Motion control system and method for mechanical arm of metro vehicle bottom inspection robot

Abstract

The invention discloses a motion control system and method for a mechanical arm of a metro underbody inspection robot, and the system comprises an obtaining module which is used for obtaining the pose information collected by at least one sensing device, the pose information comprises the pose information of a target object, and a control module which is used for employing a preset initial mechanical arm control model. The standard Euler attitude angle of the current operation point is obtained by the standard attitude angle obtaining module when the next operation inflection point is used as the moving target of the end tool under the robot coordinate system, and then the acting force acting on the end tool at the current operation point is detected in real time by the force detection module; then the acting force is decomposed into three angle directions of the standard Euler attitude angle, an acting component force facing the next operation inflection point is selected, and then the tail end tool is controlled to move to the next operation inflection point based on the acting component force.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robot technology, in particular to a motion control system and method of a subway car bottom inspection robot mechanical arm. BACKGROUND

[0002] The subway intelligent inspection robot is an intelligent maintenance device. The robot integrates advanced technologies such as walking robot, multi-degree-of-freedom mechanical arm, AI image recognition, and can autonomously complete panoramic scanning, three-dimensional identification and abnormality judgment of key components of the car bottom. Compared with traditional manual inspection, its work efficiency is expected to increase by more than 30%, and it can assist in completing more than 65% of the off-car detection tasks. Significantly improve the digital level of rail transit maintenance, remote control mechanical arm has become the primary choice to replace manual work, and the effective operation of the control end to the mechanical arm can be realized by using a local area network, so the management and operation are convenient, and it can adapt to many scenes that humans cannot easily handle, and has the advantage of higher efficiency than manual work. However, in actual application, when the motion state of the mechanical arm with redundant degrees of freedom is determined, there are countless kinematic inverse solutions, so when the mechanical arm imitates the action, there may be many cases of the action of the specific object and the mechanical arm itself, and the motion planning of the mechanical arm may have uncertainty, resulting in a relatively complex calculation process and a long time-consuming, making it difficult to meet the interactive needs of the control of the mechanical arm. SUMMARY

[0003] The purpose of the present application is to provide a motion control system and method of a subway car bottom inspection robot mechanical arm to solve the problems raised in the background art.

[0004] To achieve the above purpose, the present application provides the following technical scheme: a motion control system of a subway car bottom inspection robot mechanical arm, the system comprising an acquisition module for acquiring pose information collected by at least one sensing device, the pose information comprising information about the pose of a target object.

[0005] A control module is used to control the model of a preset mechanical arm in a work environment to perform multiple interactive tasks on the model of a preset workpiece in the work environment using a preset initial mechanical arm control model, and generate multiple interactive trajectory data.

[0006] Each interactive trajectory data is the trajectory data in an interactive task performed by the model of the preset mechanical arm.

[0007] A target interface module is used to obtain a target interface corresponding to at least one function class included in a mechanical arm control program, and expose the target interface corresponding to the at least one function class.

[0008] The target interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional classes is used to control the target robotic arm to achieve the corresponding function.

[0009] Preferably, the system further includes a data acquisition module for sampling multiple transfer data from each interaction trajectory data, wherein the multiple transfer data includes process data at multiple time points when the virtual model of the preset robotic arm performs the same interaction task.

[0010] The information acquisition module is used to acquire target custom programs and / or target library files that meet the user's robotic arm control requirements after the target interface is exposed.

[0011] The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inherited and implemented target plugin interface.

[0012] The mapping module is used to obtain the mapping relationship between the pose information and the joints in the robotic arm.

[0013] Preferably, the system further includes a determination module for determining a set of candidate playback targets for the plurality of time points from the plurality of transfer data.

[0014] The candidate playback target set for each time point includes: the completed target position of the model of the preset workpiece at at least one time point after each time point.

[0015] The program pointer determination module is used to obtain the instantiation pointer corresponding to the target custom program based on the target custom program and / or the target library file.

[0016] Preferably, the system further includes an update module, used to update the completed target positions in the transfer data corresponding to each time point according to the candidate replay target set at each time point, so as to obtain an updated transfer data.

[0017] The robotic arm control module is used to run the target custom program based on the instantiation pointer corresponding to the target custom program in order to control the target robotic arm.

[0018] Preferably, the acquisition module is further configured to sample the interactive observation data of the preset robotic arm model at multiple time points from each interactive trajectory data.

[0019] The motion data of the preset robotic arm model at multiple time points are sampled from each of the interaction trajectory data.

[0020] The motion data includes the pose of the model of the preset workpiece and the pose of the end effector of the model of the preset robotic arm.

[0021] Reward data for completing the same interactive task at the multiple time points is sampled from each interaction trajectory data.

[0022] The target location of the same interaction task at the multiple time points is sampled from each interaction trajectory data.

[0023] The completed target positions at the plurality of time points are sampled from each of the interaction trajectory data.

[0024] Based on the interactive observation data, action data, reward data, target location, and completed target location at the multiple time points, the multiple transfer data are generated respectively.

[0025] Preferably, multiple sample pairs are obtained, each sample pair including a sample image and environmental state information corresponding to the sample image.

[0026] The sample images in each sample pair are enhanced to obtain corresponding enhanced sample images, and the enhanced sample images are input into the initial robotic arm control student model to obtain the first control action information.

[0027] The environmental state information from the sample pairs is input into the robotic arm control teacher model to obtain the second control action information.

[0028] Based on the first and second control action information corresponding to each sample pair, the model parameters of the initial robotic arm control student model are updated to obtain the robotic arm control student model.

[0029] Acquire pose information collected by at least one sensing device, the pose information including information about the pose of a target object.

[0030] Obtain the mapping relationship between the pose information and the joints in the robotic arm.

[0031] The current arm angle value of the robotic arm is determined based on the pose information and the mapping relationship.

[0032] Based on the relationship between the joint angle and arm angle of each of the M joints, the pose information, and the current arm angle value, the target angle value of each joint is obtained.

[0033] Obtain the target interface corresponding to at least one functional class contained in the robotic arm control program, and expose the target interface corresponding to the at least one functional class.

[0034] The target plug-in interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional classes is used to control the target robotic arm to achieve the corresponding function.

[0035] After exposing the target interface, obtain the target custom program and / or target library file that meets the user's robotic arm control requirements.

[0036] The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inherited and implemented target plugin interface.

[0037] Using a preset initial robotic arm control model, the model of the preset robotic arm in the virtual work environment is controlled to perform multiple interactive tasks on the model of the preset workpiece in the work environment, generating multiple interactive trajectory data.

[0038] Each interactive trajectory data is the trajectory data of the model of the preset robotic arm performing an interactive task.

[0039] Preferably, the step of enhancing the sample image in the sample pair to obtain the corresponding enhanced sample image includes inputting the sample image and the background sample image into a pre-constructed context-aware enhancement model, and enhancing the sample image based on the background sample image to obtain the enhanced sample image.

[0040] The context-aware enhancement model is generated based on training with multiple image sample pairs.

[0041] A robotic arm control device includes a memory and a processor.

[0042] The memory is used to store programs.

[0043] The processor is configured to execute the program to implement the various steps of the robotic arm control method as described in any one of claims 6 to 7.

[0044] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the robotic arm control method as described in any one of claims 6 to 7.

[0045] A readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the robotic arm control method as described in any one of claims 6 to 7.

[0046] The motion control system and method for the robotic arm used in subway car undercarriage inspection proposed in this invention have the following advantages:

[0047] This application obtains the standard Euler attitude angle of the end effector at the current working point in the robot coordinate system when the next working inflection point is the target motion point using a standard attitude angle acquisition module. Then, a force detection module detects the force acting on the end effector at the current working point in real time. The force is then decomposed into three angular directions of the standard Euler attitude angle, and a component force towards the next working inflection point is selected. Based on this component force, the end effector is controlled to move to the next working inflection point. This system enables the operation of the work object, such as grinding, solving the shortcomings of traditional collaborative robots such as shaking, offset, and tool tip bounce when encountering hard objects. It also provides a more flexible operating feel. By acquiring pose information collected by at least one sensor, the action to be imitated by the robotic arm can be determined. Based on the corresponding pose information, the arm angle can be used to obtain definite solutions for each joint of the robotic arm, allowing for more reasonable control of the robotic arm's movement. This results in higher computational efficiency when the robotic arm imitates actions based on the pose information from the sensor, and the motion planning of the robotic arm is more reasonable and better meets the requirements of action imitation. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the motion control system and method of a robotic arm for inspecting the undercarriage of a subway car, according to the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see Figure 1 This invention provides a motion control system and method for a robotic arm used for inspecting the undercarriage of a subway car. The system includes an acquisition module for acquiring pose information collected by at least one sensing device, wherein the pose information includes information about the pose of a target object.

[0051] The control module is used to control the model of the preset robotic arm in the working environment to perform multiple interactive tasks on the model of the preset workpiece in the working environment, using a preset initial robotic arm control model, and generating multiple interactive trajectory data.

[0052] Each interactive trajectory data is the trajectory data of the model of the preset robotic arm performing an interactive task.

[0053] The target interface module is used to obtain the target interface corresponding to at least one functional class contained in the robotic arm control program, and expose the target interface corresponding to the at least one functional class.

[0054] The target interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional classes is used to control the target robotic arm to achieve the corresponding function.

[0055] The initial robotic arm control model is a deep neural network with initialized parameters. It can be used to enable a preset robotic arm in a virtual work environment to perform interactive tasks with a virtual model of a preset workpiece in the virtual work environment. However, the probability that the virtual model controlling the preset robotic arm will successfully place the virtual model of the preset workpiece at a preset position is random. Here, the preset position refers to the pre-specified location where the virtual model of the preset workpiece needs to be placed.

[0056] As a preferred embodiment, the system further includes a data acquisition module for sampling multiple transfer data from each interaction trajectory data. The multiple transfer data includes process data at multiple time points when the virtual model of the preset robotic arm performs the same interaction task.

[0057] The information acquisition module is used to acquire target custom programs and / or target library files that meet the user's robotic arm control requirements after the target interface is exposed.

[0058] The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inherited and implemented target plugin interface.

[0059] The mapping module is used to obtain the mapping relationship between the pose information and the joints in the robotic arm.

[0060] The existing robotic arm control program is adjusted according to the functions of each robotic arm contained therein to obtain a robotic arm control program containing at least one functional class. Then, the target plug-in interface corresponding to at least one functional class is obtained. The obtained target plug-in interface is then exposed so that users can call at least one functional class based on the exposed target plug-in interface to write a target custom program that meets the user's robotic arm control requirements.

[0061] Here, the robotic arm control program includes at least one functional class, which is divided according to the control requirements of the target robotic arm, such as forward and inverse kinematics, motion control, peripheral I / O control, robot status acquisition, robot power on / off, and collaborative robot safety control. In this step, each functional class is used to control the robotic arm to achieve the corresponding function.

[0062] As a preferred embodiment, the system further includes a determination module for determining a set of candidate playback targets for the plurality of time points from the plurality of transfer data.

[0063] The candidate playback target set for each time point includes: the completed target position of the model of the preset workpiece at at least one time point after each time point.

[0064] The program pointer determination module is used to obtain the instantiation pointer corresponding to the target custom program based on the target custom program and / or the target library file.

[0065] The robotic arm can have multiple degrees of freedom, such as 4, 5, 6, or 7 degrees of freedom. The redundant degrees of freedom are those beyond the degrees of freedom required for the robotic arm to perform a specified operation. The movement of the robotic arm in each degree of freedom direction can be achieved through joints, and the M joints control the movement of the robotic arm in their respective M degrees of freedom. Each joint can correspond to one degree of freedom. The joints can be connected sequentially, or several joints can be combined into one. The joints can include one or more structural forms such as rotational axes, linear axes, and linkages. There are various ways to arrange the joints, and the structural forms of the joints can also be varied; no limitation is placed here. For example, the robotic arm may have 7 degrees of freedom, in which case the robotic arm may include 7 joints; and the end effector of the robotic arm may have 6 degrees of freedom, such as degrees of freedom for linear motion along the x-axis, degrees of freedom for rotational motion along the x-axis, degrees of freedom for linear motion along the y-axis, degrees of freedom for rotational motion along the y-axis, degrees of freedom for linear motion along the z-axis, and degrees of freedom for rotational motion along the z-axis. In this case, the robotic arm has one more degree of freedom than the degree of freedom required for the end effector of the robotic arm. Therefore, the robotic arm may be a robotic arm with one redundant degree of freedom and a total of seven degrees of freedom.

[0066] As a preferred embodiment, the system further includes an update module, which is used to update the completed target positions in the transfer data corresponding to each time point based on the candidate replay target set at each time point, so as to obtain an updated transfer data.

[0067] The robotic arm control module is used to run the target custom program based on the instantiation pointer corresponding to the target custom program in order to control the target robotic arm.

[0068] The update module is also used to calculate the distance parameters corresponding to each candidate location based on each candidate location at each time point and the target location of the same interaction task; to calculate the similarity parameters corresponding to each candidate location based on each candidate location at each time point and the completed target location in the transfer data at each time point; to perform weighted processing based on the distance parameters and similarity parameters corresponding to each candidate location to obtain the distance evaluation parameters of each candidate location; and to select the candidate location with the highest distance evaluation parameters as the replay target location for each time point based on the distance evaluation parameters of each candidate location.

[0069] As a preferred embodiment, the acquisition module is further configured to sample the interactive observation data of the preset robotic arm model at multiple time points from each interactive trajectory data.

[0070] The motion data of the preset robotic arm model at multiple time points are sampled from each of the interaction trajectory data.

[0071] The motion data includes the pose of the model of the preset workpiece and the pose of the end effector of the model of the preset robotic arm.

[0072] Reward data for completing the same interactive task at the multiple time points is sampled from each interaction trajectory data.

[0073] The target location of the same interaction task at the multiple time points is sampled from each interaction trajectory data.

[0074] The completed target positions at the plurality of time points are sampled from each of the interaction trajectory data.

[0075] Based on the interactive observation data, action data, reward data, target location, and completed target location at the multiple time points, the multiple transfer data are generated respectively.

[0076] If the computing device executing the embodiments of this application is part of the robotic arm, then the computing device can issue relevant instructions to the robotic arm to drive the corresponding joints to move according to the corresponding target angle value. Of course, in some embodiments, if the computing device is a terminal coupled to the robotic arm, then the other terminal can send corresponding control instructions to the robotic arm through a preset information transmission method to instruct the robotic arm to perform corresponding operations. In the embodiments of this application, the specific method of controlling the robotic arm is not limited here.

[0077] As a preferred embodiment, further, it includes acquiring multiple sample pairs, each sample pair including a sample image and environmental state information corresponding to the sample image.

[0078] The sample images in each sample pair are enhanced to obtain corresponding enhanced sample images, and the enhanced sample images are input into the initial robotic arm control student model to obtain the first control action information.

[0079] The environmental state information from the sample pairs is input into the robotic arm control teacher model to obtain the second control action information.

[0080] Based on the first and second control action information corresponding to each sample pair, the model parameters of the initial robotic arm control student model are updated to obtain the robotic arm control student model.

[0081] Acquire pose information collected by at least one sensing device, the pose information including information about the pose of a target object.

[0082] Obtain the mapping relationship between the pose information and the joints in the robotic arm.

[0083] The current arm angle value of the robotic arm is determined based on the pose information and the mapping relationship.

[0084] Based on the relationship between the joint angle and arm angle of each of the M joints, the pose information, and the current arm angle value, the target angle value of each joint is obtained.

[0085] Obtain the target interface corresponding to at least one functional class contained in the robotic arm control program, and expose the target interface corresponding to the at least one functional class.

[0086] The target plug-in interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional classes is used to control the target robotic arm to achieve the corresponding function.

[0087] After exposing the target interface, obtain the target custom program and / or target library file that meets the user's robotic arm control requirements.

[0088] The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inherited and implemented target plugin interface.

[0089] Using a preset initial robotic arm control model, the model of the preset robotic arm in the virtual work environment is controlled to perform multiple interactive tasks on the model of the preset workpiece in the work environment, generating multiple interactive trajectory data.

[0090] Each interactive trajectory data is the trajectory data of the model of the preset robotic arm performing an interactive task.

[0091] If, after traversing the preset angle range, there is no arm angle value to be detected such that the joint angle to be verified for each joint corresponding to the arm angle value to be detected is within the corresponding joint movement range, then for joints whose corresponding target angle value is not within the corresponding joint movement range, the control angle of the joint is determined to be the angle with the smallest difference from the corresponding target angle value while being within the corresponding joint movement range.

[0092] In some embodiments of this application, the pose information may include elbow pose information, wrist pose information, and shoulder pose information. In this case, by controlling the arm angle of the robotic arm, the similarity between the configuration of the robotic arm and the overall shape of the target object to be imitated can be ensured. However, the range of motion of the wrist of the target object (such as a human arm) is often small. Therefore, it is easy for at least one of the lower joints to have a target angle value that is not within the corresponding joint movement range. Therefore, if at least one of the lower joints has a target angle value that is not within the corresponding joint movement range, and after traversing the second preset angle interval, there is no arm angle value to be detected such that the joint angles of each joint corresponding to the arm angle value to be detected satisfy the corresponding joint constraint conditions, then the target included angle value obtained from the pose information can be used to control the value of the arm angle of the robotic arm, ensuring the similarity between the configuration of the robotic arm and the overall shape of the target object to be imitated. At the same time, the control angle of the joint that does not meet the conditions is determined to be the angle that meets the corresponding joint constraint conditions while having the smallest difference from the corresponding target angle value, thereby appropriately sacrificing the imitation accuracy of the lower joint to ensure the accuracy of the shape of the robotic arm.

[0093] As a preferred embodiment, further, the step of enhancing the sample image in the sample pair to obtain the corresponding enhanced sample image includes inputting the sample image and the background sample image into a pre-constructed context-aware enhancement model, and enhancing the sample image based on the background sample image to obtain the enhanced sample image.

[0094] The context-aware enhancement model is generated based on training with multiple image sample pairs.

[0095] When controlling the movement of the robotic arm, the environmental state information obtained after the robotic arm executes the current control action can be input into the robotic arm control student model. Since the robotic arm control student model is trained by combining enhanced sample images and the robotic arm control teacher model, it has strong robustness. Therefore, the robotic arm control student model can better predict the next control action of the robotic arm. Thus, when controlling the movement of the robotic arm based on the next control action, the accuracy of robotic arm control can be improved.

[0096] A robotic arm control device includes a memory and a processor.

[0097] The memory is used to store programs.

[0098] The processor is used to execute the program to implement the various steps of the robotic arm control method.

[0099] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor performing a robotic arm control method.

[0100] A readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the various steps of a robotic arm control method.

[0101] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A motion control system for a robotic arm used for inspecting the undercarriage of a subway car, characterized in that: The system includes: An acquisition module is used to acquire pose information collected by at least one sensing device, the pose information including information about the pose of a target object; The control module is used to control the model of the preset robotic arm in the working environment to perform multiple interactive tasks on the model of the preset workpiece in the working environment using a preset initial robotic arm control model, and generate multiple interactive trajectory data. Each interactive trajectory data is the trajectory data of the model of the preset robotic arm performing an interactive task once; The target interface module is used to obtain the target interface corresponding to at least one functional class contained in the robotic arm control program, and expose the target interface corresponding to the at least one functional class. The target interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional class is used to control the target robotic arm to achieve the corresponding function.

2. The motion control system for a subway car undercarriage inspection robot arm according to claim 1, characterized in that: The system also includes: The acquisition module is used to sample multiple transfer data from each interaction trajectory data, the multiple transfer data including: The virtual model of the preset robotic arm executes process data at multiple time points corresponding to the same interactive task; The information acquisition module is used to acquire target custom programs and / or target library files that meet the user's robotic arm control requirements after exposing the target interface; The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inheritance and implementation of the target plugin interface; The mapping module is used to obtain the mapping relationship between the pose information and the joints in the robotic arm.

3. The motion control system for a subway car undercarriage inspection robot arm according to claim 2, characterized in that: The system also includes: The determining module is used to determine a set of candidate playback targets for the multiple time points from the multiple transfer data; The candidate playback target set for each time point includes: the completed target position of the model of the preset workpiece at at least one time point after each time point; The program pointer determination module is used to obtain the instantiation pointer corresponding to the target custom program based on the target custom program and / or the target library file.

4. The motion control system of the robotic arm for inspecting the undercarriage of a subway car according to claim 3, characterized in that: The system also includes: The update module is used to update the completed target positions in the transfer data corresponding to each time point according to the candidate replay target set for each time point, so as to obtain an updated transfer data; The robotic arm control module is used to run the target custom program based on the instantiation pointer corresponding to the target custom program in order to control the target robotic arm.

5. The motion control system for a metro car undercarriage inspection robot arm according to claim 2, characterized in that: The acquisition module is also used to sample the interactive observation data of the model of the preset robotic arm at multiple time points from each interactive trajectory data; The motion data of the preset robotic arm model at multiple time points are sampled from each of the interaction trajectory data; The action data includes: The pose of the model of the preset workpiece and the pose of the end effector of the model of the preset robotic arm; Reward data for completing the same interaction task at multiple time points is sampled from each interaction trajectory data; Sample the target location of the same interaction task at the multiple time points from each interaction trajectory data; Sample the completed target positions at the plurality of time points from each of the interaction trajectory data; Based on the interactive observation data, action data, reward data, target location, and completed target location at the multiple time points, the multiple transfer data are generated respectively.

6. A motion control method for a robotic arm of a subway car undercarriage inspection robot, characterized in that: include: Multiple sample pairs are acquired, each sample pair including a sample image and environmental state information corresponding to the sample image; Enhancement processing is performed on the sample images in the sample pairs for each sample to obtain corresponding enhanced sample images, and the enhanced sample images are input into the initial robotic arm control student model to obtain the first control action information; The environmental state information from the sample pairs is input into the robotic arm control teacher model to obtain the second control action information; Based on the first and second control action information corresponding to each sample pair, the model parameters of the initial robotic arm control student model are updated to obtain the robotic arm control student model. Acquire pose information collected by at least one sensing device, the pose information including information about the pose of a target object; Obtain the mapping relationship between the pose information and the joints in the robotic arm; Based on the pose information and the mapping relationship, the current arm angle value of the robotic arm is determined; Based on the relationship between the joint angle and arm angle of each of the M joints, the pose information, and the current arm angle value, the target angle value of each joint is obtained. Obtain the target interface corresponding to at least one functional class contained in the robotic arm control program, and expose the target interface corresponding to the at least one functional class; The target plug-in interface is used to call the program of the at least one functional class contained in the robotic arm control program through the instantiation pointer of the at least one functional class, and any one of the at least one functional classes is used to control the target robotic arm to achieve the corresponding function. After exposing the target interface, obtain the target custom program and / or target library file that meets the user's robotic arm control requirements; The library file is a library file compiled based on the target custom program, and the target custom program refers to a custom program obtained based on the inheritance and implementation of the target plugin interface; Using a preset initial robotic arm control model, the model of the preset robotic arm in the virtual work environment is controlled to perform multiple interactive tasks on the model of the preset workpiece in the work environment, generating multiple interactive trajectory data. Each interactive trajectory data is the trajectory data of the model of the preset robotic arm performing an interactive task.

7. The motion control system for a subway car undercarriage inspection robot arm according to claim 6, characterized in that: The enhancement process for the sample images in the sample pair to obtain the corresponding enhanced sample images includes: The sample image and the background sample image are input into a pre-built context-aware enhancement model, and the sample image is enhanced based on the background sample image to obtain the enhanced sample image; The context-aware enhancement model is generated based on training with multiple image sample pairs.

8. A robotic arm control device, characterized in that: Including memory and processor; The memory is used to store programs; The processor is configured to execute the program to implement the various steps of the robotic arm control method as described in any one of claims 6 to 7.

9. A computing 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 computer program, it implements the robotic arm control method as described in any one of claims 6 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the various steps of the robotic arm control method as described in any one of claims 6 to 7.

Images