; represents the position of a specific image feature point. This represents the horizontal and vertical pixel coordinates in the image. Note the coordinates in the visual sensor coordinate system. Axis perpendicular to A mapping relationship is established between the visual sensor coordinate system and the image coordinate system to determine the correspondence between the target object in the visual sensor coordinate system and the two-dimensional image features in the image coordinate system. This is done for a set of three-dimensional points fixed in the visual sensor coordinate system C. , Its corresponding two-dimensional image features are It represents a three-dimensional point. The corresponding coordinates of the feature points in the two-dimensional image. The effect of visual sensor motion on feature coordinates on the image plane is determined by... Modeling, This represents the velocity of a point in the image plane, where = , It is the focal length of the vision sensor. Represents the spatial point to the vision sensor Axial distance; Indicates the first The horizontal and vertical coordinates of each image feature.

[0039] An interaction matrix is ​​introduced to characterize the effect of visual sensor motion on the position changes of feature points on the image plane, thereby obtaining the current feature information of the target object. The interaction matrix is ​​as follows:

[0040]

[0041] The interaction matrix establishes the mathematical relationship between the linear and angular velocities of the visual sensor motion and the feature points on the image plane. It also projects the motion of the visual sensor in three-dimensional space onto the image plane and expresses it as the rate of change of the coordinates of the feature points on the image plane over time. Used to describe the response of image points to motion of a vision sensor.

[0042] Step S30: Based on the current feature information of the target body and the motion information of the visual sensor, model the extended state space, adopt discretized model predictive control, and obtain the optimal control input by solving the quadratic programming problem;

[0043] Specifically, Represents the linear speed of the vision sensor and angular velocity , The subscripts correspond to the linear velocity and angular velocity axes, respectively. It is a set of velocities. Simultaneously, the overall image feature vector is defined. The concatenated vector of coordinates of all image feature points. Indicates the first The pixel coordinates of each image feature point. Therefore, the dynamics of the feature coordinates are given by the following formula:

[0044]

[0045] Represents the time derivative (velocity) of all image feature points; It is the overall interaction matrix, a combination of all points' interaction matrices; Indicates the speed of motion of the vision sensor;

[0046] This represents unknown interference related to the speed of the tracked object. Given a column vector with each term equal to 1, a dynamic equation is used to describe the process of feature changes as the visual sensor moves and unknown interferences occur. Interferences are treated as augmented variables, thus constructing an extended state-space model.

[0047] The model is transformed into the following form:

[0048]

[0049] The controller takes the form of: The form of interference is and It is the measurement output. Indicates the speed of image features; This refers to the control input, specifically the feature changes caused by the movement of the visual sensor, determined by the visual servo controller. This represents an unknown interference term, an interference vector related to the target's motion; This indicates the output measurement, specifically the location of the observed image features. Indicates the state of image features.

[0050] Treating the disturbance d(t) as an augmented variable, the new state is defined as follows: The system can then be expanded to:

[0051]

[0052] in ,

[0053]

[0054] This indicates how the system state evolves over time; Represents the state transition matrix; Represents the control input matrix; Represents the input matrix of the interference derivative; Indicates the output matrix; This indicates the system augmentation state, which includes the image state and interference terms; This indicates the controller output, which is a system control quantity (such as motion from a vision sensor). The derivative of the disturbance represents the rate of change of the disturbance, taking into account its dynamic nature; This represents the measured value, the actual observed value of the image feature points.

[0055] like Figure 3 As shown, (1) Construct the MPC controller

[0056] The model performs model predictions:

[0057] First step prediction:

[0058]

[0059] in, Represents any one of these steps. , For the sake of convenience, the prediction interval is omitted. The system state vector at time t (including image features and augmented interference, etc.); Indicates time The system state vector; The control input at time t is derived from the output of the model predictive controller. Indicates time The system output, i.e., the system measurement value; This represents the state observation matrix, used to extract measurement outputs from the state vector.

[0060] , Yes The discretized form of matrix B, which uses forward finite difference in this section, is as follows:

[0061] in, Let A represent the state transition matrix in discrete time, which is used to describe the state evolution relationship in the discrete time domain; B represents the control input matrix in discrete time, describing the influence of the control input on the state; The sampling interval is not necessarily related to the prediction interval. In this section, the prediction interval and the sampling interval are set to be the same, i.e. .

[0062] Second step prediction:

[0063]

[0064] Indicates time The predicted value of the system state vector; Representation matrix The square of; The system state vector at time t represents the system state vector. Indicates time The predicted value of the control input; Control input indicating time 𝑘; Indicates time The system output predicted value; Indicates time The predicted value of the system state vector;

[0065] No. Step prediction:

[0066]

[0067] Indicates the prediction step size; Indicates the first The state prediction value of the step; express of The exponentiation indicates state propagation. The cumulative impact of each step; Indicates control input go through The effect matrix after the state propagation step; Indicates at time Predicted time The system output;

[0068] In summary:

[0069]

[0070] in

[0071]

[0072] Reference signal The design is as follows:

[0073]

[0074] Design cost function :

[0075]

[0076] Indicates the future The output predictions for each step are in column vector form; Indicates the future The expected reference output of the step, controlling the target; This represents the future control input sequence and optimizes the variables. The prediction matrix represents the influence of the system's initial state on the future output Y. The control input mapping matrix represents the influence of the control input U on the future output Y. This represents the cost function, which measures the trade-off between prediction output deviation and control input deviation; Q is the weight of the prediction tracking error, and R is the weight of the control input. To obtain the required control sequence, the next step is to obtain the desired control sequence. :

[0077]

[0078] in For the required reference state and It is the pseudo-inverse of matrix M. For this MPC structure, the required system constraints are applied and transformed into a set of constraints. The constraints can be selected according to the actual situation. In this experiment, the constraints are mainly for safety considerations, setting the action execution body, that is, the maximum running speed of the robotic arm.

[0079] The cost equation is rewritten as:

[0080]

[0081] in , is the coefficient matrix of the quadratic term; A constant term that is independent of the optimization variable; The first-order term vector (gradient);

[0082] The results are summarized into a quadratic programming problem:

[0083]

[0084] The optimal solution U obtained is:

[0085]

[0086] Next, we will optimize online in real time. In this system, the optimal control input sequence is obtained by solving a quadratic programming problem. Then, after each sampling, the quadratic programming problem is repeatedly solved, and the new sequence is used... It is applied to the system.

[0087] (2) Constructing a discrete-time disturbance observer

[0088] Another method for mitigating interference is introduced to estimate the disturbance, namely, constructing a disturbance observer (DO). Since this system is a digital sampling system, a discrete-time disturbance observer can be designed to estimate the lumped disturbance of the system in real time. First, we give the definitions of the two components of lumped interference. The lumped disturbance must satisfy the following assumptions.

[0089] Assuming interference components The first-order difference is bounded and satisfies the following conditions:

[0090]

[0091] in, , representing the first-order difference of the disturbance component. It is a positive constant, representing the maximum rate of change (upper bound) of the disturbance component. Ts represents the time interval for continuous sampling in the discrete-time system.

[0092] In the field of engineering control, the simple boundary pre-setting method has theoretical limitations in addressing the coupling effect of complex disturbances and their infinitesimal components on the system state. Nevertheless, disturbance observation and cancellation strategies are still widely used in engineering practice, fundamentally because closed-loop control systems possess inherent stability protection mechanisms. The basic control architecture typically ensures that the system state remains in a stable operating domain. Under this premise, secondary uncertainties caused by the evolution of the system state often exhibit weak disturbance characteristics, with energy levels insufficient to breach the system's stability boundaries. This inherent resilience of the physical system provides an engineering feasibility basis for disturbance observer design. By establishing a dynamic disturbance model to progressively compensate for unmodeled factors, both overly idealistic assumptions in theoretical analysis and effective suppression of intrinsic uncertainties such as parameter perturbations in actual operation are avoided. To estimate lumped disturbances... Construct a discrete-time disturbance observer.

[0093]

[0094] in, The state vector of the disturbance observer is a diagonal matrix. This represents the gain matrix of the interference observer. This represents an estimate of the lumped disturbance. This represents the control input matrix, a mapping matrix between input channels and state variables; The control input vector at time t; Represents the interference input matrix; Represents system state variables; It represents the recursive expression of the state variables of the disturbance observer, which is used to dynamically track and estimate the changing trend of the system under unknown disturbances.

[0095] The estimation error of lumped disturbance is defined as: , Indicates the first The estimation error of each disturbance component; combining the system model and the disturbance observer, the error equation for disturbance estimation can be obtained:

[0096]

[0097] in, The first-order difference represents the lumped disturbance.

[0098] Select the interference observer gain for Then the interference estimate It always converges to the true value. Within its bounded neighborhood; It is represented as a diagonal matrix, with the convergence factor corresponding to each dimension; It is represented as the inverse of the interference input matrix; where Meet the conditions Indicates the first The convergence coefficient of the observer in each dimension must be less than 1 to ensure stability;

[0099] Increase the gain of the interference observer Substituting these equations into the disturbance estimation error equation, we can obtain the dynamic equations for the disturbance estimation error components:

[0100]

[0101] The estimated error components can be obtained by calculating the iterative method:

[0102]

[0103] This represents the error at the initial moment; This represents the historical contribution factor of the error, which decays over time. The convergence factor representing the perturbation estimation error In the The exponent of a time step; Indicates the first The disturbance component in the first... First-order difference at each time step;

[0104] Based on the assumptions, the range of the lumped disturbance estimation error is:

[0105]

[0106] Based on the above proof, the estimation error of the discrete-time disturbance observer constructed above will converge to a bounded range.

[0107] As mentioned above, the derivative describing the state change over time in the extended state-space model is approximated using the forward finite difference method. After obtaining the approximation of the derivative, the relationship between the state and the control input in the continuous-time state equation is transformed into a state update relationship in discrete time. In this way, a linear relationship between the predicted output and the control input is constructed. By designing a reference trajectory and a cost function, and introducing a pseudo-inverse method to calculate the optimal control sequence, the model predictive control problem is transformed into a constrained quadratic programming optimization problem for solution, thus obtaining the optimal control input.

[0108] In step S40, the optimal control input is sent as a motion control command to the motion actuator, which then performs the corresponding action on the target body. In some embodiments, the motion actuator includes a robotic arm.

[0109] This invention acquires target images in real time from a visual sensor and extracts the target's current feature information based on the target image, which can comprehensively grasp the state changes of the target and greatly improve the accuracy of target position and posture recognition.

[0110] This invention models and expands the state space based on the current feature information of the target and the motion information of the visual sensor. It adopts discretized model predictive control and obtains the optimal control input by solving a quadratic programming problem. This expands the unknown disturbances and uncertainties in the system into a new state variable. By measuring the system output, the system state and the expanded state variable, i.e. the unknown disturbance, are estimated in real time. This improves the anti-interference ability and robustness of the entire system, eliminates the need for an exact model, improves control performance, and is easy to implement.

[0111] This invention sends the optimal control input as a motion control command to the action executor, which then performs the corresponding action on the target body. Since the motion control command is generated through the optimization of image processing algorithms and control strategies, the tracking speed of the action executor is improved.

[0112] This invention solves the technical problem that most current control systems with industrial robotic arms can only perform specified actions and lack the ability to track targets, which makes it difficult for the industrial robotic arm of the control system to complete the target task if the mechanical parts are not in the specified position.

[0113] This application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the control method mentioned above.

[0114] This application provides a control system, including: a memory for storing computer-executable instructions; and a processor for executing the computer-executable instructions stored in the memory to implement the control method mentioned above.

[0115] In some embodiments, the control system can be integrated onto the robot, specifically the Universal Robots UR5, with an RGB-D end-effector camera. To reduce image processing complexity, the dynamically tracked target is represented by a square box manufactured by APRILTAG, driven by a lead screw motor. The software architecture is based on the robot operating system ROS. All programs run on a Lenovo laptop with an i7-10750H CPU and 32GB of RAM.

[0116] The camera sampling rate is 0.12s, and the observer gain is... MPC parameters are , With R=1.5, the maximum safe operating speed of the robotic arm only needs to be greater than the speed of the target being tracked. The target speed is designed to be 0.12 m / s, and the constraint is designed to be 0.3 m / s. The experiment mainly compares the dynamic tracking situation without MPC+ESO and with MPC+ESO. In the traditional system without the MPC+ESO composite control architecture, the target trajectory tracking exhibits a significant lag. The root cause is the failure to effectively handle the dynamic coupling characteristics of the system and the coupling effect of external disturbances. Experimental data shows that when the target is moving, the open-loop prediction mechanism cannot correct the pose deviation caused by model mismatch in time, resulting in a cumulative divergence trend in the tracking error. Especially when there are sudden external disturbances, the adjustment bandwidth of the traditional PID controller is insufficient, and the response delay exceeds the system stability margin, causing significant overshoot oscillations.

[0117] In comparison, the MPC+ESO collaborative control architecture adopted in this application exhibits significant advantages. The ESO-based anti-interference strategy developed in this invention effectively improves the stability and robustness of the system. This invention employs an Extended State Observer (ESO) to construct a global disturbance estimation model, unifying external environmental disturbances, robotic arm joint friction, and visual measurement noise into extended state variables for real-time observation. By designing a gain adjustment mechanism, the observer bandwidth can be adjusted according to the disturbance intensity, ensuring both rapid tracking capability for high-frequency disturbances and avoiding excessive amplification of observation noise. At the control strategy level, a synergistic mechanism of disturbance feedforward compensation and feedback suppression is proposed: the disturbance estimate output by the ESO directly cancels the impact of disturbances on the system through the feedforward channel, while simultaneously forming a composite anti-interference control architecture combined with a visual servo feedback loop. Specifically targeting dynamic occlusion interference experienced by the visual sensor, an interference recognition algorithm based on motion continuity is developed to ensure that the ESO can maintain effective observation even when some features are lost.

[0118] The MPC module employs a rolling time-domain optimization strategy to solve a multi-objective optimization problem, including robotic arm dynamics constraints, in each control cycle, achieving predictive tracking of the desired trajectory. The ESO (Extended State Observer) constructs a dynamic model for disturbance estimation, treating unmodeled system dynamics and external disturbances as extended state variables for real-time observation and compensation. Notably, this architecture ensures the global asymptotic convergence of the closed-loop system through Lyapunov stability analysis. In high-speed motion experiments, when the target acceleration reaches the robotic arm's dynamic limit, the MPC's constraint handling capability successfully prevents actuator saturation, while the ESO's disturbance estimation error remains within a preset tolerance range. Furthermore, through an online parameter adaptation mechanism, the system maintains tracking accuracy even under disturbed conditions, verifying the strong robustness of the control strategy. Specific experimental figures are shown in Part III. Figure 6 and Figure 7It is quite obvious that the effect is outstanding.

[0119] The technical solutions involved in this invention include not only the technical solutions disclosed above, but also technical solutions composed of equivalent substitutions of the technical features involved in this invention. Matters not covered in this invention are common knowledge to those skilled in the art.

Claims

1. A control method, characterized in that, include: Acquire images of the target object using a visual sensor; Based on the image of the target object, extract the current feature information of the target object; Based on the current feature information of the target and the motion information of the vision sensor, an extended state space is modeled, and a discretized model predictive control is adopted. The optimal control input is obtained by solving a quadratic programming problem. The optimal control input is sent as a motion control command to the action executor, which then performs the corresponding action on the target body.

2. The control method according to claim 1, characterized in that, The step of extracting the current feature information of the target object based on the image of the target object includes: A mapping relationship is established between the visual sensor coordinate system and the image coordinate system to determine the correspondence between the target body in the visual sensor coordinate system and the two-dimensional image features in the image coordinate system. The axis perpendicular to the image plane in the visual sensor coordinate system passes through the center of the image, and an interaction matrix is ​​introduced to characterize the influence of the visual sensor motion on the position changes of feature points on the image plane, so as to obtain the current feature information of the target body.

3. The control method according to claim 1, characterized in that: The interaction matrix establishes the mathematical relationship between the linear and angular velocities of the visual sensor motion and the feature points on the image plane.

4. The control system according to claim 1, characterized in that: The interaction matrix projects the motion of the visual sensor in three-dimensional space onto the image plane and expresses it as the rate of change of the coordinates of feature points on the image plane over time.

5. The control method according to claim 1, characterized in that, The modeling extended state space includes: The process of feature changes with the motion of the visual sensor and unknown interference is described by dynamic equations, and the interference is regarded as an augmented variable, thereby constructing an extended state-space model.

6. The control method according to claim 5, characterized in that, The discretized model predictive control, which obtains the optimal control input by solving a quadratic programming problem, includes: The derivatives describing the state's change over time in the extended state-space model are approximated using the forward finite difference method. After obtaining the approximation of the derivatives, the relationship between the state and the control input in the continuous-time state equation is transformed into a discrete-time state update relationship. In this way, a linear relationship between the predicted output and the control input is constructed. By designing a reference trajectory and a cost function, and introducing a pseudo-inverse method to calculate the optimal control sequence, the model predictive control problem is transformed into a constrained quadratic programming optimization problem for solution, thus obtaining the optimal control input.

7. The control system according to claim 1, characterized in that: The actuator includes a robotic arm.

8. A computer-readable storage medium storing computer-executable instructions, characterized in that: When the computer-executable instructions are executed by the processor, they implement the control method according to any one of claims 1 to 7.

9. A control system, characterized in that, include: Memory is used to store executable instructions for a computer; A processor, when executing computer-executable instructions stored in the memory, implements the control method according to any one of claims 1 to 8.

Images