A knowledge graph-based software gripper control method and related device

Abstract

This invention relates to the field of robotic arm technology, specifically to a knowledge graph-based soft gripper control method and related device. Before using a robotic arm composed of various soft grippers to manipulate a target object, this invention first checks the air pressure of each soft gripper to remove those with insufficient air pressure, leaving only those with adequate air pressure. Then, it searches the knowledge graph for the claw shape required for a valid gripper to manipulate a target object with a specific appearance. Finally, it controls the valid gripper to manipulate the target object with the claw shape. From the above analysis, it can be seen that this invention, by removing soft grippers with low air pressure before using them to manipulate an object, and allowing only those with adequate air pressure to participate in the manipulation of the target object, ensures precise control of the target object through the soft gripper.

Description

Technical Field

[0001] This invention relates to the field of robotic arm technology, specifically to a knowledge graph-based software gripper control method and related devices. Background Technology

[0002] The soft gripper mainly consists of three air bladders arranged side by side, with the central air bladder compressed by the air bladders on either side. As an end effector for robots, the soft gripper can simulate human gripping actions and adapt to the grasping and handling of objects of various shapes. Soft grippers have been widely used in industrial automation, medical and health care, rescue and disaster relief, and other fields. Current technology directly uses soft grippers to grasp objects; however, if the air pressure in the soft gripper is low, it will be difficult for the soft gripper to accurately manipulate the object.

[0003] In summary, existing technologies using soft grippers make it difficult to precisely manipulate objects.

[0004] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a knowledge graph-based software gripper control method, which solves the problem of difficulty in accurately manipulating objects using existing software grippers.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a software gripper control method based on knowledge graphs, comprising:

[0008] Detect the gripper air pressure of each software gripper, and select the non-failed grippers from among the software grippers based on the gripper air pressure;

[0009] Obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph;

[0010] Control the unfailed gripper to manipulate the target object in the claw shape.

[0011] In one implementation, the soft gripper includes a first airbag and a second airbag compressed between the first airbags. The step of detecting the gripper air pressure of each soft gripper and selecting non-failed grippers from among the soft grippers based on the gripper air pressure includes:

[0012] The sensor is detected as being disabled, the disabling being used to characterize the sensor's failure to detect air pressure, the sensor being located in the second airbag;

[0013] When the sensor is not disabled, the range of air pressure changes collected by the sensor is continuously monitored;

[0014] Based on the air pressure change range, select the non-disabled first airbags from each of the first airbags of each of the soft grippers;

[0015] The soft gripper containing the non-disabled first airbag is considered as an unfailed gripper.

[0016] In one implementation, the step of selecting non-disabled first airbags from each of the first airbags of each of the various soft grippers based on the air pressure variation range includes:

[0017] Obtain the standard range of air pressure change corresponding to the target object, wherein the standard range of air pressure change is the air pressure collected by the sensor when the soft gripper successfully grasps the target object;

[0018] Based on the air pressure change range and the air pressure change standard range, non-disabled first airbags are selected from each of the first airbags of each of the soft grippers.

[0019] In one implementation, obtaining the appearance information of the target object to be grasped and determining the claw shape corresponding to the appearance information based on a knowledge graph includes:

[0020] Obtain the object shape and / or object size from the appearance information;

[0021] Based on the object shape and / or object size and the number of non-failed grippers, and using the knowledge graph, target grippers are selected from the non-failed grippers, and the claw shape of the target grippers is determined.

[0022] In one implementation, the knowledge graph is constructed in the following ways:

[0023] Determine the gripper combination consisting of the individual software grippers;

[0024] By applying a convolutional neural network model to the distribution information of each soft gripper and the gripper combination, various predetermined claw shapes formed by the grippers of the gripper combination are obtained.

[0025] A convolutional neural network model is applied to the object's appearance information and various predetermined claw shapes to filter out the preferred claw shape for the object from the various predetermined claw shapes;

[0026] A knowledge graph is constructed using the object's appearance information and the corresponding preferred claw shape.

[0027] In one implementation, controlling the non-failed gripper to manipulate the target object in the claw shape includes:

[0028] Obtain the object position of the target object;

[0029] Control the undamaged gripper to move to the object's location and manipulate the target object in a claw shape.

[0030] One implementation also includes:

[0031] Monitor whether the target object is in a stable state when the non-failed gripper manipulates the target object in the claw shape;

[0032] When the target object is not in a stable state, adjust the air pressure of the non-disabled first airbag of the non-failed gripper until the target object is in a stable state.

[0033] When the target object is in a stable state, the air pressure of the non-disabled first airbag is collected and recorded as the update air pressure;

[0034] The pressure change standard range is updated with the updated pressure.

[0035] Secondly, embodiments of the present invention also provide a knowledge graph-based software gripper control device, wherein the device comprises the following components:

[0036] The disabled gripper screening module is used to detect the gripper air pressure of each software gripper and screen out the non-failed grippers from each software gripper based on the gripper air pressure;

[0037] The claw shape generation module is used to obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph;

[0038] A control module is used to control the unfailed gripper to manipulate the target object in the claw shape.

[0039] Thirdly, embodiments of the present invention also provide a terminal device, wherein the terminal device includes a memory, a processor, and a knowledge graph-based software gripper control program stored in the memory and executable on the processor. When the processor executes the knowledge graph-based software gripper control program, it implements the steps of the knowledge graph-based software gripper control method described above.

[0040] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a knowledge graph-based software gripper control program. When the knowledge graph-based software gripper control program is executed by a processor, it implements the steps of the knowledge graph-based software gripper control method described above.

[0041] Beneficial Effects: Before using a robotic arm composed of various soft grippers to manipulate a target object, this invention first checks the air pressure of each soft gripper to remove those with insufficient air pressure (soft grippers with low internal air pressure), leaving only the soft grippers with adequate air pressure (undamaged grippers). Then, it searches a knowledge graph for the claw shape required for the undamaged grippers to manipulate a target object with a specific appearance. Finally, it controls the undamaged grippers to manipulate the target object in a claw shape. From the above analysis, it can be seen that this invention, by removing soft grippers with low air pressure before using them to manipulate an object, and only allowing the soft grippers with adequate air pressure to participate in the manipulation of the target object, can ensure precise manipulation of the target object through the soft grippers. Attached Figure Description

[0042] Figure 1 This is an overall flowchart of the present invention;

[0043] Figure 2 This is a structural diagram of the robotic arm in an embodiment of the present invention;

[0044] Figure 3 This is a flowchart of the software gripper failure determination process in an embodiment of the present invention;

[0045] Figure 4 This is a diagram of the R-CNN structure in an embodiment of the present invention;

[0046] Figure 5 This is a schematic diagram illustrating the use of knowledge graphs to capture objects in an embodiment of the present invention;

[0047] Figure 6 This is a schematic diagram of fault-tolerant control—knowledge graph reasoning in an embodiment of the present invention;

[0048] Figure 7 This is a block diagram illustrating the internal structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0050] Research has revealed that soft grippers primarily consist of three side-by-side air bladders, with the central bladder compressed by the air bladders on either side. As an end effector for robots, the soft gripper can simulate human gripping movements, adapting to the grasping and handling of objects of various shapes. Soft grippers are widely used in industrial automation, healthcare, and disaster relief. However, current technologies that directly use soft grippers to grasp objects suffer from limitations; low air pressure can hinder precise object manipulation.

[0051] To address the aforementioned technical problems, this invention provides a knowledge graph-based soft gripper control method, solving the issue of precise object manipulation using existing soft grippers. Specifically, the method first detects the gripper air pressure of each soft gripper, and then selects the compliant grippers based on this pressure. Next, it acquires the appearance information of the target object to be grasped, and determines the corresponding claw shape based on the knowledge graph. Finally, it controls the compliant grippers to manipulate the target object using the claw shape. This invention enables precise manipulation of target objects using compliant grippers.

[0052] For example, soft robotic arms, such as Figure 2 As shown, the robotic arm includes six soft grippers (inflatable grippers), numbered 1, 2, 3, 4, 5, and 6 sequentially in a circular pattern. Before using the soft robotic arm to grasp a target object, the air pressure of these six soft grippers is first tested. If the air pressure of soft grippers 2 and 4 is very low, they are considered faulty or disabled grippers. Conversely, if the air pressure of soft grippers 1, 3, 5, and 6 is stable and sufficient to grasp the target object, then these four soft grippers are considered valid grippers. The grippers 1, 3, 5, and 6 can be used in combination (the structure of the robotic arm dictates that soft grippers 1, 3, 5, and 6 cannot be used simultaneously; the preferred combination is grippers 1, 3, and 5). Then, the appearance information of the target object (such as shape and size) is combined with the knowledge graph (which records the correspondence between gripper combination, object appearance, and claw shape) to determine which claw shape the soft grippers 1, 3, and 5 should use to grasp the target object.

[0053] Exemplary methods

[0054] The knowledge graph-based software gripper control method of this embodiment can be applied to terminal devices, which can be terminal products with mechanical control functions, such as computers. In this embodiment, as... Figure 1 As shown, the knowledge graph-based software gripper control method specifically includes the following steps:

[0055] S100, detect the gripper air pressure of each software gripper, and select non-failed grippers from each software gripper based on the gripper air pressure.

[0056] The soft gripper consists of two first airbags and one second airbag. The second airbag is compressed between the two first airbags and contains sensors to monitor the air pressure inside. The first airbags act as actuators, undertaking most of the work of grasping objects, and the actuators perform the grasping action through pneumatic drive.

[0057] In one embodiment, step S100 includes the following steps S101 to S105:

[0058] S101, detect whether the sensor is disabled, the disabling is used to characterize the sensor's failure to detect air pressure, the sensor is located in the second airbag.

[0059] First, determine the status of the soft gripper's sensors. If the sensors are faulty, meaning they cannot detect the internal air pressure, then the entire soft gripper is considered disabled.

[0060] S102, when the sensor is not disabled, continuously detect the range of air pressure changes collected by the sensor.

[0061] If the sensor is not disabled, the system continues to determine whether the first airbag is disabled by measuring the pressure change range of the second airbag collected by the sensor.

[0062] S103, obtain the standard range of air pressure change corresponding to the target object, wherein the standard range of air pressure change is the air pressure collected by the sensor when the soft gripper successfully grasps the target object.

[0063] The standard range of air pressure change corresponds one-to-one with the shape and size of the target object.

[0064] S104, based on the air pressure change range and the air pressure change standard range, select the non-disabled first airbags from each of the first airbags of each of the soft grippers.

[0065] If the sensor is intact, the air pressure information is fed back normally, and the status of the actuator is then determined. The actuator failure is determined by the air pressure change over the feedback period: if the air pressure fluctuates greatly and the actuator's air pressure is below the standard range before it contacts the target object, the actuator (first airbag) is determined to be faulty, and thus the finger (soft gripper) is determined to be faulty; if the air pressure is stable and the actuator's air pressure is above the standard range before it contacts the target object, the actuator is determined to be intact, and thus the finger is intact and has the ability to perform actions.

[0066] S105, the soft gripper where the non-disabled first airbag is located is regarded as a non-failed gripper.

[0067] The software gripper whose first airbag is not disabled is determined to be the non-failed gripper.

[0068] by Figure 3 For example, let's explain how to determine if a software gripper has failed:

[0069] A 30 kPa air pressure is input to sensors S1-S6 (i.e., the six sensors inside the six sets of second airbags on the six soft grippers). The air pressure corresponding to each of the six sets of sensors (S1-S6) is checked to see if it is less than 10 kPa. (For example, when checking if the air pressure corresponding to one of the actuators is less than 10 kPa, since the second airbag containing the sensor is compressed by the first airbag, the air pressure collected by the sensor inside the second airbag can reflect the air pressure of the first airbag.) Soft grippers containing sensors with pressure less than 10 kPa are determined to be faulty grippers. For software grippers with sensors of not less than 10 kPa, further determine whether the air pressure of the actuator (first airbag) is less than 35 kPa (35 kPa is obtained through the sensor inside the second airbag. For example, when determining whether the air pressure corresponding to one set of actuators is less than 35 kPa, since the second airbag where the sensor is located is compressed by the first airbag, the air pressure collected by the sensor inside the second airbag can reflect the air pressure of the first airbag. Therefore, the air pressure of the first airbag can be obtained through the air pressure collected by the sensor). Software grippers with actuators of less than 35 kPa are determined to be faulty grippers; otherwise, they are determined to be non-faulty grippers.

[0070] S200: Obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph.

[0071] In one embodiment, as shown below Figure 4 The R-CNN convolutional neural network shown identifies the appearance information of the target object, including object shape classification (including cube, cylinder, sphere), object size classification (including large, medium, small), and information about the object's surrounding environment.

[0072] In this embodiment, the R-CNN convolutional neural network is a trained convolutional neural network, and the training process is as follows:

[0073] Image of a sample object is acquired using a visual sensor and input into an R-CNN model. Candidate regions are selected using bounding boxes, and the bounding boxes are redefined to a fixed size before being input into the CNN model to obtain a feature vector. This feature vector is then input into an SVM classifier for classification and recognition of the sample object image (classifying based on the object's shape). Simultaneously, a boundary regression model corrects the position of the bounding boxes. The CNN model classifies the objects within the corrected bounding boxes, and the classification result output by the CNN model is obtained. This classification result is compared with the true classification result of the sample object image. If the two are different, the parameters of the CNN model are adjusted, and finally, the trained R-CNN model is established.

[0074] In another embodiment, the specific process of step S200 is as follows: obtain the object shape and / or object size in the appearance information; based on the object shape and / or object size and the number of non-failed grippers, select target grippers from the non-failed grippers and determine the claw shape of the target grippers based on the knowledge graph.

[0075] For example, if the non-disabled grippers are numbered 1, 3, and 5, and the target object is cylindrical and relatively small, then the non-disabled grippers 1, 3, and 5 will be controlled to converge into a circle to grasp the target object.

[0076] by Figure 5 For example, let's illustrate how knowledge graphs can be used to manipulate software grippers to grasp objects:

[0077] First, the target object's type, size, and currently active grippers are input into the knowledge graph. Then, the knowledge graph provides actions that match the object's type and size. For example, if the active grippers are 1, 2, 3, 4, 5, and 6, the knowledge graph will provide combinations of 1, 3, and 5 or 2, 4, and 6 (these two combinations are the first choice, hence the knowledge graph will provide these two combinations). Then, grippers 1, 3, and 5 or grippers 2, 4, and 6 will grasp the object. If the grasp is not firm during the grasping process, the air pressure of the gripper's first airbag will be adjusted using an air pump. The air pressure corresponding to a firm grasp will be used to update the standard range of air pressure changes required for grasping the object recorded in the knowledge graph.

[0078] In one embodiment, the construction of the knowledge graph includes the following steps S201, S202, S203, and S204:

[0079] S201, determine the gripper combination consisting of each software gripper.

[0080] S202, apply a convolutional neural network model to the distribution information of each soft gripper and the gripper combination to obtain various predetermined claw shapes formed by the grippers of the gripper combination.

[0081] For example, if the soft grippers are 1, 2, 3, 4, 5, and 6, possible gripper combinations include 1 and 3, 3 and 5, 5 and 6, 1 and 3 and 5, 3 and 5 and 6, 3 and 6, 1 and 5, 4 and 6, etc. Figure 6 As shown, the above gripper combinations and gripper distribution information (i.e., which grippers) are input into the ConvE model (convolutional neural network model), and the ConvE model outputs various predetermined claw shapes. Figure 6 (The available claw shapes in the model). For example, if you input claw combinations 1, 3, and 5 into the ConvE model, the ConvE model will output predefined claw shapes such as circles and triangles.

[0082] S203, apply a convolutional neural network model to the object's appearance information and the various predetermined claw shapes to filter out the preferred claw shape for the object from the various predetermined claw shapes.

[0083] If the object is a cylinder, then by inputting the appearance information of the cylinder and the predetermined claw shapes such as circles and triangles into the convolutional neural network model, the model will output the preferred claw shape, which is a circle, that matches the cylinder.

[0084] S204. Construct a knowledge graph using the object's appearance information and the preferred claw shape corresponding to the object's appearance information.

[0085] A knowledge graph is used to record the correspondence between the appearance information of an object and the preferred claw shape. In other words, the knowledge graph records what kind of appearance an object needs what kind of claw shape to grasp.

[0086] Knowledge graphs enable machines to better understand and utilize knowledge. The information acquired during the grasping process of a soft gripper includes: the gripper's status information (whether it is malfunctioning), the attribute information of the target to be grasped, and real-time air pressure monitoring information. By constructing a knowledge graph, different types of information, such as image information and text information, can be standardized, stored, and managed. The unstructured knowledge required for constructing a knowledge graph is often processed using deep learning techniques. Deep learning techniques use multi-layer neural networks to automatically learn and extract features from data, achieving end-to-end processing of unstructured knowledge.

[0087] S300, control the non-failed gripper to manipulate the target object in the claw shape.

[0088] The pneumatic control system moves the robotic arm to the location of the target object. Then, the pneumatic control system controls the intact grippers on the robotic arm to manipulate the target object using a claw shape provided by the knowledge graph. During the object grasping process, the pneumatic control system also controls the gripper's pose to complete the grasping operation.

[0089] In one embodiment, step S300 is performed as follows: obtaining the object position of the target object; controlling the non-failed gripper to move to the object position and manipulating the target object in the claw shape.

[0090] In one embodiment, the standard range of air pressure change set for the object in the knowledge graph is updated based on the situation that occurs during the manipulation of the target object by the non-failed gripper. The specific update process includes the following steps S401 to S404:

[0091] S401, monitor whether the target object is in a stable state when the non-failed gripper manipulates the target object in the claw shape.

[0092] S402, when the target object is not in a stable state, adjust the air pressure of the non-disabled first airbag of the non-failed gripper until the target object is in a stable state.

[0093] S403, when the target object is in a stable state, the air pressure of the non-disabled first airbag is collected and recorded as the update air pressure.

[0094] S404, update the standard range of pressure changes with the updated pressure.

[0095] In this embodiment, the vision sensor monitors in real time during the grasping process. If the grasping fails or becomes unstable, the grasped image is fed back to the knowledge graph in real time. After receiving the feedback information, the knowledge graph confirms the current air pressure status of the gripper and adjusts the air pressure based on the ConvE model, according to the grasping target object and the grasping effect, to ensure stable grasping.

[0096] Exemplary device

[0097] This embodiment also provides a knowledge graph-based software gripper control device, which includes the following components:

[0098] The disabled gripper screening module is used to detect the gripper air pressure of each software gripper and screen out the non-failed grippers from each software gripper based on the gripper air pressure;

[0099] The claw shape generation module is used to obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph;

[0100] A control module is used to control the unfailed gripper to manipulate the target object in the claw shape.

[0101] Based on the above embodiments, the present invention also provides a terminal device, the principle block diagram of which can be as follows: Figure 7 As shown, the terminal device includes a processor, memory, network interface, display screen, and temperature sensor connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a knowledge graph-based software gripper control method. The display screen can be an LCD screen or an e-ink screen. The temperature sensor is pre-installed inside the terminal device to detect the operating temperature of the internal components.

[0102] Those skilled in the art will understand that Figure 7 The block diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0103] In one embodiment, a terminal device is provided, comprising a memory, a processor, and a knowledge graph-based software gripper control program stored in the memory and executable on the processor. When the processor executes the knowledge graph-based software gripper control program, it implements the following operation instructions:

[0104] Detect the gripper air pressure of each software gripper, and select the non-failed grippers from among the software grippers based on the gripper air pressure;

[0105] Obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph;

[0106] Control the unfailed gripper to manipulate the target object in the claw shape.

[0107] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A software gripper control method based on knowledge graphs, characterized in that, include: Detect the gripper air pressure of each software gripper, and select the non-failed grippers from among the software grippers based on the gripper air pressure; Obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph; Control the intact gripper to manipulate the target object in the claw shape; The soft gripper includes a first airbag and a second airbag compressed between the first airbags. The step of detecting the gripper air pressure of each soft gripper and selecting non-failed grippers from among the soft grippers based on the gripper air pressure includes: The sensor is detected as being disabled, the disabling being used to characterize the sensor's failure to detect air pressure, the sensor being located in the second airbag; When the sensor is not disabled, the range of air pressure changes collected by the sensor is continuously monitored; Obtain the standard range of air pressure change corresponding to the target object, wherein the standard range of air pressure change is the air pressure collected by the sensor when the soft gripper successfully grasps the target object; Based on the air pressure change range and the air pressure change standard range, select the non-disabled first airbags from each of the first airbags of each of the soft grippers; The soft gripper containing the non-disabled first airbag is considered as an unfailed gripper.

2. The knowledge graph-based software gripper control method as described in claim 1, characterized in that, The step of acquiring the appearance information of the target object to be grasped, and determining the claw shape corresponding to the appearance information based on a knowledge graph, includes: Obtain the object shape and / or object size from the appearance information; Based on the object shape and / or object size and the number of non-failed grippers, and using the knowledge graph, target grippers are selected from the non-failed grippers, and the claw shape of the target grippers is determined.

3. The knowledge graph-based software gripper control method as described in claim 1, characterized in that, The knowledge graph is constructed in the following ways: Determine the gripper combination consisting of the individual software grippers; By applying a convolutional neural network model to the distribution information of each soft gripper and the gripper combination, various predetermined claw shapes formed by the grippers of the gripper combination are obtained. A convolutional neural network model is applied to the object's appearance information and various predetermined claw shapes to filter out the preferred claw shape for the object from the various predetermined claw shapes; A knowledge graph is constructed using the object's appearance information and the corresponding preferred claw shape.

4. The knowledge graph-based software gripper control method as described in claim 1, characterized in that, The control of the non-failed gripper to manipulate the target object in the claw shape includes: Obtain the object position of the target object; Control the undamaged gripper to move to the object's location and manipulate the target object in a claw shape.

5. The knowledge graph-based software gripper control method as described in claim 1, characterized in that, Also includes: Monitor whether the target object is in a stable state when the non-failed gripper manipulates the target object in the claw shape; When the target object is not in a stable state, adjust the air pressure of the non-disabled first airbag of the non-failed gripper until the target object is in a stable state. When the target object is in a stable state, the air pressure of the non-disabled first airbag is collected and recorded as the update air pressure; The pressure change standard range is updated with the updated pressure.

6. A knowledge graph-based software gripper control device, characterized in that, The device comprises the following components: The disabled gripper screening module is used to detect the gripper air pressure of each software gripper and screen out the non-failed grippers from each software gripper based on the gripper air pressure; The claw shape generation module is used to obtain the appearance information of the target object to be grasped, and determine the claw shape corresponding to the appearance information based on the knowledge graph; The control module is used to control the non-failed gripper to manipulate the target object in the claw shape; The soft gripper includes a first airbag and a second airbag compressed between the first airbags. The step of detecting the gripper air pressure of each soft gripper and selecting non-failed grippers from among the soft grippers based on the gripper air pressure includes: The sensor is detected as being disabled, the disabling being used to characterize the sensor's failure to detect air pressure, the sensor being located in the second airbag; When the sensor is not disabled, the range of air pressure changes collected by the sensor is continuously monitored; Obtain the standard range of air pressure change corresponding to the target object, wherein the standard range of air pressure change is the air pressure collected by the sensor when the soft gripper successfully grasps the target object; Based on the air pressure change range and the air pressure change standard range, select the non-disabled first airbags from each of the first airbags of each of the soft grippers; The soft gripper containing the non-disabled first airbag is considered as an unfailed gripper.

7. A terminal device, characterized in that, The terminal device includes a memory, a processor, and a knowledge graph-based software gripper control program stored in the memory and executable on the processor. When the processor executes the knowledge graph-based software gripper control program, it implements the steps of the knowledge graph-based software gripper control method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a knowledge graph-based software gripper control program, which, when executed by a processor, implements the steps of the knowledge graph-based software gripper control method as described in any one of claims 1-5.

Images