Manipulator control method and device and electronic equipment

Abstract

The invention discloses a manipulator control method and device and electronic equipment, and the method comprises the steps that a first task is acquired, a first neural network is determined according to the first task, a plurality of pre-trained neural networks are configured in a control unit of a manipulator, and the first neural network is a neural network matched with the first task in the neural networks; reconstructing the first task by using a first neural network to obtain a control strategy of the reconstructed first task, the control strategy comprising a series of basic control instructions; control parameters for controlling the manipulator are obtained based on the control strategy, the manipulator is controlled to execute the control action according to the control parameters, and the control parameters comprise parameters corresponding to the basic control instruction. Due to the fact that spatial position control is completed through the neural network configured in the manipulator control unit, the method can well adapt to application scenes without networking, and computing resources of the control unit do not need to be additionally consumed.

Description

Technical Field

[0001] The embodiments of the present application belong to the field of robot technology, and in particular, relate to a robot control method, device and electronic equipment. Background Art

[0002] There are many control methods for manipulators, such as traditional PID control (proportional, integral, differential control), fuzzy control and other methods. When a system and the controlled object are not fully understood, or when the system parameters cannot be obtained through effective measurement methods, PID control technology is most suitable. PID control, in practice, also has PI and PD control. The PID controller calculates the control quantity based on the system error using proportion, integration, and differentiation. In fuzzy control, the input quantity is fuzzy quantized to become a fuzzy variable, and the fuzzy variable is inferred by fuzzy rules to obtain a fuzzy output, and the clear output quantity is obtained through defuzzification for control.

[0003] Spatial computing has many applications in robot control. However, how to control the spatial position of a robot without an Internet connection, that is, when the robot's computing power is limited, is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present application aims to solve at least one of the technical problems existing in the related art. To this end, the present application proposes a manipulator control method, device and electronic device, which can realize flexible and changeable self-learning control of the manipulator without networking, and effectively solve the problem of limited computing power of the manipulator.

[0005] In a first aspect, the present application provides a manipulator control method, the method comprising:

[0006] Acquire a first task, and determine a first neural network according to the first task, wherein the control unit is configured with a plurality of pre-trained neural networks, and the first neural network is a neural network among the neural networks that matches the first task;

[0007] Reconstructing the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions;

[0008] Based on the control strategy, control parameters for controlling the manipulator are acquired, and the manipulator is controlled to perform a control action according to the control parameters, wherein the control parameters include parameters corresponding to the basic control instructions.

[0009] In some embodiments, the obtaining of the first task of the manipulator comprises:

[0010] Acquire the first task and the environment status information corresponding to the first task, wherein the environment status information includes the surrounding environment information and the self status information of the manipulator in the simulation environment;

[0011] The step of reconstructing the first task using a pre-trained first neural network to obtain a control strategy for the reconstructed first task includes:

[0012] The surrounding environment information, the self-state information and the first task are input into the first neural network to obtain a reconstructed control strategy for the first task, wherein the control strategy includes the basic control instructions and the scene control instructions.

[0013] In some embodiments, the acquiring control parameters for controlling the manipulator based on the control strategy, and controlling the manipulator to perform a control action according to the control parameters, includes:

[0014] Acquire control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions;

[0015] Predicting the spatial pre-operation trajectory of the manipulator according to the control parameters;

[0016] When it is determined by using the scene parameters that there are no running obstacles in the spatial pre-operation trajectory, the manipulator is controlled to run according to the spatial pre-operation trajectory.

[0017] In some embodiments, if the scene parameters are used to determine that there are operating obstacles in the spatial pre-operation trajectory, the spatial reconstruction operation trajectory of the manipulator is obtained according to the control parameters and the scene parameters, and the manipulator is controlled to operate according to the spatial reconstruction operation trajectory.

[0018] In some embodiments, obtaining the spatial reconstruction operation trajectory of the manipulator according to the control parameters and the scene parameters, and controlling the manipulator to operate according to the spatial reconstruction operation trajectory, includes:

[0019] The control parameters are added or deleted according to the scene parameters to obtain reconstructed control parameters, and the spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

[0020] In some embodiments, if the spatially reconstructed operation trajectory obtained based on the reconstructed control parameters still has operation obstacles, an alarm message indicating that the first task cannot be completed is issued.

[0021] In some embodiments, the acquiring control parameters for controlling the manipulator based on the control strategy, and controlling the manipulator to perform a control action according to the control parameters, includes:

[0022] Acquire control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions;

[0023] Based on the similarity scene measurement algorithm, the corresponding historical control parameters in the historical similar scenes are obtained according to the scene parameters;

[0024] The control parameters are reconstructed according to the historical control parameters to obtain reconstructed control parameters, and a spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

[0025] In some embodiments, each of the control parameters corresponds to a basic action, and obtaining the spatial reconstruction operation trajectory according to the reconstructed control parameters, and controlling the manipulator to operate according to the spatial reconstruction operation trajectory, includes:

[0026] The basic actions of the manipulator and the execution order of each basic action are obtained according to the reconstructed control parameters, and the manipulator is controlled to execute the basic actions in sequence according to the execution order to complete the operation of the spatially reconstructed operation trajectory.

[0027] In a second aspect, the present application provides a manipulator control device, the device comprising:

[0028] an acquisition module, configured to acquire a first task, and determine a first neural network according to the first task, wherein the control unit is configured with a plurality of pre-trained neural networks, and the first neural network is a neural network among the neural networks that matches the first task;

[0029] A reconstruction module, used to reconstruct the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions;

[0030] A control module is used to obtain control parameters for controlling the manipulator based on the control strategy, and control the manipulator to perform control actions according to the control parameters, wherein the control parameters include parameters corresponding to the basic control instructions.

[0031] In some embodiments, the acquisition module is further used to acquire the first task and the environmental status information corresponding to the first task, wherein the environmental status information includes the surrounding environment information and the self-state information of the manipulator in the simulation environment;

[0032] The reconstruction module is further used to input the surrounding environment information, the self-state information and the first task into the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes the basic control instructions and the scene control instructions.

[0033] In some embodiments, the control module is also used to obtain control parameters and scene parameters for controlling the manipulator based on the control strategy, the scene parameters being parameters corresponding to the scene control instructions; predicting the spatial pre-operation trajectory of the manipulator according to the control parameters; if it is determined by the scene parameters that there are no operating obstacles to the spatial pre-operation trajectory, controlling the manipulator to operate according to the spatial pre-operation trajectory; if it is determined by the scene parameters that there are operating obstacles to the spatial pre-operation trajectory, then obtaining the spatially reconstructed operating trajectory of the manipulator according to the control parameters and the scene parameters, and controlling the manipulator to operate according to the spatially reconstructed operating trajectory.

[0034] In some embodiments, the control module is also used to add or delete the control parameters according to the scene parameters to obtain reconstructed control parameters, and obtain the spatial reconstructed operation trajectory according to the reconstructed control parameters, and control the manipulator to operate according to the spatial reconstructed operation trajectory.

[0035] In some embodiments, the control module is further configured to issue a warning message indicating that the first task cannot be completed if there are still operation obstacles in the spatially reconstructed operation trajectory obtained based on the reconstructed control parameters.

[0036] In some embodiments, the control module is also used to obtain control parameters and scene parameters for controlling the manipulator based on the control strategy, where the scene parameters are parameters corresponding to the scene control instructions; based on a similarity scene measurement algorithm, corresponding historical control parameters in historical similar scenes are obtained according to the scene parameters; the control parameters are reconstructed according to the historical control parameters to obtain reconstructed control parameters, and a spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

[0037] In some embodiments, the control module is also used to obtain the basic actions of the manipulator and the execution order of each basic action according to the reconstructed control parameters, and control the manipulator to execute the basic actions in sequence according to the execution order to complete the operation of the spatial reconstruction operation trajectory.

[0038] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the manipulator control method as described in the first aspect above is implemented.

[0039] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the manipulator control method as described in the first aspect above.

[0040] In a fifth aspect, the present application provides a chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the manipulator control method as described in the first aspect.

[0041] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the manipulator control method as described in the first aspect above.

[0042] The above one or more technical solutions in the embodiments of the present application have at least the following technical effects:

[0043] The manipulator control method, device and electronic device provided in the embodiments of the present application, the manipulator control method provided in the embodiments of the present application, obtains a first task through the control unit of the manipulator, selects and determines a first neural network according to the first task, the first neural network is a neural network that matches the first task among multiple pre-trained neural networks configured in the control unit, and reconstructs the first task using the first neural network to obtain a control strategy for the reconstructed first task, the control strategy includes a series of basic control instructions, control parameters for controlling the manipulator are obtained based on the series of basic control instructions, and finally the manipulator is controlled according to the control parameters to perform control actions and complete corresponding movements. Since the method uses the neural network configured in the manipulator control unit to complete spatial position control, it can be well adapted to application scenarios without Internet access, and does not require additional consumption of computing resources of the control unit itself, and can effectively solve the problem of limited computing power of the manipulator during spatial position control.

[0044] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0046] Figure 1It is a schematic diagram of the structural framework of the manipulator provided in an embodiment of the present application;

[0047] Figure 2 It is a flowchart of a manipulator control method provided in an embodiment of the present application;

[0048] Figure 3 is a schematic diagram of the structure of a manipulator control device provided in an embodiment of the present application;

[0049] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0051] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0052] First, the overall description of the structure of the robot is given in Figure 1 , Figure 1 The figure shows a schematic diagram of the structural framework of the manipulator. The manipulator includes a control unit, which includes multiple neural networks and a skill library corresponding to each application network, so that after inputting the corresponding task, the corresponding neural network is matched from the multiple neural networks, and the corresponding neural network is used to select the skill to complete the task from the corresponding skill library. It can be understood that pre-configuring the neural network and the corresponding skill library in the control unit will not affect the original computing resources of the control unit, and it shares the pressure of the computing resources of the control unit itself to calculate the control position.

[0053] In conjunction with the accompanying drawings, the manipulator control method, device and electronic device provided by the embodiment of the present application are described in detail through specific embodiments and their application scenarios. Among them, the manipulator control method can be applied to the manipulator, and can be specifically executed by hardware or software in the manipulator.

[0054] like Figure 2 As shown, the manipulator control method is applied to a control unit of the manipulator, and the method comprises the following steps:

[0055] S100, obtaining a first task of the manipulator, and determining a first neural network based on the first task, wherein a control unit is configured with a plurality of pre-trained neural networks, and the first neural network is a neural network among the neural networks that matches the first task.

[0056] It is understandable that the first task in the embodiment of the present application can be obtained by the control unit of the manipulator, or it can be directly obtained by the service first and then transmitted to the control unit of the manipulator. Specifically, the manipulator / server receives the first task input by the operator, that is, the first task can be input through the receiving interface of the hardware device; it can also be a pre-input task, and the manipulator / server can select the first task from the pre-input tasks; it can also be a pre-set task generation rule, so that the manipulator can generate the first task by itself according to the generation rule. In some embodiments, the generation rule can be self-learned to form tasks with increasing difficulty. As an example, for example, if the initial task is for the manipulator to make a fist, the first task generated with a greater difficulty than the initial task can be for the manipulator to switch between a fist and an open fist, switching every 3 seconds; for another example: the first task with a further increased difficulty can be for the manipulator to pick up a cup from the table. In some embodiments, the first task can be a control instruction, and the control instruction can be an action instruction such as picking up, lifting, dragging, and pulling.

[0057] It should be noted that the manipulator in the embodiment of the present application may be an intelligent robot. The control unit of the manipulator may include a main processor, an AI processor or an image processor, etc., which is not limited here.

[0058] S200, reconstructing the first task using a first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions.

[0059] In some embodiments, after acquiring the first task, the server determines the first neural network according to the first task, and then uses the first neural network to reconstruct the first task to obtain the control strategy corresponding to the reconstructed first task. It can be understood that the first neural network can be understood as being used to obtain the path and result of completing the first task from the knowledge base or skill base, and disassemble it into basic control instructions, so that the control strategy of the first task is composed of these basic control instructions. Each skill in the skill base or knowledge base can be specifically expressed as a neural network, or as an operation rule. As an example, the skills in the knowledge base or skill base can be specifically for picking up an empty cup with one hand, or picking up a watermelon with both hands, or picking up a weight of more than 10 kg with both hands, etc.

[0060] In some embodiments, the first neural network can be pre-trained with different first tasks for its skill library or knowledge base and then configured to the control unit of the manipulator. It should be noted here that before using the first neural network, a step of obtaining the first neural network can also be included, that is, pre-downloading multiple neural networks from the server and configuring them to the control unit of the manipulator. Of course, they can also be trained directly in the control unit of the manipulator. For example, the control unit of the manipulator is configured with 5 skill libraries and 5 neural networks corresponding to the 5 skill libraries, which are respectively used to perform the 5 tasks of single-handed grasping, single-handed lifting, single-handed throwing, single-handed scratching, and single-handed pushing and pulling. The first task determined by the control unit of the manipulator is single-handed grasping, then the manipulator needs to obtain the neural network corresponding to the single-handed grasping. The example here is not used to limit this solution.

[0061] It can be understood that a variety of neural networks and skill libraries can be stored in the control unit. The first neural network can be a neural network trained based on the simulation environment corresponding to the second task, that is, the first neural network can be a mature neural network among at least one pre-trained neural network. Correspondingly, the control unit can determine the skill library corresponding to the first neural network as the skill library. More specifically, the operator can select the first neural network from at least one pre-trained neural network, and then the control unit obtains the first neural network selected by technical personnel in this field; or the control unit can autonomously select the first neural network from at least one pre-trained neural network, wherein the semantic information of the first task is similar to the semantic information of the second task. Specifically, the similarity between the semantic information of the first task and the semantic information of the second task may refer to the use of a neural network to obtain the semantic information of the first task and the second task and compare them to determine that the semantic information of the first task and the semantic information of the second task are similar; it may also be that the constraints obtained by decomposing the first task and the second task are similar. As an example, if the result of decomposing the first task is to lift something, and the constraint condition obtained by decomposing the second task is a specific thing that is particularly heavy or particularly light, then the semantic information of the first task can be regarded as similar to the semantic information of the second task; it may also be that the operating environment of the first task and the second task is similar. As an example, if the first task is to lift something and the second task is to hold up something, then the semantic information of the first task can be regarded as similar to the semantic information of the second task, etc. Of course, other methods can also be used to determine that the semantic information of the first task and the semantic information of the second task are similar. The examples given here are only for the convenience of understanding of this solution, and all implementation methods are not exhaustive.

[0062] In some embodiments, the control unit may also initialize a first neural network after determining the first task and the neural network type of the first neural network, and initially train a skill library based on the simulation environment corresponding to the first task and using a reinforcement learning algorithm; in another implementation method, the control unit may also initialize a first neural network after determining the first task and the neural network type of the first neural network, and then the technical personnel in this field may configure at least one skill in the skill library according to the first task, and so on. Since the skills in the skill library may be expanded in subsequent steps, the number of skills in the skill library does not need to be particularly large, and the skill library contains some basic action components of the embodied robot, such as drinking water, carrying, walking, etc.

[0063] In some embodiments, step S100 includes: acquiring the first task and environmental status information corresponding to the first task, wherein the environmental status information includes surrounding environment information and self-status information of the manipulator in a simulation environment.

[0064] In this embodiment, the control unit inputs the environmental state information related to the first task into the first neural network to obtain the skills selected by the first neural network from the skill library. Specifically, the environmental state information includes the surrounding environment information of the manipulator in the simulation environment corresponding to the first task and the manipulator's own state information. Specifically, it can include map information around the intelligent robot, interaction information of the intelligent robot, movement information of adjacent intelligent robots, etc. As an example, the environmental state information of the embodiment of the present application may include positioning information, map information, space information, interaction information, task information, operation information, etc. Specifically, information such as posture, position, coordinates, etc. can be obtained by an inertial measurement unit, motion sensor, gravity sensor, etc. provided on the manipulator.

[0065] Step S200 includes: inputting the surrounding environment information, the self-state information and the first task into the first neural network to obtain a reconstructed control strategy for the first task, wherein the control strategy includes the basic control instructions and the scene control instructions.

[0066] In this embodiment, in addition to considering the instruction information of the first task itself, the surrounding environment information and the self-state information of the manipulator are also fully considered. The control strategy reconstructed by considering this information can better fit the operation of the manipulator and realize accurate motion control of the manipulator. For example: the first task is to reach a certain destination. If it is not combined with the surrounding environment information, it will directly move in a straight line. This way of moving may cause the manipulator to encounter obstacles during operation and be unable to move. The control strategy after considering the surrounding environment information will avoid obstacles and take the straight + turn instructions to control the manipulator to complete the action; and considering the manipulator's own state information can avoid conflicts between the two tasks. For example, if the manipulator is currently executing a task, it needs to wait for the task to be completed before executing the first task.

[0067] In some embodiments, the environmental state information may be in the form of a picture, a sequence data, or other data forms. When the environmental state information is input into the first neural network, the neural network type of the first neural network may be determined according to the data type of the input data. For example, if the input data is picture data, the first neural network may use a convolutional neural network (CNN). For example, if the input data is sequence data, the first neural network may use a recurrent neural network (RNN). Other cases are not listed here one by one.

[0068] In some embodiments, when the first neural network outputs a control strategy, the first neural network predicts the skill to be selected based on the input first task and the environmental state information corresponding to the first task, and outputs the serial number of the corresponding skill, so that the server can select the corresponding skill from the skill library based on the output serial numbers. It is understandable that a corresponding unique serial number can be configured in advance for each skill in the skill library. Each skill can be understood as the most basic action control instruction when the manipulator performs an action, such as forward, backward, left turn, right turn, etc.

[0069] It can be understood that the process of reconstructing the first task by the first neural network is equivalent to the process of disassembling and reorganizing the first task to obtain a series of basic control instructions.

[0070] S300, acquiring control parameters for controlling the manipulator based on the control strategy, so as to control the manipulator to perform a control action according to the control parameters.

[0071] In some embodiments, step S300 includes: acquiring basic actions for controlling the manipulator and the execution order of each basic action according to the control parameters, and controlling the manipulator to perform each basic action in sequence according to the execution order.

[0072] After obtaining the control strategy of the robot, it is equivalent to obtaining a series of basic control instructions of the robot. Each control instruction corresponds to a control parameter. By adjusting the control parameters of the robot, the robot can perform each basic action in sequence according to the execution order to achieve the set path and / or action operation.

[0073] Furthermore, in some embodiments, the basic actions of the manipulator and the execution order of each basic action are obtained according to the control parameters and / or the scene parameters, and the manipulator is controlled to perform the basic actions in sequence according to the execution order to complete the corresponding operation. Specifically, when the manipulator is controlled to perform the basic actions in sequence according to the execution order, the performance parameters such as the time, computing power, and resources used and the energy, strength, angle, speed, arm / finger degrees of freedom, etc. used by the basic actions are estimated, and the manipulator is controlled according to the estimated performance parameters. Specifically, when outputting the control strategy, the degree of freedom of the manipulator needs to be considered. That is, when determining the control strategy, the relationship between the degree of freedom required to complete the first task and the degree of freedom of the manipulator itself is judged. When the control strategy determines that the degree of freedom required to complete the first task is greater than the degree of freedom of the manipulator itself, the control strategy is further simplified; when the control strategy determines that the degree of freedom required to complete the first task is less than the degree of freedom of the manipulator itself, the control strategy of the first task is executed. When the control strategy determines that the degrees of freedom required to complete the first task are less than the degrees of freedom of the manipulator itself, it means that the degrees of freedom of the control strategy for the first task are redundant and belong to a better control strategy, which can be learned more by the neural network.

[0074] In some embodiments, step S300 includes: obtaining control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions. Predicting the spatial pre-operation trajectory of the manipulator according to the control parameters. If it is determined by the scene parameters that there are no operating obstacles to the spatial pre-operation trajectory, the manipulator is controlled to operate according to the spatial pre-operation trajectory. If it is determined by the scene parameters that there are operating obstacles to the spatial pre-operation trajectory, the spatial reconstruction operation trajectory of the manipulator is obtained according to the control parameters and the scene parameters, and the manipulator is controlled to operate according to the spatial reconstruction operation trajectory.

[0075] In this embodiment, the control strategy includes basic control instructions and scene control instructions. Control parameters are obtained based on the basic control instructions, and scene parameters are obtained based on the scene control instructions. When the manipulator is controlled based on the control parameters and the scene parameters, the scene parameters are first used to determine whether there are any operating obstacles in the spatial pre-operation trajectory predicted by the control parameters. If so, the spatial operation trajectory of the manipulator is reconstructed according to the control parameters and the scene parameters to obtain a spatial reconstructed operation trajectory, and the manipulator is controlled to operate according to the spatial reconstructed operation trajectory. If not, the manipulator is directly operated according to the spatial pre-operation trajectory without reconstruction, thereby avoiding unnecessary parameter addition and deletion processes and improving the operation efficiency of the manipulator.

[0076] Specifically, in some embodiments, the spatially reconstructed operation trajectory of the manipulator is obtained according to the control parameters and the scene parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory, including: adding and deleting the control parameters according to the scene parameters to obtain reconstructed control parameters, and obtaining the spatially reconstructed operation trajectory according to the reconstructed control parameters, and controlling the manipulator to operate according to the spatially reconstructed operation trajectory.

[0077] If the spatial pre-operation trajectory of the manipulator obtained by the control parameters has obstacles, the space and the operation trajectory need to be reconstructed. During reconstruction, the original control parameters will be added or deleted according to the scene parameters to obtain the reconstructed control parameters, and the spatial reconstructed operation trajectory is obtained from the reconstructed control parameters, and then the manipulator is controlled to run according to the spatial reconstructed operation trajectory to achieve obstacle avoidance of the manipulator. For example, if the manipulator is to run from point A to point B, the spatial pre-operation trajectory predicted by the control parameters is a straight line operation trajectory of AB, but combined with the scene parameters, it is found that there are obstacles in the straight line operation trajectory between AB, then the original straight control parameters need to be deleted and changed to right turn, straight, left turn, straight, left turn, straight, so as to achieve unimpeded operation of the manipulator. Another example: the action of drinking water, the goal is to drink water, the result is to drink water, the process is to lift the cup to drink water, and at the same time, it is necessary to pay attention to whether there is water in the cup, the color, material, position, size, quality of the cup, and whether there are other objects around the cup, etc. If it is known that there is only one cup on the table and who owns the cup in combination with the scene parameters, the parameters such as the color, material, size, quality of the cup, and whether there are other objects around the cup can be deleted, and only the control parameters of the cup position recognition and grabbing and lifting the cup are retained to avoid overly complex and unnecessary movements. Another example: the most basic control of the above-mentioned drinking action is to lift the cup. If it is known from the scene parameters that the current scene is a banquet scene, the basic control action of toasting is added to match the current environment.

[0078] It is worth mentioning that the reconstruction is performed only once, because if the reconstruction is performed more times, the prototype of the first task may be lost and the first task will not be able to be executed.

[0079] In some embodiments, if the spatially reconstructed operation trajectory obtained based on the reconstructed control parameters still has operation obstacles, an alarm message indicating that the first task cannot be completed is issued.

[0080] In some scenarios, such as scenarios with dense spatial objects, obstacles cannot be avoided even after multiple reconstructions. In this case, a warning message is issued that the first task cannot be completed, informing the user that a suitable path cannot be found to complete the first task and that the user needs to re-enter the first task or abandon the first task.

[0081] In some embodiments, step S300 further includes: based on a similarity scene measurement algorithm, obtaining historical control parameters corresponding to historical similar scenes according to scene parameters, wherein the scene parameters are parameters corresponding to the scene control instructions. Reconstructing the control parameters according to the historical control parameters to obtain reconstructed control parameters, and obtaining the spatial reconstruction operation trajectory according to the reconstructed control parameters, and controlling the manipulator to operate according to the spatial reconstruction operation trajectory.

[0082] This embodiment can be understood as an equivalent embodiment of the aforementioned use of scene parameters to determine whether there are any operating obstacles in the spatial pre-operation trajectory predicted by the control parameters. This embodiment simplifies the idea of ​​determining the operation trajectory by scene parameters, that is, directly using scene parameters to find the corresponding historical control parameters in historical similar scenes, and reconstructing the control parameters according to the historical control parameters, and obtaining the spatially reconstructed operation trajectory according to the obtained reconstructed control parameters, and controlling the manipulator to operate according to the spatially reconstructed operation trajectory. This embodiment is more efficient and quick, and improves the control efficiency of the manipulator.

[0083] In some embodiments, step S300 also includes: obtaining the basic actions of the manipulator and the execution order of each basic action according to the reconstructed control parameters, and controlling the manipulator to execute the basic actions in sequence according to the execution order to complete the operation of the spatially reconstructed operation trajectory.

[0084] The robot control method provided in the embodiment of the present application obtains a first task through the control unit of the robot, selects and determines a first neural network according to the first task, and the first neural network is a neural network that matches the first task among multiple pre-trained neural networks configured in the control unit, and reconstructs the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions, and control parameters for controlling the robot are obtained based on the series of basic control instructions, and finally the robot is controlled according to the control parameters to perform control actions and complete corresponding movements. Since the method uses the neural network configured in the control unit of the robot to complete spatial position control, it can adapt well to application scenarios without Internet access, and does not require additional consumption of computing resources of the control unit itself, and can effectively solve the problem of limited computing power of the robot during spatial position control; and using neural networks, self-learning of robot control can also be achieved to meet the flexible and changeable control needs of the robot.

[0085] In some embodiments, based on a reinforcement learning algorithm, the control strategy is used to train a neural network model, obtain feedback parameters, and adjust the feedback parameters to obtain a trained second neural network.

[0086] Using reinforcement learning algorithms, control strategies are memorized and trained to enable rapid control execution the next time a similar scenario is encountered.

[0087] It should be noted that when the control unit controls the action of the manipulator based on the simulation environment corresponding to the first task, it will first complete the first task through at least one skill selected by the first neural network. After the control unit controls the intelligent manipulator to perform the first task, it will obtain the data for executing the first task (such as control parameters, scene parameters), and use the reinforcement learning algorithm to train the parameters of the first neural network again to obtain an updated first neural network. Of course, the above training process can also be implemented on the server side, thereby reducing the computing pressure of the manipulator.

[0088] More specifically, the server inputs the environmental state information into the first neural network to obtain the skills selected by the first neural network. The environmental state information includes the environmental information around the intelligent manipulator in the simulation environment corresponding to the first task and the state information of the intelligent manipulator itself. The skills for executing the first task are selected by the first neural network, and then the basic control instructions are obtained. After that, the intelligent manipulator can be controlled in the simulator to perform operations corresponding to the basic control instructions. During the execution process, the server performs an execution state acquisition operation for the skills selected by the first neural network every preset time until the execution state of the skills selected by the first neural network is the end of execution; the server obtains the data generated during the operation corresponding to the execution of the control instruction by the intelligent manipulator, and the data includes any one or more of the operation path, operation speed or operation destination of the intelligent manipulator; the server updates the parameters of the first neural network based on the data using the reinforcement learning algorithm. Among them, the concepts of intelligent manipulator, preset time, and execution state are all introduced in detail in the above description and will not be repeated here. In an embodiment of the present application, the server determines whether the skill selected by the first neural network has been completed by obtaining the execution status of the skill selected by the first neural network once every preset time period, so that the server can iteratively update the new skill strategy and the parameters of the new skill in a timely manner according to the operation behavior information of the intelligent manipulator, which is conducive to improving the accuracy of the training process.

[0089] In some embodiments, the method further includes: after each basic action is executed, determining whether the first task is completed; if completed, ending the first task; if not completed, returning to step S100 so as to timely adjust the control action of the manipulator according to the current environmental state parameters of the manipulator.

[0090] When judging whether the first task is completed, it can be determined based on whether the last basic action of the first task is completed, or it can be determined based on the operating state of the manipulator collected by the motion sensor of the manipulator.

[0091] In some embodiments, when reconstructing the control parameters of the manipulator according to the first task, an execution timestamp of each basic action can be generated, and the manipulator can be controlled to complete the corresponding basic action within the corresponding execution timestamp to improve execution efficiency. If the execution timestamp corresponding to the last basic action ends, the first task is determined to be completed.

[0092] The robot control method provided in the embodiment of the present application can be executed by a robot control device. In the embodiment of the present application, the robot control device provided in the embodiment of the present application is described by taking the robot control method executed by the robot control device as an example.

[0093] The present application also provides a robot control device, such as Figure 3 As shown, the manipulator control device includes: an acquisition module 100, used to acquire a first task, and determine a first neural network according to the first task, wherein the control unit is configured with multiple pre-trained neural networks, and the first neural network is a neural network in the neural network that matches the first task.

[0094] The reconstruction module 200 is used to reconstruct the first task by using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions.

[0095] The control module 300 is used to obtain control parameters for controlling the manipulator based on the control strategy, and control the manipulator to perform control actions according to the control parameters, wherein the control parameters include parameters corresponding to the basic control instructions.

[0096] The manipulator control device provided in the embodiment of the present application and the manipulator control method provided in the embodiment of the present application obtain a first task through the control unit of the manipulator, select and determine a first neural network according to the first task, the first neural network is a neural network matching the first task among multiple pre-trained neural networks configured in the control unit, and reconstruct the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions, and control parameters for controlling the manipulator are obtained based on the series of basic control instructions, and finally the manipulator is controlled according to the control parameters to perform control actions and complete corresponding movements. Since the method uses the neural network configured in the manipulator control unit to complete spatial position control, it can adapt well to application scenarios without networking and does not require additional consumption of computing resources of the control unit itself, and can effectively solve the problem of limited computing power of the manipulator during spatial position control; and using neural networks, self-learning of manipulator control can also be achieved to meet the flexible and changeable control needs of the manipulator.

[0097] In some embodiments, the acquisition module is further used to acquire the first task and the environmental status information corresponding to the first task, wherein the environmental status information includes the surrounding environment information and the self-state information of the manipulator in the simulation environment;

[0098] The reconstruction module is further used to input the surrounding environment information, the self-state information and the first task into the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes the basic control instructions and the scene control instructions.

[0099] In some embodiments, the control module is also used to obtain control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions; predict the spatial pre-operation trajectory of the manipulator according to the control parameters; if it is determined by the scene parameters that there are no operating obstacles to the spatial pre-operation trajectory, control the manipulator to operate according to the spatial pre-operation trajectory; if it is determined by the scene parameters that there are operating obstacles to the spatial pre-operation trajectory, obtain the spatially reconstructed operating trajectory of the manipulator according to the control parameters and the scene parameters, and control the manipulator to operate according to the spatially reconstructed operating trajectory.

[0100] In some embodiments, the control module is also used to add or delete the control parameters according to the scene parameters to obtain reconstructed control parameters, and obtain the spatially reconstructed operation trajectory according to the reconstructed control parameters, and control the manipulator to operate according to the spatially reconstructed operation trajectory; if there are still operation obstacles based on the spatially reconstructed operation trajectory obtained based on the reconstructed control parameters, an alarm message indicating that the first task cannot be completed is issued.

[0101] In some embodiments, the control module is also used to obtain control parameters and scene parameters for controlling the manipulator based on the control strategy, where the scene parameters are parameters corresponding to the scene control instructions; based on a similarity scene measurement algorithm, corresponding historical control parameters in historical similar scenes are obtained according to the scene parameters; the control parameters are reconstructed according to the historical control parameters to obtain reconstructed control parameters, and a spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

[0102] In some embodiments, each of the control parameters corresponds to a basic action, and the control module is also used to obtain the basic actions of the manipulator and the execution order of each basic action according to the reconstructed control parameters, and control the manipulator to execute the basic actions in sequence according to the execution order to complete the operation of the spatial reconstructed operation trajectory.

[0103] In some embodiments, the model training module, based on a reinforcement learning algorithm, uses the control strategy to train the first neural network, obtains feedback parameters and adjusts the first neural network according to the feedback parameters to obtain an updated first neural network.

[0104] The manipulator control device in the embodiment of the present application can be an electronic device, or a component in the electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or it can be other devices other than a terminal. Exemplarily, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, a vehicle-mounted electronic device, a mobile Internet device (Mobile Internet Device, MID), an augmented reality (augmented reality, AR) / virtual reality (virtual reality, VR) device, a robot, a wearable device, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a netbook or a personal digital assistant (personal digital assistant, PDA), etc. It can also be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., and the embodiment of the present application is not specifically limited.

[0105] The manipulator control device in the embodiment of the present application may be a device having an operating system. The operating system may be a Microsoft (Windows) operating system, an Android (Android) operating system, an IOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0106] The robot control device provided in the embodiment of the present application can achieve Figure 2 To avoid repetition, the various processes implemented in the illustrated method embodiment will not be described again here.

[0107] The robot control method provided in the embodiment of the present application can be executed by an electronic device or a functional module or functional entity in the electronic device that can implement the control method. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras and wearable devices. The control method provided in the embodiment of the present application is described below using the electronic device as an example of the execution subject.

[0108] In some embodiments, Figure 4As shown, an embodiment of the present application also provides an electronic device 800, including a processor 801, a memory 802, and a computer program stored in the memory 802 and executable on the processor 801. When the program is executed by the processor 801, each process of the above-mentioned manipulator control method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0109] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0110] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned robot control method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0111] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0112] An embodiment of the present application also provides a computer program product, including a computer program, which implements the above-mentioned manipulator control method when executed by a processor.

[0113] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0114] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned robot control method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0115] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0116] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0117] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0118] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0119] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0120] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A control method for a manipulator, characterized in that: Applied to the control unit of the manipulator, the method comprises: Acquire a first task, and determine a first neural network according to the first task, wherein the control unit is configured with a plurality of pre-trained neural networks, and the first neural network is a neural network among the neural networks that matches the first task; Reconstructing the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions; Based on the control strategy, control parameters for controlling the manipulator are acquired, and the manipulator is controlled to perform a control action according to the control parameters, wherein the control parameters include parameters corresponding to the basic control instructions.

2. The method according to claim 1, characterized in that The obtaining of the first task comprises: Acquire the first task and the environment status information corresponding to the first task, wherein the environment status information includes the surrounding environment information and the self status information of the manipulator in the simulation environment; The step of reconstructing the first task using a pre-trained first neural network to obtain a control strategy for the reconstructed first task includes: The surrounding environment information, the self-state information and the first task are input into the first neural network to obtain a reconstructed control strategy for the first task, wherein the control strategy includes the basic control instructions and the scene control instructions.

3. The method according to claim 2, characterized in that The acquiring control parameters for controlling the manipulator based on the control strategy, and controlling the manipulator to perform a control action according to the control parameters, comprises: Acquire control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions; Predicting the spatial pre-operation trajectory of the manipulator according to the control parameters; When it is determined by using the scene parameters that there are no running obstacles in the spatial pre-operation trajectory, the manipulator is controlled to run according to the spatial pre-operation trajectory.

4. The method according to claim 3, characterized in that If it is determined by using the scene parameters that there are running obstacles in the spatial pre-running trajectory, the spatial reconstruction running trajectory of the manipulator is obtained according to the control parameters and the scene parameters, and the manipulator is controlled to run according to the spatial reconstruction running trajectory.

5. The method according to claim 4, characterized in that The step of obtaining the spatial reconstruction operation trajectory of the manipulator according to the control parameters and the scene parameters, and controlling the manipulator to operate according to the spatial reconstruction operation trajectory includes: The control parameters are added or deleted according to the scene parameters to obtain reconstructed control parameters, and the spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

6. The method according to claim 5, characterized in that If the spatially reconstructed operation trajectory obtained based on the reconstructed control parameters still has operation obstacles, an alarm message indicating that the first task cannot be completed is issued.

7. The method according to claim 2, characterized in that The acquiring control parameters for controlling the manipulator based on the control strategy, and controlling the manipulator to perform a control action according to the control parameters, comprises: Acquire control parameters and scene parameters for controlling the manipulator based on the control strategy, wherein the scene parameters are parameters corresponding to the scene control instructions; Based on the similarity scene measurement algorithm, the corresponding historical control parameters in the historical similar scenes are obtained according to the scene parameters; The control parameters are reconstructed according to the historical control parameters to obtain reconstructed control parameters, and a spatially reconstructed operation trajectory is obtained according to the reconstructed control parameters, and the manipulator is controlled to operate according to the spatially reconstructed operation trajectory.

8. The method according to claim 5 or 7, characterized in that: Each of the control parameters corresponds to a basic action, and obtaining the spatial reconstruction operation trajectory according to the reconstructed control parameters, and controlling the manipulator to operate according to the spatial reconstruction operation trajectory, includes: The basic actions of the manipulator and the execution order of each basic action are obtained according to the reconstructed control parameters, and the manipulator is controlled to execute the basic actions in sequence according to the execution order to complete the operation of the spatially reconstructed operation trajectory.

9. A robot control device, characterized in that: The device comprises: an acquisition module, configured to acquire a first task, and determine a first neural network according to the first task, wherein the control unit is configured with a plurality of pre-trained neural networks, and the first neural network is a neural network among the neural networks that matches the first task; A reconstruction module, used to reconstruct the first task using the first neural network to obtain a control strategy for the reconstructed first task, wherein the control strategy includes a series of basic control instructions; A control module is used to obtain control parameters for controlling the manipulator based on the control strategy, and control the manipulator to perform control actions according to the control parameters, wherein the control parameters include parameters corresponding to the basic control instructions.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the robot control method as described in any one of claims 1-8 is implemented.

Images