Model training, mechanical equipment control methods, devices, processing equipment and media

Abstract

The present application provides a model training, mechanical equipment control method, device, processing equipment and medium, relating to the field of intelligent control technology. The method includes: obtaining sample motion data of a first mechanical arm at multiple moments in the mechanical equipment to be controlled, sample motion states of a second mechanical arm at multiple moments, and sample control data of the first mechanical arm at multiple moments; the first mechanical arm and the second mechanical arm are mechanical arms with a coupling relationship in the mechanical equipment to be controlled; based on the sample motion data of the first mechanical arm at multiple moments, the sample motion states of the second mechanical arm at multiple moments, and the sample control data of the first mechanical arm at multiple moments, an initial control model is trained to obtain a control model for the first mechanical arm. This makes the control model for the first mechanical arm obtained by training more reliable, and the control parameters for the first mechanical arm output by the control model more accurate.

Description

Technical Field

[0001] The present application relates to the field of intelligent control technology, and more specifically, to a model training, mechanical equipment control method, device, processing equipment and medium. Background Art

[0002] With the development of science and technology, there are more and more various mechanical equipment, such as excavators, forklifts, pharmaceutical machinery, plastic processing machinery, etc. Automatic control of the robotic arms of mechanical equipment can ensure the automated operation of mechanical equipment. The control of mechanical equipment has also become a research hotspot.

[0003] In the related art, for each robotic arm of a mechanical device, a neural network model corresponding to the robotic arm is independently trained based on sample data of the robotic arm, and the model of the robotic arm can subsequently be used to output the control parameters of the robotic arm.

[0004] However, in the related art, the neural network model corresponding to the robotic arm is independently trained, resulting in inaccurate control parameters output by the trained neural network model. Summary of the Invention

[0005] The purpose of this application is to provide a model training, mechanical equipment control method, device, processing equipment and medium to address the above-mentioned technical problems existing in the related technology.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of the present application are as follows:

[0007] In a first aspect, an embodiment of the present application provides a model training method, comprising:

[0008] Acquiring sample motion data of a first robotic arm in a mechanical device to be controlled at multiple moments, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments; the first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the sample motion data of the first robotic arm at each moment includes: the sample motion state of the first robotic arm at the multiple moments and a target sample motion state at a preset future moment after the multiple moments;

[0009] The initial control model is trained based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0010] In a second aspect, an embodiment of the present application further provides a mechanical equipment control method, comprising:

[0011] Obtaining a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm are robotic arms in a coupled relationship in the to-be-controlled mechanical device; and the second moment is a preset future moment after the first moment;

[0012] Using a control model for the first robotic arm, according to a motion state of the first robotic arm at the first moment, a target motion state of the first robotic arm at the second moment, and a motion state of the second robotic arm at the first moment, a control parameter of the first robotic arm at the first moment is obtained;

[0013] According to the control parameters of the first robotic arm at the first moment, the first robotic arm is controlled so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the method described in any one of claims 1 to 7 above.

[0014] In a third aspect, an embodiment of the present application further provides a model training device, comprising:

[0015] An acquisition module is configured to acquire sample motion data of a first robotic arm in a mechanical device to be controlled at multiple moments, sample motion states of a second robotic arm at said multiple moments, and sample control data of the first robotic arm at said multiple moments; the first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the sample motion data of the first robotic arm at each moment includes: the sample motion state of the first robotic arm at said each moment and a target sample motion state at a preset future moment after said each moment;

[0016] A training module is used to train an initial control model based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments, so as to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0017] In a fourth aspect, an embodiment of the present application further provides a mechanical equipment control device, comprising:

[0018] an acquisition module, configured to acquire a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm being robotic arms in a coupled relationship in the to-be-controlled mechanical device; the second moment being a preset future moment after the first moment; and employing a control model for the first robotic arm to obtain control parameters of the first robotic arm at the first moment based on the motion state of the first robotic arm at the first moment, the target motion state of the first robotic arm at the second moment, and the motion state of the second robotic arm at the first moment;

[0019] A control module is used to control the first robotic arm according to the control parameters of the first robotic arm at the first moment, so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the method described in any one of the first aspects above.

[0020] In the fifth aspect, an embodiment of the present application also provides a processing device, including: a memory and a processor, wherein the memory stores a computer program executable by the processor, and when the processor executes the computer program, it implements the model training method described in any one of the first aspects above, or the mechanical equipment control method described in the second aspect above.

[0021] In the sixth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is read and executed, it implements the model training method described in any one of the first aspects above, or the mechanical equipment control method described in the second aspect above.

[0022] The beneficial effects of the present application are as follows: the embodiment of the present application provides a model training method, comprising: obtaining sample motion data of a first manipulator at multiple moments in a mechanical device to be controlled, sample motion states of a second manipulator at multiple moments, and sample control data of the first manipulator at multiple moments; the first manipulator and the second manipulator are manipulators in a coupled relationship in the mechanical device to be controlled; the sample motion data of the first manipulator at each moment includes: the sample motion state of the first manipulator at each moment and the target sample motion state of a preset future moment after each moment; based on the sample motion data of the first manipulator at multiple moments, the sample motion states of the second manipulator at multiple moments, and the sample control data of the first manipulator at multiple moments, an initial control model is trained to obtain a control model of the first manipulator, the control model of the first manipulator is used to output control parameters for the first manipulator. When performing model training, not only the sample motion data and sample control data of the first manipulator are used, but also the sample motion data of the second manipulator that has an influence on the first manipulator is used, and the mutual influence and coupling relationship between the first manipulator and the second manipulator are taken into account, so that the control model for the first manipulator obtained by training is more reliable and the control parameters for the first manipulator output by the control model are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 A schematic diagram of a model training method provided in this embodiment of the application Figure 1 ;

[0025] Figure 2 A schematic diagram of a model training method provided in this embodiment of the application Figure 2 ;

[0026] Figure 3 A schematic diagram of a model training method provided in this embodiment of the application Figure 3 ;

[0027] Figure 4 A schematic flow chart of a mechanical equipment control method provided in an embodiment of the present application;

[0028] Figure 5 A schematic diagram of the structure of a model training device provided in an embodiment of the present application;

[0029] Figure 6A schematic structural diagram of a mechanical equipment control device provided in an embodiment of the present application;

[0030] Figure 7 A schematic diagram of the structure of a processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0032] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for protection, but merely represents selected embodiments of the present application. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments in the present application without creative work are within the scope of protection of the present application.

[0033] In the description of this application, it should be noted that if the terms "upper", "lower", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the accompanying drawings, or is the orientation or position relationship in which the product of the application is usually placed when in use. It is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on this application.

[0034] In addition, the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] It should be noted that, in the absence of conflict, the features in the embodiments of this application can be combined with each other.

[0036] An embodiment of the present application provides a model training method, which is applied to a processing device, which can be a terminal device, a server or a controller in a mechanical device. If the electronic device is a terminal device, the terminal device can be at least one of the following: a desktop computer, a laptop computer, a tablet computer, a smart phone, etc. The mechanical device can be an excavator, an industrial robot or other mechanical device with a robotic arm. The embodiment of the present application does not impose specific restrictions on this.

[0037] The following explains a model training method provided in an embodiment of the present application.

[0038] Figure 1 A schematic diagram of a model training method provided in this embodiment of the application Figure 1 ,like Figure 1 As shown, the method includes:

[0039] S101 : Acquire sample motion data of a first robotic arm at multiple moments, sample motion states of a second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments in a robotic device to be controlled.

[0040] The first robotic arm and the second robotic arm are robotic arms having a coupling relationship in the mechanical device to be controlled. For example, the first robotic arm can be a large robotic arm and the second robotic arm can be a small robotic arm, or the first robotic arm can be a small robotic arm and the second robotic arm can be a large robotic arm.

[0041] In addition, the sample motion data of the first robotic arm at each moment includes: the sample motion state of the first robotic arm at each moment and the target sample motion state at a preset future moment after each moment.

[0042] In an embodiment of the present application, multiple moments refer to multiple moments with a sequence of moments. The target sample motion state of a preset future moment after each moment can be determined from the sample motion state of the first robotic arm at multiple moments.

[0043] For example, multiple moments can be represented as: t1, t2, t3 and t4. For moment t1, the preset future moment can be t3; for moment t2, the preset future moment can be t4; then the sample motion data of the first robotic arm at moment t1 can include: the sample motion state of the first robotic arm at moment t1, and the target sample motion state of the first robotic arm at moment t3. Similarly, the sample motion data at moment t2 can include: the sample motion state of the first robotic arm at moment t2, and the target sample motion state of the first robotic arm at moment t4. And so on. The sample motion data of the first robotic arm at multiple moments can be obtained, which will not be repeated here.

[0044] It should be noted that, at the same moment, the sample motion state of the first robotic arm, the sample motion state of the second robotic arm, and the sample control data of the first robotic arm are all in a corresponding relationship.

[0045] S102 . Train an initial control model based on sample motion data of the first robotic arm at multiple moments, sample motion states of the second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments to obtain a control model of the first robotic arm.

[0046] The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0047] In some embodiments, sample motion data of the first robotic arm at multiple moments, sample motion states of the second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments can be input into an initial control model, and the initial control model can be trained until the training process meets preset conditions. The training is terminated and the model parameters are saved to obtain a control model for the first robotic arm.

[0048] It is worth noting that the initial control model can be a neural network model.

[0049] In an embodiment of the present application, after the control model of the first robotic arm is obtained through training, the control model of the first robotic arm can be used to output control parameters for the first robotic arm, and accurate control of the first robotic arm can be achieved based on the control parameters of the first robotic arm.

[0050] In summary, an embodiment of the present application provides a model training method, comprising: obtaining sample motion data of a first manipulator at multiple moments in a mechanical device to be controlled, sample motion states of a second manipulator at multiple moments, and sample control data of the first manipulator at multiple moments; the first manipulator and the second manipulator are manipulators having a coupling relationship in the mechanical device to be controlled; the sample motion data of the first manipulator at each moment includes: the sample motion state of the first manipulator at each moment and the target sample motion state of a preset future moment after each moment; based on the sample motion data of the first manipulator at multiple moments, the sample motion states of the second manipulator at multiple moments, and the sample control data of the first manipulator at multiple moments, an initial control model is trained to obtain a control model of the first manipulator, the control model of the first manipulator is used to output control parameters for the first manipulator. When performing model training, not only the sample motion data and sample control data of the first manipulator are used, but also the sample motion data of the second manipulator that has an impact on the first manipulator is used, and the mutual influence and coupling relationship between the first manipulator and the second manipulator are taken into account, so that the control model for the first manipulator obtained by training is more reliable and the control parameters for the first manipulator output by the control model are more accurate.

[0051] Moreover, the motion state of the second robotic arm can be integrated into the control model of the first robotic arm, and the coupling relationship between the first robotic arm and the second robotic arm can be learned during the model training process.

[0052] Optional, Figure 2 A schematic diagram of a model training method provided in this embodiment of the application Figure 2 ,like Figure 2 As shown, the process of obtaining sample motion data of the first robotic arm at multiple moments, sample motion states of the second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments in the mechanical device to be controlled in S101 may include:

[0053] S201: Obtain the motion trajectory of the mechanical equipment to be controlled.

[0054] The motion trajectory includes: a plurality of trajectory points in a sequential order, and the plurality of trajectory points have corresponding sample motion states of the first robotic arm, sample control data of the first robotic arm, and sample motion states of the second robotic arm.

[0055] In some embodiments, the motion trajectory of the mechanical equipment to be controlled can be collected according to a preset sampling frequency. The sample data of the trajectory points in the motion trajectory refers to the sample data collected at the corresponding sampling time. Among them, due to the influence of the preset sampling frequency, different trajectory points correspond to different sampling times.

[0056] It should be noted that there may be multiple motion trajectories, each of which includes multiple trajectory points in a sequential order. For example, the number of motion trajectories may be 95.

[0057] For example, multiple trajectory points in a motion trajectory can be represented as: a, b, c, d. The sampling time corresponding to each trajectory point is different. The sampling time corresponding to trajectory point a can be t1, the sampling time corresponding to trajectory point b can be t2, the sampling time corresponding to trajectory point c can be t3, and the sampling time corresponding to trajectory point d can be t4.

[0058] S202 : Determine sample motion data of the first robotic arm at multiple moments according to the sample motion states of the first robotic arm corresponding to the multiple trajectory points.

[0059] Among them, since the sampling time corresponding to each trajectory point is different, the sample motion state of the first robot arm corresponding to multiple trajectory points can be considered as the sample motion state of the first robot arm at each moment. Similarly, the target sample motion state of the future moment preset after each moment can be determined from the sample motion state of the first robot arm at each moment.

[0060] In an embodiment of the present application, the sample motion state of the first robotic arm corresponding to the first trajectory point among multiple trajectory points can be regarded as the sample motion state of the first robotic arm at the first moment among multiple moments, and the preset number of second trajectory points after the first trajectory point among the multiple trajectory points can be regarded as the sample motion state of the first robotic arm at a preset future moment after the first moment. In this way, the sample motion data of the first robotic arm at the first moment can be obtained.

[0061] S203 : Using the sample control data of the first robotic arm corresponding to the multiple trajectory points as the sample control data of the first robotic arm at multiple moments.

[0062] S204 : Using the sample motion states of the second manipulator corresponding to the multiple trajectory points as the sample motion states of the second manipulator at multiple moments.

[0063] Similarly, since each trajectory point corresponds to a different sampling moment, executing the above-mentioned processes S203 and S204 can obtain the sample control data of the first robot arm at multiple moments and the sample motion states of the second robot arm at multiple moments.

[0064] Optionally, the process of obtaining sample motion data of the first robotic arm at multiple moments, sample motion states of the second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments in the mechanical device to be controlled in S101 may include:

[0065] The position and angular velocity of the first robotic arm at each moment are obtained, and after each moment, the target sample motion state of the future moment is preset, the position and angular velocity of the second robotic arm at multiple moments, and the sample pulse width modulation data of the first robotic arm at multiple moments are obtained.

[0066] The sample pulse width modulation data of the first robotic arm at multiple moments may also be referred to as sample PWM (Pulse Width Modulation) data of the first robotic arm at multiple moments.

[0067] Optionally, the first robotic arm presets a target sample motion state at a future moment after each moment, including: a target position and / or a target angular velocity.

[0068] It is worth noting that the target sample motion state includes only the target position, or only the target angular velocity, or both the target position and the target angular velocity, and the embodiments of the present application do not impose specific restrictions on this.

[0069] Optionally, the time interval between two adjacent moments in the multiple moments is the sampling time interval between two adjacent trajectory points in the motion trajectory;

[0070] The time interval between two adjacent moments in the multiple moments is smaller than the time interval between each moment and the preset future moment.

[0071] The time interval is determined by the sampling frequency. For example, the sampling frequency can be 100 Hz, and a trajectory point is sampled every 0.01 seconds. The moment after each moment after the preset future time is used as the preset future moment. That is, the time interval between each time and the preset future moment is the preset future time. For example, the preset future time can be 1 second.

[0072] In the embodiment of the present application, the reason why the target sample motion state preset at the future moment after each moment is used as the training object is to take into account the delay problem of mechanical equipment control. After the instruction is issued, the mechanical equipment needs a certain amount of time to respond.

[0073] In some embodiments, the initial control model can be trained based on the position and angular velocity of the first robotic arm at each moment, the target position and / or target angular velocity at future moments preset after each moment, the position and angular velocity of the second robotic arm at multiple moments, and the sample pulse width modulation data of the first robotic arm at multiple moments to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0074] The control parameter for the first robotic arm may be a pulse width modulation parameter for the first robotic arm.

[0075] Optional, Figure 3 A schematic diagram of a model training method provided in this embodiment of the application Figure 3 ,like Figure 3 As shown, the process of training the initial control model in the above S102 based on the sample motion data of the first manipulator at multiple moments, the sample motion states of the second manipulator at multiple moments, and the sample control data of the first manipulator at multiple moments to obtain the control model of the first manipulator may include:

[0076] S301. Input the sample motion states of the first robotic arm at multiple moments, the target sample motion states of the future moments preset after each moment, the sample motion states of the second robotic arm at multiple moments, and the sample control data of the first robotic arm at multiple moments into the initial control model to obtain the sample prediction control data of multiple moments output by the initial control model.

[0077] The input dimension of the initial control model may be a preset dimension, for example, 104 dimensions.

[0078] In an embodiment of the present application, sample data of multiple moments can be input in batches, wherein the sample data of part of the moments are output each time, or the sample data of multiple moments can be output in full. This embodiment of the present application does not impose any specific restrictions on this.

[0079] S302. Update the model parameters of the initial control model based on the sample prediction control data at multiple moments, the sample control data of the first robotic arm at multiple moments, and the preset loss function until the number of iterations is greater than or equal to the preset number of iterations, and obtain the control model for the first robotic arm.

[0080] Among them, the sample control data of the first robotic arm at multiple moments can be considered as labels in the model training process.

[0081] In some embodiments, a preset loss function is used to calculate the root mean square error based on sample prediction control data at multiple moments and sample control data of the first robotic arm at multiple moments, and the model parameters of the initial control module are updated based on the root mean square error. The iteration is performed, and when the number of iterations is greater than or equal to the preset number of iterations, the model parameters at this time are saved to obtain a control model for the first robotic arm.

[0082] Optionally, the initial control model includes: a plurality of fully connected layers connected in sequence, wherein the number of neurons in the plurality of fully connected layers decreases in sequence. The initial control model also includes: an input layer and an output layer, wherein the input layer is connected to the first fully connected layer of the plurality of fully connected layers, and the last fully connected layer of the plurality of fully connected layers is connected to the output layer;

[0083] Among them, the number of fully connected layers can be 4, which not only ensures that the extracted features are richer, but also avoids the risk of overfitting. Of course, this is just an example and can be set according to actual needs. This application does not impose specific restrictions on this.

[0084] In addition, the number of neurons in multiple fully connected layers can be reduced proportionally. For example, the number of neurons in four fully connected layers are 256, 128, 64, and 32 respectively.

[0085] In some embodiments, sample motion data of the first robotic arm at multiple moments, sample motion states of the second robotic arm at multiple moments, and sample control data of the first robotic arm at multiple moments are input into the input layer; the first fully connected layer among multiple fully connected layers is used to perform feature extraction processing on the sample motion data of the first robotic arm at multiple moments and the sample motion states of the second robotic arm at multiple moments, and the first fully connected layer outputs the first sample features of the first robotic arm at multiple moments, and the first sample features of the second robotic arm at multiple moments; the next fully connected layer is used to perform feature extraction processing on the first sample features of the first robotic arm at multiple moments and the first sample features of the second robotic arm at multiple moments to obtain the second sample features of the first robotic arm at multiple moments, and the second sample features of the second robotic arm at multiple moments, until the last fully connected layer outputs the target sample features of the first robotic arm at multiple moments, and the target sample features of the second robotic arm at multiple moments.

[0086] Correspondingly, the output layer is used to determine the sample prediction control data of the first robotic arm at multiple moments based on the target sample features of the first robotic arm at multiple moments and the target sample features of the second robotic arm at multiple moments; then, based on the sample prediction control data of the first robotic arm at multiple moments, the sample control data of the first robotic arm at multiple moments, and the preset loss function, the model parameters of the multiple fully connected layers are updated until the number of iterations is greater than or equal to the preset number of iterations, thereby obtaining a control model for the first robotic arm.

[0087] It should be noted that the sample motion data of the first robotic arm at multiple moments and the sample motion states of the second robotic arm at multiple moments are subjected to feature extraction processing in sequence through multiple fully connected layers, so that the target feature features of the first robotic arm at multiple moments and the target features of the second robotic arm at multiple moments that are finally extracted are richer and more detailed.

[0088] Optionally, the mechanical equipment can be a hydraulically driven excavator. Most existing excavators use hydraulic transmission, and it is difficult to establish an accurate model for a hydraulically driven excavator. In the excavator's loading task, the excavator needs to be able to accurately track a given target trajectory. The excavator needs to accurately track the target trajectory. Since the excavator is hydraulically driven, it is difficult to model the excavator as a whole and construct a dynamic model and a kinematic model. In related technologies, model training can model a single robotic arm, but in fact, the control of multiple robotic arms of an excavator needs to take into account the coupling relationship between the robotic arms, such as the mutual influence between the upper arm and the lower arm. When tracking the target trajectory, the coupling relationship between the excavator's bucket and cabin and the robotic arm is relatively small, so this application mainly considers the coupling relationship between the robotic arms.

[0089] An embodiment of the present application also provides a mechanical equipment control method, which is applied to a processing device. The processing device can be a terminal device, a server or a controller in a mechanical device. If the electronic device is a terminal device, the terminal device can be at least one of the following: a desktop computer, a laptop computer, a tablet computer, a smart phone, etc. The mechanical equipment can be an excavator, an industrial robot or other mechanical equipment with a robotic arm. The embodiment of the present application does not impose specific restrictions on this.

[0090] The following explains a mechanical equipment control method provided in an embodiment of the present application.

[0091] Optional, Figure 4 A flow chart of a mechanical equipment control method provided in an embodiment of the present application is shown as follows: Figure 4 As shown, the method includes:

[0092] S401: Acquire the motion state of a first robotic arm at a first moment, the target motion state of the first robotic arm at a second moment, and the motion state of the second robotic arm at the first moment in the mechanical device to be controlled.

[0093] The first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the second moment is a preset future moment after the first moment; and the duration between the second moment and the first moment is a preset duration, for example, 1 second.

[0094] In some embodiments, the first moment may be the current moment, and the position and speed of the first robotic arm in the mechanical equipment to be controlled at the current moment, the target position and / or target speed of the first robotic arm at the second moment, and the position and speed of the second robotic arm at the current moment are obtained; wherein, the target position and / or target speed of the first robotic arm at the second moment refers to the motion state required for the first robotic arm at a future moment.

[0095] In an embodiment of the present application, the mechanical equipment to be controlled needs to follow a preset target trajectory during operation and complete the task corresponding to the target trajectory. The target trajectory includes the motion states of multiple trajectory points. The target motion state of the first robotic arm at the second moment can be the motion state of the first robotic arm at a trajectory point.

[0096] S402. Using a control model for the first robotic arm, according to the motion state of the first robotic arm at the first moment, the target motion state of the first robotic arm at the second moment, and the motion state of the second robotic arm at the first moment, obtain control parameters of the first robotic arm at the first moment.

[0097] The control parameter of the first robotic arm at the first moment may be a PWM value.

[0098] In an embodiment of the present application, the position and speed of the first robotic arm at the current moment, the target position and / or target speed of the first robotic arm at the second moment, and the position and speed of the second robotic arm at the current moment are input into the control model for the first robotic arm. The control model for the first robotic arm can output the control parameters of the first robotic arm at the current moment.

[0099] S403 : Control the first robotic arm according to the control parameters of the first robotic arm at the first moment, so that the motion state of the first robotic arm at the second moment reaches the target motion state.

[0100] Among them, the control model of the first robotic arm is a model trained using the above-mentioned model training method.

[0101] It should be noted that, by controlling the first robotic arm according to the control parameters of the first robotic arm at the first moment, the position of the first robotic arm at the second moment can be the target position, and the speed of the first robotic arm at the second moment can be the target speed, thereby achieving accurate control of the first robotic arm of the mechanical equipment.

[0102] In summary, an embodiment of the present application provides a method for controlling a mechanical device, comprising: obtaining the motion state of a first mechanical arm in a mechanical device to be controlled at a first moment and the target motion state of the first mechanical arm at a second moment, taking the motion state of the second mechanical arm at the first moment as the first mechanical arm and the second mechanical arm as mechanical arms with a coupling relationship in the mechanical device to be controlled; the second moment is a preset future moment after the first moment; using a control model for the first mechanical arm, according to the motion state of the first mechanical arm at the first moment, the target motion state of the first mechanical arm at the second moment, and the motion state of the second mechanical arm at the first moment, obtaining the control parameters of the first mechanical arm at the first moment; according to the control parameters of the first mechanical arm at the first moment, controlling the first mechanical arm so that the motion state of the first mechanical arm at the second moment reaches the target motion state. When training the control model for the first mechanical arm, the mutual influence between the first mechanical arm and the second mechanical arm is taken into account, so that the control model for the first mechanical arm obtained by training is more reliable, and the control parameters of the first mechanical arm at the first moment output by the control model are also more accurate, thereby achieving accurate control of the first mechanical arm.

[0103] Furthermore, the joint modeling of the first and second robotic arms takes into account the mutual influence between them, enabling coordinated control when the multi-joint motion of the mechanical device is performed, thus improving the tracking effect of the target trajectory. In a real-world environment, the method provided in the embodiment of the present application was used to control the first robotic arm. After testing five trajectories, the average error at the end of the display was 0.27 meters, while the average error at the end of the method using the related art was 0.298 meters, which significantly improved the tracking effect.

[0104] The following describes the model training device, processing equipment, storage medium, etc. used to execute the model training method provided in this application. Its specific implementation process and technical effects can be found in the relevant content of the above-mentioned model training method, which will not be repeated below.

[0105] Figure 5 A schematic diagram of the structure of a model training device provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, the device includes:

[0106] An acquisition module 501 is configured to acquire sample motion data of a first robotic arm in a mechanical device to be controlled at multiple time instants, sample motion states of a second robotic arm at said multiple time instants, and sample control data of the first robotic arm at said multiple time instants; the first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the sample motion data of the first robotic arm at each time instant includes: the sample motion state of the first robotic arm at said each time instant and a target sample motion state at a preset future time instant after said each time instant;

[0107] The training module 502 is used to train the initial control model based on the sample motion data of the first robotic arm at the multiple moments, the sample motion state of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain the control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0108] Optionally, the acquisition module 501 is specifically used to obtain the motion trajectory of the mechanical equipment to be controlled, and the motion trajectory includes: a plurality of trajectory points in a sequence of moments, the plurality of trajectory points having corresponding sample motion states of the first robotic arm, sample control data of the first robotic arm, and sample motion states of the second robotic arm; determining the sample motion data of the first robotic arm at multiple moments according to the sample motion states of the first robotic arm corresponding to the plurality of trajectory points; using the sample control data of the first robotic arm corresponding to the plurality of trajectory points as the sample control data of the first robotic arm at the multiple moments; and using the sample motion states of the second robotic arm corresponding to the plurality of trajectory points as the sample motion states of the second robotic arm at the multiple moments.

[0109] Optionally, the acquisition module 501 is specifically used to obtain the position and angular velocity of the first robotic arm at each moment, the target sample motion state at a preset future moment after each moment, the position and angular velocity of the second robotic arm at the multiple moments, and the sample pulse width modulation data of the first robotic arm at the multiple moments.

[0110] Optionally, the target sample motion state of the first robotic arm preset at a future moment after each moment includes: a target position and / or a target angular velocity.

[0111] Optionally, the training module 502 is specifically used to input the sample motion state of the first robotic arm at the multiple moments, the target sample motion state at the preset future moment after each moment, the sample motion state of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments into the initial control model to obtain the sample prediction control data of the multiple moments output by the initial control model; according to the sample prediction control data of the multiple moments, the sample control data of the first robotic arm at the multiple moments, and the preset loss function, the model parameters of the initial control model are updated until the number of iterations is greater than or equal to the preset number of iterations, thereby obtaining a control model for the first robotic arm.

[0112] Optionally, the initial control model includes: multiple fully connected layers connected in sequence, and the number of neurons in the multiple fully connected layers decreases in sequence.

[0113] Optionally, the time interval between two adjacent moments in the multiple moments is a sampling time interval between two adjacent trajectory points in the motion trajectory;

[0114] The time interval between two adjacent moments in the multiple moments is smaller than the time interval between each moment and the preset future moment.

[0115] The following describes the mechanical equipment control device, processing equipment, storage medium, etc. used to execute the mechanical equipment control method provided in this application. Its specific implementation process and technical effects can be found in the relevant content of the above-mentioned model training method, which will not be repeated below.

[0116] Figure 6 A schematic diagram of the structure of a model training device provided in an embodiment of the present application is shown in FIG. Figure 6 As shown, the device includes:

[0117] An acquisition module 601 is configured to acquire a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm being coupled robotic arms in the to-be-controlled mechanical device; the second moment being a preset future moment after the first moment; and employing a control model for the first robotic arm to obtain control parameters of the first robotic arm at the first moment based on the motion state of the first robotic arm at the first moment, the target motion state of the first robotic arm at the second moment, and the motion state of the second robotic arm at the first moment.

[0118] The control module 602 is used to control the first robotic arm according to the control parameters of the first robotic arm at the first moment, so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the method described in any one of the first aspects above.

[0119] The above-mentioned device is used to execute the method provided in the above-mentioned embodiment. Its implementation principle and technical effect are similar and will not be repeated here.

[0120] The above modules can be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASICs), one or more digital singnal processors (DSPs), or one or more field programmable gate arrays (FPGAs). For another example, when a module is implemented by scheduling program code through a processing element, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0121] Figure 7 A schematic diagram of the structure of a processing device provided in an embodiment of the present application is shown in FIG. Figure 7 As shown, the device includes: a processor 701 and a memory 702.

[0122] The memory 702 is used to store programs, and the processor 701 calls the programs stored in the memory 702 to execute the above-mentioned model training method embodiment or mechanical equipment control method embodiment. The specific implementation method and technical effect are similar and will not be repeated here.

[0123] For example, the model training method includes:

[0124] Acquiring sample motion data of a first robotic arm in a mechanical device to be controlled at multiple moments, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments; the first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the sample motion data of the first robotic arm at each moment includes: the sample motion state of the first robotic arm at the multiple moments and a target sample motion state at a preset future moment after the multiple moments;

[0125] The initial control model is trained based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0126] Optionally, the acquiring of sample motion data of a first robotic arm at multiple moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments includes:

[0127] Acquire a motion trajectory of the to-be-controlled mechanical device, the motion trajectory comprising: a plurality of trajectory points in a sequential time sequence, the plurality of trajectory points corresponding to sample motion states of the first mechanical arm, sample control data of the first mechanical arm, and sample motion states of the second mechanical arm;

[0128] determining sample motion data of the first manipulator at multiple moments according to the sample motion states of the first manipulator corresponding to the multiple trajectory points;

[0129] using the sample control data of the first robotic arm corresponding to the plurality of trajectory points as the sample control data of the first robotic arm at the plurality of moments;

[0130] The sample motion states of the second robotic arm corresponding to the multiple trajectory points are used as the sample motion states of the second robotic arm at the multiple moments.

[0131] Optionally, the acquiring of sample motion data of a first robotic arm at multiple moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments includes:

[0132] Obtain the position and angular velocity of the first robotic arm at each moment, the target sample motion state at a preset future moment after each moment, the position and angular velocity of the second robotic arm at the multiple moments, and the sample pulse width modulation data of the first robotic arm at the multiple moments.

[0133] Optionally, the target sample motion state of the first robotic arm preset at a future moment after each moment includes: a target position and / or a target angular velocity.

[0134] Optionally, the training of the initial control model based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain the control model of the first robotic arm includes:

[0135] Inputting the sample motion states of the first manipulator at the multiple moments, the target sample motion states at the future moments preset after each moment, the sample motion states of the second manipulator at the multiple moments, and the sample control data of the first manipulator at the multiple moments into the initial control model, to obtain the sample prediction control data at the multiple moments output by the initial control model;

[0136] Based on the sample prediction control data at the multiple moments, the sample control data of the first robotic arm at the multiple moments, and the preset loss function, the model parameters of the initial control model are updated until the number of iterations is greater than or equal to the preset number of iterations, so as to obtain a control model for the first robotic arm.

[0137] Optionally, the initial control model includes: multiple fully connected layers connected in sequence, and the number of neurons in the multiple fully connected layers decreases in sequence.

[0138] Optionally, the time interval between two adjacent moments in the multiple moments is a sampling time interval between two adjacent trajectory points in the motion trajectory;

[0139] The time interval between two adjacent moments in the multiple moments is smaller than the time interval between each moment and the preset future moment.

[0140] For example, the mechanical equipment control method includes:

[0141] Obtaining a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm are robotic arms in a coupled relationship in the to-be-controlled mechanical device; and the second moment is a preset future moment after the first moment;

[0142] Using a control model for the first robotic arm, according to a motion state of the first robotic arm at the first moment, a target motion state of the first robotic arm at the second moment, and a motion state of the second robotic arm at the first moment, a control parameter of the first robotic arm at the first moment is obtained;

[0143] According to the control parameters of the first robotic arm at the first moment, the first robotic arm is controlled so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the above-mentioned model training method.

[0144] In summary, when training the control model for the first robotic arm, the mutual influence between the first robotic arm and the second robotic arm is taken into account, so that the trained control model for the first robotic arm is more reliable, and the control parameters of the first robotic arm output by the control model at the first moment are also more accurate, thereby achieving accurate control of the first robotic arm.

[0145] Optionally, the present application also provides a program product, such as a computer-readable storage medium, comprising a program, which, when executed by a processor, is used to execute the above-mentioned model training method embodiment or mechanical equipment control method embodiment.

[0146] For example, the model training method includes:

[0147] Acquiring sample motion data of a first robotic arm in a mechanical device to be controlled at multiple moments, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments; the first robotic arm and the second robotic arm are coupled robotic arms in the mechanical device to be controlled; the sample motion data of the first robotic arm at each moment includes: the sample motion state of the first robotic arm at the multiple moments and a target sample motion state at a preset future moment after the multiple moments;

[0148] The initial control model is trained based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

[0149] Optionally, the acquiring of sample motion data of a first robotic arm at multiple moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments includes:

[0150] Acquire a motion trajectory of the to-be-controlled mechanical device, the motion trajectory comprising: a plurality of trajectory points in a sequential time sequence, the plurality of trajectory points corresponding to sample motion states of the first mechanical arm, sample control data of the first mechanical arm, and sample motion states of the second mechanical arm;

[0151] determining sample motion data of the first manipulator at multiple moments according to the sample motion states of the first manipulator corresponding to the multiple trajectory points;

[0152] using the sample control data of the first robotic arm corresponding to the plurality of trajectory points as the sample control data of the first robotic arm at the plurality of moments;

[0153] The sample motion states of the second robotic arm corresponding to the multiple trajectory points are used as the sample motion states of the second robotic arm at the multiple moments.

[0154] Optionally, the acquiring of sample motion data of a first robotic arm at multiple moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments includes:

[0155] Obtain the position and angular velocity of the first robotic arm at each moment, the target sample motion state at a preset future moment after each moment, the position and angular velocity of the second robotic arm at the multiple moments, and the sample pulse width modulation data of the first robotic arm at the multiple moments.

[0156] Optionally, the target sample motion state of the first robotic arm preset at a future moment after each moment includes: a target position and / or a target angular velocity.

[0157] Optionally, the training of the initial control model based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain the control model of the first robotic arm includes:

[0158] Inputting the sample motion states of the first manipulator at the multiple moments, the target sample motion states at the future moments preset after each moment, the sample motion states of the second manipulator at the multiple moments, and the sample control data of the first manipulator at the multiple moments into the initial control model, to obtain the sample prediction control data at the multiple moments output by the initial control model;

[0159] Based on the sample prediction control data at the multiple moments, the sample control data of the first robotic arm at the multiple moments, and the preset loss function, the model parameters of the initial control model are updated until the number of iterations is greater than or equal to the preset number of iterations, so as to obtain a control model for the first robotic arm.

[0160] Optionally, the initial control model includes: multiple fully connected layers connected in sequence, and the number of neurons in the multiple fully connected layers decreases in sequence.

[0161] Optionally, the time interval between two adjacent moments in the multiple moments is a sampling time interval between two adjacent trajectory points in the motion trajectory;

[0162] The time interval between two adjacent moments in the multiple moments is smaller than the time interval between each moment and the preset future moment.

[0163] For example, the mechanical equipment control method includes:

[0164] Obtaining a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm are robotic arms in a coupled relationship in the to-be-controlled mechanical device; and the second moment is a preset future moment after the first moment;

[0165] Using a control model for the first robotic arm, according to a motion state of the first robotic arm at the first moment, a target motion state of the first robotic arm at the second moment, and a motion state of the second robotic arm at the first moment, a control parameter of the first robotic arm at the first moment is obtained;

[0166] According to the control parameters of the first robotic arm at the first moment, the first robotic arm is controlled so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the above-mentioned model training method.

[0167] In summary, when training the control model for the first robotic arm, the mutual influence between the first robotic arm and the second robotic arm is taken into account, so that the trained control model for the first robotic arm is more reliable, and the control parameters of the first robotic arm output by the control model at the first moment are also more accurate, thereby achieving accurate control of the first robotic arm.

[0168] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0169] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0170] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0171] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (English: Read-Only Memory, abbreviated: ROM), a random access memory (English: Random Access Memory, abbreviated: RAM), a disk or an optical disk, and other media that can store program code.

[0172] The above are merely preferred embodiments of the present application and are not intended to limit the present application. Those skilled in the art will readily appreciate that various modifications and variations are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A model training method, characterized in that: include: Acquire sample motion data of a first robotic arm at multiple moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the multiple moments, and sample control data of the first robotic arm at the multiple moments; The first robotic arm and the second robotic arm are robotic arms in a coupling relationship in the mechanical device to be controlled; The sample motion data of the first manipulator at each moment includes: the sample motion state of the first manipulator at each moment and the target sample motion state at a preset future moment after each moment; The initial control model is trained based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

2. The method according to claim 1, characterized in that The acquiring of sample motion data of a first robotic arm at a plurality of moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the plurality of moments, and sample control data of the first robotic arm at the plurality of moments comprises: Acquire a motion trajectory of the to-be-controlled mechanical device, the motion trajectory comprising: a plurality of trajectory points in a sequential time sequence, the plurality of trajectory points corresponding to sample motion states of the first mechanical arm, sample control data of the first mechanical arm, and sample motion states of the second mechanical arm; determining sample motion data of the first manipulator at multiple moments according to the sample motion states of the first manipulator corresponding to the multiple trajectory points; using the sample control data of the first robotic arm corresponding to the plurality of trajectory points as the sample control data of the first robotic arm at the plurality of moments; The sample motion states of the second robotic arm corresponding to the multiple trajectory points are used as the sample motion states of the second robotic arm at the multiple moments.

3. The method according to claim 1, characterized in that The acquiring of sample motion data of a first robotic arm at a plurality of moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the plurality of moments, and sample control data of the first robotic arm at the plurality of moments comprises: Obtain the position and angular velocity of the first robotic arm at each moment, the target sample motion state at a preset future moment after each moment, the position and angular velocity of the second robotic arm at the multiple moments, and the sample pulse width modulation data of the first robotic arm at the multiple moments.

4. The method according to claim 1, wherein The target sample motion state of the first robotic arm preset at a future moment after each moment includes: a target position and / or a target angular velocity.

5. The method according to claim 1, wherein The training of the initial control model based on the sample motion data of the first manipulator at the multiple moments, the sample motion states of the second manipulator at the multiple moments, and the sample control data of the first manipulator at the multiple moments to obtain the control model of the first manipulator includes: Inputting the sample motion states of the first manipulator at the multiple moments, the target sample motion states at the future moments preset after each moment, the sample motion states of the second manipulator at the multiple moments, and the sample control data of the first manipulator at the multiple moments into the initial control model, to obtain the sample prediction control data at the multiple moments output by the initial control model; Based on the sample prediction control data at the multiple moments, the sample control data of the first robotic arm at the multiple moments, and the preset loss function, the model parameters of the initial control model are updated until the number of iterations is greater than or equal to the preset number of iterations, so as to obtain a control model for the first robotic arm.

6. The method according to claim 1, characterized in that The initial control model includes: a plurality of fully connected layers connected in sequence, and the number of neurons in the plurality of fully connected layers decreases in sequence.

7. The method according to claim 2, characterized in that The time interval between two adjacent moments in the multiple moments is the sampling time interval between two adjacent trajectory points in the motion trajectory; The time interval between two adjacent moments in the multiple moments is smaller than the time interval between each moment and the preset future moment.

8. A method for controlling a mechanical device, characterized in that: include: Obtaining a motion state of a first robotic arm in a mechanical device to be controlled at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm are robotic arms in a coupled relationship in the mechanical device to be controlled; the second moment is a preset future moment after the first moment; Using a control model for the first robotic arm, according to a motion state of the first robotic arm at the first moment, a target motion state of the first robotic arm at the second moment, and a motion state of the second robotic arm at the first moment, a control parameter of the first robotic arm at the first moment is obtained; According to the control parameters of the first robotic arm at the first moment, the first robotic arm is controlled so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the method described in any one of claims 1 to 7 above.

9. A model training device, characterized in that: include: an acquisition module, configured to acquire sample motion data of a first robotic arm at a plurality of moments in the mechanical device to be controlled, sample motion states of a second robotic arm at the plurality of moments, and sample control data of the first robotic arm at the plurality of moments; The first robotic arm and the second robotic arm are robotic arms in a coupling relationship in the mechanical device to be controlled; The sample motion data of the first manipulator at each moment includes: the sample motion state of the first manipulator at each moment and the target sample motion state at a preset future moment after each moment; A training module is used to train an initial control model based on the sample motion data of the first robotic arm at the multiple moments, the sample motion states of the second robotic arm at the multiple moments, and the sample control data of the first robotic arm at the multiple moments, so as to obtain a control model of the first robotic arm. The control model of the first robotic arm is used to output control parameters for the first robotic arm.

10. A mechanical equipment control device, characterized in that: include: an acquisition module, configured to acquire a motion state of a first robotic arm in a to-be-controlled mechanical device at a first moment, a target motion state of the first robotic arm at a second moment, and a motion state of a second robotic arm at the first moment; the first robotic arm and the second robotic arm being robotic arms in a coupled relationship in the to-be-controlled mechanical device; the second moment being a preset future moment after the first moment; and employing a control model for the first robotic arm to obtain control parameters of the first robotic arm at the first moment based on the motion state of the first robotic arm at the first moment, the target motion state of the first robotic arm at the second moment, and the motion state of the second robotic arm at the first moment; A control module is used to control the first robotic arm according to the control parameters of the first robotic arm at the first moment, so that the motion state of the first robotic arm at the second moment reaches the target motion state, wherein the control model of the first robotic arm is a model trained using the method described in any one of claims 1 to 7 above.

11. A processing device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program executable by the processor, and when the processor executes the computer program, it implements the model training method described in any one of claims 1 to 7, or the mechanical equipment control method described in claim 8.

12. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is read and executed, it implements the model training method described in any one of claims 1 to 7, or the mechanical equipment control method described in claim 8.

Images