Robot-based articulated object control method, system, device, and medium

Abstract

This application provides a robot-based control method, system, device, and medium for articulated objects. The method first obtains the initial articulation attributes of the target articulated object based on its articulation type, thereby initially determining the object's motion characteristics when articulation is triggered. Then, based on the initial articulation attributes, the robot is controlled to perform trajectory control on the target articulated object to obtain the initial trajectory execution result in this control process. Finally, the initial articulation attributes of the object are corrected using a preset ASSG algorithm and the initial trajectory execution result, and the robot is controlled to repeatedly perform trajectory control on the object based on the obtained corrected articulation attributes. This method aims to fully understand the physical properties and articulation motion characteristics of the target articulated object through iterative optimization based on single real-time trajectory results, thereby generating a stable robot articulation control method and effectively improving the robot's control performance for articulated objects.

Description

Technical Field

[0001] This application relates to the field of reinforcement learning technology, and in particular to a method, system, device and medium for controlling articulated objects based on robots. Background Technology

[0002] Currently, automated control of rigid objects using robots is widely used in related technologies. However, due to the complex physical structure and limited motion space of articulated objects, the automated control of articulated objects by robots cannot adapt to various types of articulated objects, resulting in poor control performance.

[0003] Therefore, how to improve the control effect of robots on articulated objects has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of the above problems, in order to improve the control effect of robots on articulated objects, this application provides a robot-based control method, system, device and medium for articulated objects.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, embodiments of this application provide a robot-based control method for articulated objects, including:

[0007] Based on the hinge type of the target hinge object, the initial hinge attributes of the target hinge object are obtained; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when hinge motion is triggered.

[0008] Based on the initial articulation properties, the robot is controlled to perform motion trajectory control on the target articulated object in order to obtain the initial trajectory execution result;

[0009] Based on the preset ASSG algorithm and the initial trajectory execution result, the initial articulation attribute is modified to obtain the articulation correction attribute, and the robot is controlled to repeatedly perform motion trajectory control on the target articulated object through the articulation correction attribute; the preset ASSG algorithm is an online reinforcement learning algorithm based on the single trajectory execution result.

[0010] In one possible implementation, obtaining the initial hinge properties of the target hinge-type object includes:

[0011] Acquire RGB and depth images of the target articulated object;

[0012] The RGB image is segmented to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface.

[0013] Based on the first region position and the second region position, the RGB image is aligned with the depth image to determine the hinge type of the target hinged object;

[0014] The initial hinge attributes are determined based on the hinge type, the first region position, the second region position, and the depth image.

[0015] In one possible implementation, the hinge type includes: rotary hinge and translational hinge;

[0016] Determining the initial hinge attributes based on the hinge type, the first region position, the second region position, and the depth image includes:

[0017] When the hinge type is the rotary hinge, a first hinge attribute and a second hinge attribute are determined based on the first region position, the second region position, and the spatial geometric information in the depth image, and the first hinge attribute and the second hinge attribute are determined as the initial hinge attribute;

[0018] When the hinge type is the translational hinge, the second hinge attribute is determined based on the position of the first region, the position of the second region, and the spatial geometric information in the depth image, and the second hinge attribute is determined as the initial hinge attribute;

[0019] The first hinge attribute is the spatial position information of the center point of the edge of the object's surface region, and the second hinge attribute is the hinge motion direction of the object.

[0020] In one possible implementation, the initial trajectory execution result includes: trajectory execution completion degree and average execution tension during the motion trajectory control;

[0021] The initial articulation attributes are corrected based on the preset ASSG algorithm and the initial trajectory execution result to obtain articulation correction attributes, including:

[0022] Based on the completion rate of the trajectory execution, the execution distance of the motion trajectory is determined;

[0023] The motion trajectory execution distance, the average execution tension, and the expected trajectory execution distance for the target articulated object are imported into the preset ASSG algorithm to determine the execution feedback reward; the execution feedback reward is used to characterize the machine's motion trajectory control effect on the target articulated object;

[0024] Based on the execution feedback reward, the initial articulation attribute is modified to obtain the modified articulation attribute.

[0025] In one possible implementation, a force sensor is provided at the end of the robot's robotic arm, and the method further includes:

[0026] When the robot controls the motion trajectory of the target articulated object, the force sensor acquires the pulling force in real time.

[0027] When the pulling force is reduced to zero or exceeds a preset safety threshold, the trajectory control operation of the robot is terminated.

[0028] In one possible implementation, controlling the robot to perform motion trajectory control on the target articulated object based on the initial articulation attributes includes:

[0029] Based on the location of the first region and the initial hinge attributes, a motion execution trajectory is generated;

[0030] Based on the motion execution trajectory, the robot is controlled to perform motion trajectory control on the target articulated object.

[0031] Secondly, embodiments of this application provide a robot-based control system for articulated objects, comprising:

[0032] The attribute acquisition module is used to acquire the initial hinge attributes of the target hinge object according to the hinge type of the target hinge object; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when the hinge motion is triggered.

[0033] The first trajectory control module is used to control the robot to perform motion trajectory control on the target articulated object based on the initial articulation attributes, so as to obtain the initial trajectory execution result;

[0034] The second trajectory control module is used to correct the articulation attributes based on the preset ASSG algorithm and the initial trajectory execution result, obtain the articulation correction attribute, and control the robot to repeatedly perform motion trajectory control on the target articulated object through the articulation correction attribute; the preset ASSG algorithm is an online reinforcement learning algorithm based on the single trajectory execution result.

[0035] In one possible implementation, the attribute acquisition module is specifically used for:

[0036] Acquire RGB and depth images of the target articulated object;

[0037] The RGB image is segmented to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface.

[0038] Based on the first region position and the second region position, the RGB image is aligned with the depth image to determine the hinge type of the target hinged object;

[0039] The initial hinge properties are determined based on the hinge type, the first region position, and the second region position.

[0040] Thirdly, embodiments of this application provide an electronic device, the device including: a processor, a memory, and a system bus;

[0041] The processor and the memory are connected via the system bus;

[0042] The memory is used to store one or more programs, the one or more programs including instructions that, when executed by the processor, cause the processor to perform any possible robot-based articulated object control method in the first aspect.

[0043] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any possible robot-based articulated object control method described in the first aspect.

[0044] Compared to existing technologies, this application offers the following advantages: This application provides a robot-based control method, system, device, and medium for articulated objects. The method first obtains the initial articulation attributes of the target articulated object based on its articulation type, thereby initially determining the object's motion characteristics when articulation is triggered. Further, based on the initial articulation attributes, the robot is controlled to perform trajectory control on the target articulated object, thus obtaining the robot's initial trajectory execution result in this control process. Finally, the initial articulation attributes of the object are corrected using a preset ASSG algorithm and the robot's initial trajectory execution result. Based on the obtained corrected articulation attributes, the robot is controlled to repeatedly perform trajectory control on the object. This process of correcting the articulation attributes is repeated, and each correction is based on the robot's actual trajectory execution result, thereby fully understanding the physical properties and articulation motion characteristics of the target articulated object, and generating a stable robot articulation control method, effectively improving the robot's control effect on articulated objects. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A flowchart illustrating a robot-based articulated object control method provided in this application embodiment;

[0047] Figure 2 This is a schematic diagram of the structure of a vehicle temperature abnormality alarm system provided in an embodiment of this application;

[0048] Figure 3 A schematic diagram of a robot-based articulated object control method provided in an embodiment of this application;

[0049] Figure 4 A schematic diagram of the architecture of an ASSG algorithm provided in an embodiment of this application;

[0050] Figure 5 A flowchart illustrating a hinge property correction method provided in an embodiment of this application;

[0051] Figure 6 A schematic diagram illustrating the relationship between execution resistance and the number of optimization attempts provided in this application embodiment;

[0052] Figure 7 A schematic diagram of a robot-based articulated object control system provided in this application embodiment;

[0053] Figure 8 This is a schematic diagram of a robot-based articulated object control electronic device provided in an embodiment of this application. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and accompanying drawings. It should be particularly noted that the embodiments described in this application are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0055] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0056] As described earlier, automated control of rigid objects using robots is widely used in current technologies. However, due to the complex physical structure and limited motion space of articulated objects, current automated control of articulated objects by robots cannot adapt to various types of articulated objects, resulting in poor control performance.

[0057] Therefore, how to improve the control effect of robots on articulated objects has become a technical problem that urgently needs to be solved by those skilled in the art.

[0058] To address the aforementioned issues, this application provides a robot-based control method, system, device, and medium for articulated objects. The method first obtains the initial articulation attributes of the target articulated object based on its articulation type, thereby initially determining the object's motion characteristics when articulation is triggered. Further, based on the initial articulation attributes, the robot is controlled to perform trajectory control on the target articulated object, thus obtaining the robot's initial trajectory execution result in this control process. Finally, the initial articulation attributes of the object are corrected using a preset ASSG algorithm and the robot's initial trajectory execution result, and the robot is controlled to repeatedly perform trajectory control on the object based on the obtained corrected articulation attributes. This process of correcting the articulation attributes is repeated, with each correction based on the robot's actual trajectory execution result, thereby fully understanding the physical properties and articulation motion characteristics of the target articulated object, and generating a stable robot articulation control method, effectively improving the robot's control performance for articulated objects.

[0059] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0060] See Figure 1 The figure is a flowchart illustrating a robot-based articulated object control method according to an embodiment of this application, specifically including the following steps:

[0061] S101: Based on the hinge type of the target hinge object, obtain the initial hinge attributes of the target hinge object; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when the hinge motion is triggered.

[0062] As described in the background section, articulated objects often have complex physical structures. When robots perform automated operations such as pulling and closing on articulated objects (e.g., doors, refrigerators, drawers), the automated operation is constrained by the complex physical structure of the object, making it impossible for the robot to plan an effective articulated action execution path. Therefore, to address this problem, in the initial stage of controlling the articulated object, this embodiment needs to obtain the initial articulation attributes of the target articulated object based on its articulation type to gain a preliminary understanding of its articulation characteristics.

[0063] The initial articulation attribute is used to characterize the motion characteristics of an object when the articulation motion is triggered. By using the object's articulation attribute, the motion trajectory of the object when the articulation motion is triggered can be calculated, thereby driving the robot to execute the motion trajectory to achieve automated control of the target articulated object.

[0064] The motion characteristics of articulated objects can be understood as the motion path when the articulation is triggered, the trigger point of the articulation, and the surface area of ​​the object that triggers the articulation. For example, taking a door as an example, the form of the door's articulation trigger is usually a quarter circle, so its initial motion path is a quarter circle with the width of the door's plane as the radius. Similarly, the door handle corresponds to the trigger point of the articulation, and the plane on which the door handle is located corresponds to the surface area where the articulation is triggered.

[0065] In this step, the hinge type and initial hinge attributes of the target hinged object need to be obtained through image analysis of the object. The following will describe this process with reference to the accompanying drawings of a specific embodiment.

[0066] See Figure 2 and Figure 3 , Figure 2 This is a flowchart illustrating a method for obtaining initial hinge attributes according to an embodiment of this application. Figure 3 This is a schematic diagram illustrating a robot-based control method for articulated objects, provided in an embodiment of this application. Figure 2 As shown, the method specifically includes the following steps:

[0067] S1011: Obtain the RGB image and depth image of the target articulated object.

[0068] First, it is necessary to use a camera mounted on the outside of the robot to capture RGB and depth images of the target articulated object.

[0069] The core difference between depth images and RGB images lies in the type and purpose of the information they carry. RGB images record visual features such as color, texture, and lighting on the surface of articulated objects using three color channels: red, green, and blue. They are mainly used for semantic understanding (e.g., identifying door handles, segmenting object regions). Depth images, on the other hand, represent the distance information from each pixel to the camera using a single-channel value. They can reflect the three-dimensional spatial structure of articulated objects, such as surface geometry, the coordinates of hinge contact points, and so on.

[0070] Therefore, obtaining the hinge attributes of the target hinged object based on its depth image and RGB image can combine the object's color semantic information and three-dimensional geometric information, thereby improving the accuracy and robustness of the robot's perception of hinged objects and ensuring the accuracy of the initial hinge attribute data.

[0071] S1012: Perform image segmentation processing on the RGB image to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface.

[0072] Subsequently, image segmentation processing is performed on the RGB image to determine the location information of the hinge motion trigger point of the target hinged object in the RGB image (first region location), and the location information of the object surface to which the hinge motion trigger point belongs (second region location).

[0073] For details, please refer to the reference. Figure 3 Examples in, Figure 3 Using a door as an example of a hinged object, the SAM2 image segmentation technique is used to segment the RGB image of the door to determine the position of the door handle and the area of ​​the object surface to which the door handle belongs.

[0074] In one possible implementation, the location of the articulated motion trigger point can also be determined through image analysis of the depth image. Figure 3 The AnyGrasp method used in this embodiment can determine the point closest to the door handle in space as the gripping point (and the trigger point for hinge movement). In practical applications, the accuracy of obtaining the position of the first region can be improved by combining RGB image analysis and depth image analysis. This embodiment will not elaborate on this aspect.

[0075] S1013: Based on the first region position and the second region position, align the RGB image with the depth image to determine the hinge type of the target hinge-type object.

[0076] Based on determining the trigger point of the hinge motion and the surface of the object to which it belongs, the RGB image and the depth image are compared and analyzed. The spatial geometric information represented by the depth image is used to determine the hinge type of the target hinge object, so as to obtain specific hinge attributes according to different hinge types.

[0077] Specifically, pre-calibrated parameters (such as camera intrinsics) can be used to map the RGB pixel coordinates corresponding to the trigger point position (first region position) and surface region position (second region position) obtained after image segmentation to the coordinate system of the depth image, thereby generating three-dimensional spatial coordinates for the target articulated object, and determining the articulation type based on the obtained three-dimensional spatial coordinates.

[0078] S1014: Determine the initial hinge attributes based on the hinge type, the first region position, the second region position, and the depth image.

[0079] It is understandable that, due to the complex physical structure of articulated objects, the initial articulation properties used to characterize their articulation motion characteristics will vary significantly depending on the type of articulation. Therefore, in order to clarify the articulation characteristics of objects of different articulation types, this embodiment of the application needs to determine the initial articulation properties of the object by combining the object's articulation type, the location of the articulation motion trigger point (first region location), the region location of its surface (second region location), and the spatial geometric information represented in the depth image.

[0080] The hinge types of objects are classified into rotational hinges and translational hinges. When the hinge type is a rotational hinge, it is necessary to determine the first hinge attribute and the second hinge attribute based on the spatial geometric information in the first region position, the second region position, and the depth image, and then set the first hinge attribute and the second hinge attribute as the initial hinge attribute.

[0081] The first hinge attribute is used to characterize the spatial position information of the center point of the edge of the object's surface region, and the second hinge attribute is used to characterize the hinge motion direction of the object.

[0082] Understandably, when an object's hinge type is a rotary hinge, it's necessary to specify the axis of rotation when the rotary hinge movement is triggered, as well as the object's rotation direction. For example, for rotary hinge objects such as cabinet doors or microwave ovens, the center point of the edge of the surface containing the hinge movement trigger point (handle) is taken as the starting point p of the rotation axis. The spatial position information of this starting point p is the first hinge attribute. Correspondingly, the direction of the edge of the surface area is further obtained; this direction can be used as the direction of the rotation axis (e.g., the direction of rotation of a door around the edge of the door frame), i.e., the direction of motion q when the hinge movement is triggered, thus obtaining the second hinge attribute.

[0083] On the other hand, when the hinge type is the translational hinge, the second hinge attribute is determined based on the position of the first region, the position of the second region, and the spatial geometric information in the depth image, and the second hinge attribute is determined as the initial hinge attribute.

[0084] When an object's hinge type is a translational hinge, similar to that of a drawer, the hinge axis of a translational hinge does not have a significant impact on the physical structure like a rotational hinge. Therefore, it is only necessary to obtain the translational direction of the object when the hinge movement is triggered. Taking a drawer as an example, based on the first region position of the drawer's hinge movement trigger point and the second region position of the plane to which the trigger point belongs, the normal vector of the plane where the second region position is located can be obtained. This allows the determination of the drawer's translational direction q when the hinge movement is triggered, i.e., the second hinge attribute.

[0085] S102: Based on the initial articulation attributes, control the robot to perform motion trajectory control on the target articulated object to obtain the initial trajectory execution result.

[0086] After obtaining the initial articulation attributes, based on the location of the first region of the articulation motion trigger point and the articulation motion characteristics represented by the initial articulation attributes, a motion execution trajectory that the robot needs to refer to when controlling the target articulated object is generated, so that the robot can control the target articulated object to perform articulation motion, and obtain the initial trajectory execution result after the articulation motion is completed.

[0087] As mentioned above, the initial articulation attributes in this embodiment are obtained through image segmentation of the RGB image and alignment and analysis of the depth image. This method relies on data captured by an external camera, which may often contain a certain degree of error. Therefore, the motion trajectory generated under this error may lead to poor robot control of the object. Thus, to improve the robot's control over articulated objects, after controlling the robot's motion trajectory based on the initial articulation attributes, it is necessary to obtain the robot's initial trajectory execution result. This initial trajectory execution result characterizes the robot's control effect during automated trajectory control, allowing subsequent optimization of the articulation attributes based on real-time trajectory execution results. This ensures that the trajectory execution result closely matches the expected execution trajectory, thereby optimizing the robot's control performance.

[0088] The initial trajectory execution result includes the trajectory execution completion rate and the average execution pull force of the robot controlling the object's articulated movement during a single motion trajectory. In this embodiment, a force sensor is installed at the end of the robot's robotic arm. This force sensor can acquire the execution pull force in real time when the robot pulls the articulated movement trigger point to control the articulated movement, and monitor the magnitude of the execution pull force in real time. When the execution pull force is zero, it indicates that the robot's robotic arm gripper may detach from the object. In this case, the trajectory control operation needs to be stopped, and the trajectory execution data of this instance should be used as the trajectory execution result. Similarly, when the execution pull force exceeds a preset safety threshold, it indicates that there is a significant difference between the robot's motion execution trajectory and the actual articulated movement trajectory of the object. In this case, the trajectory control operation also needs to be stopped, and the corresponding trajectory execution result should be obtained.

[0089] Accordingly, the trajectory execution completion rate can be obtained by acquiring the current execution trajectory distance when the force sensor receives zero execution tension, and then calculating it with the preset expected trajectory execution distance to obtain the trajectory execution completion rate of the robot during this controlled articulated motion. Additionally, the average execution tension can be determined by the average value of the forces monitored by the sensors during this control process; this embodiment does not impose limitations on this.

[0090] S103: Based on the preset ASSG algorithm and the initial trajectory execution result, the initial articulation attribute is modified to obtain the articulation correction attribute, and the robot is controlled to repeatedly perform motion trajectory control on the target articulated object through the articulation correction attribute; the preset ASSG algorithm is an online reinforcement learning algorithm based on the single trajectory execution result.

[0091] This step, as the core of this application, aims to analyze the robot's trajectory execution results using a pre-designed ASSG (Accelerated Single-Step Gradient) algorithm, and thereby correct the initial articulation properties of the object. Through repeated iterations and online training, the robot's understanding of the articulation characteristics of the target articulated object is optimized, making the robot's understanding of the target object's articulation characteristics more accurate. This, in turn, generates motion control trajectories that better match the actual articulation characteristics of the object, thereby improving the control effect.

[0092] The core of the preset ASSG algorithm lies in performing online optimization and reinforcement learning of the object's articulation properties based on the trajectory execution results generated each time the robot performs articulated motion control. The aim is to gradually adjust the articulation properties through robot execution feedback and optimize the neural network strategy used for reinforcement learning within the algorithm. The specific architecture of the preset ASSG algorithm actor-critic architecture in this application embodiment can be found in [reference needed]. Figure 4 The diagram illustrates the architecture of an ASSG algorithm. In the diagram, (s, a, r) ​​represents the (state, action, reward) tuple in the Markov process; traj represents the action trajectory sent to the robot; R represents the execution feedback reward generated by the robot after generating the trajectory execution result, which characterizes the machine's control effect on the motion trajectory of the target articulated object; the Replay Buffer represents the experience replay pool, which stores the state s, the network-generated action a, and the execution feedback reward R in a packaged storage pool for random retrieval of training data during network training; MLP represents a multilayer perceptron; Q represents the Q-value used during the training of the critic network; (p, q) represents the articulation correction attribute generated by the network, used to generate the robot's action execution trajectory; and Ornstein-Uhlenbeck noise represents the noise introduced during training.

[0093] Depend on Figure 4 As can be seen, the ASSG algorithm corrects the articulation properties of the target articulated object in real time based on the robot's motion trajectory execution results for each step, thereby optimizing the robot's control trajectory. Next, step S103 will be described in detail with reference to the accompanying drawings of a specific process embodiment.

[0094] See Figure 5 The figure is a flowchart illustrating a hinge property correction method provided in an embodiment of this application, which specifically includes the following steps:

[0095] S1031: Determine the motion trajectory execution distance based on the trajectory execution completion degree.

[0096] S1032: The motion trajectory execution distance, the average execution tension, and the expected trajectory execution distance for the target articulated object are imported into the preset ASSG algorithm to determine the execution feedback reward; the execution feedback reward is used to characterize the machine's motion trajectory control effect on the target articulated object;

[0097] S1033: Based on the execution feedback reward, the initial articulation attribute is modified to obtain the modified articulation attribute.

[0098] As described above, the actual articulation attribute correction process in this embodiment needs to be determined based on the trajectory execution results of the robot when controlling the articulated motion. This process requires calculating the execution feedback reward based on the motion trajectory execution distance, average execution tension, and a pre-set expected trajectory execution distance from the trajectory execution results. The execution feedback reward characterizes the current robot's motion trajectory control effect on the target articulated object. Correcting the initial articulation attributes of the object based on the execution feedback reward can optimize the robot's trajectory execution path, thereby improving the control effect.

[0099] Specifically, the method for calculating the execution feedback reward based on the preset ASSG algorithm can be found in the following formula:

[0100]

[0101] In the formula, R represents the execution feedback reward, l represents the motion trajectory execution distance in the trajectory execution result, L represents the expected execution distance for a specific target articulated object, α represents the uniform metric hyperparameter, and F represents the average execution tension.

[0102] Additionally, the process for repairing the initial hinge properties is illustrated in the following formula:

[0103] p * =p+p a

[0104]

[0105] In the formula, p * q represents the hinge correction attribute corresponding to the first hinge attribute. * This represents the hinge correction attribute corresponding to the second hinge attribute, p represents the first hinge attribute in the initial hinge attributes, and q represents the second hinge attribute in the initial hinge attributes. a and q a These represent the correction values ​​corresponding to each hinge attribute.

[0106] Thus, by optimizing the articulation properties in real time based on the robot's trajectory execution results each time, it is possible to ensure that the robot's actual motion trajectory more closely matches the desired motion trajectory. As the number of optimizations increases, the resistance encountered by the robot during execution will gradually decrease. For details on the effects, please refer to [link to relevant documentation]. Figure 6 The diagram illustrates the relationship between execution resistance and the number of optimization attempts.

[0107] This application provides a robot-based control method for articulated objects. The method first obtains the initial articulation attributes of the target articulated object based on its articulation type, thereby initially determining the object's motion characteristics when articulation is triggered. Further, based on the initial articulation attributes, the robot is controlled to perform trajectory control on the target articulated object, thus obtaining the robot's initial trajectory execution result in this control process. Finally, the initial articulation attributes of the object are corrected using a preset ASSG algorithm and the robot's initial trajectory execution result, and the robot is controlled to repeatedly perform trajectory control on the object based on the obtained corrected articulation attributes. This process of correcting the articulation attributes is repeated, with each correction based on the robot's actual trajectory execution result, thereby fully understanding the physical properties and articulation motion characteristics of the target articulated object, and generating a stable robot articulation control method, effectively improving the robot's control performance for articulated objects.

[0108] The following describes a robot-based articulated object control system provided by an embodiment of this application. The robot-based articulated object control system described below and the robot-based articulated object control method described above can be referred to and correspond to each other.

[0109] See Figure 7 The figure is a schematic diagram of a robot-based articulated object control system provided in an embodiment of this application, specifically including the following modules:

[0110] The attribute acquisition module 100 is used to acquire the initial hinge attributes of the target hinge object according to the hinge type of the target hinge object; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when the hinge motion is triggered.

[0111] The first trajectory control module 200 is used to control the robot to perform motion trajectory control on the target articulated object based on the initial articulation attributes, so as to obtain the initial trajectory execution result;

[0112] The second trajectory control module 300 is used to perform articulation attribute correction based on the preset ASSG algorithm and the initial trajectory execution result to obtain articulation correction attributes, and to control the robot to repeatedly perform motion trajectory control on the target articulated object through the articulation correction attributes; the preset ASSG algorithm is an online reinforcement learning algorithm based on the single trajectory execution result.

[0113] In one possible implementation, the attribute acquisition module 100 is specifically used for:

[0114] Acquire RGB and depth images of the target articulated object;

[0115] The RGB image is segmented to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface.

[0116] Based on the first region position and the second region position, the RGB image is aligned with the depth image to determine the hinge type of the target hinged object;

[0117] The initial hinge properties are determined based on the hinge type, the first region position, and the second region position.

[0118] See Figure 8 The figure is a schematic diagram of a robot-based articulated object control electronic device provided in an embodiment of this application, including:

[0119] Memory 11 is used to store computer programs;

[0120] The processor 12 is used to implement the steps of the robot-based articulated object control method described in any of the above method embodiments when executing the computer program.

[0121] In this embodiment, the device can be an in-vehicle computer, a PC (Personal Computer), or a terminal device such as a smartphone, tablet computer, handheld computer, or portable computer.

[0122] The device may include a memory 11, a processor 12, and a bus 13.

[0123] The memory 11 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the device, such as the hard disk of the device. In other embodiments, the memory 11 may be an external storage device of the device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, Flash Card, etc. Furthermore, the memory 11 may include both internal and external storage units of the device. The memory 11 can be used not only to store application software and various types of data installed on the device, such as program code executing a robot-based articulated object control method, but also to temporarily store data that has been output or will be output. In some embodiments, the processor 12 may be a Central Processing Unit (CPU).

[0124] In some embodiments, processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 11 or process data, such as executing program code for a robot-based articulated object control method.

[0125] This bus 13 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0126] Furthermore, the device may also include a network interface 14, which may optionally include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), typically used to establish communication connections between the device and other electronic devices.

[0127] Optionally, the device may further include a user interface 15, which may include a display, an input unit such as a keyboard, and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the device and to display a visual user interface.

[0128] Figure 8 Only devices with components 11-15 are shown; those skilled in the art will understand that... Figure 8 The structure shown does not constitute a limitation on the device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0129] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a computer-readable storage medium storing computer instructions for causing the computer to execute the robot-based articulated object control method as described in any of the above embodiments.

[0130] The computer-readable media in this application embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0131] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the robot-based articulated object control method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0132] It should be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for methods, systems, electronic devices, and media, since they are basically similar to the method embodiments, the descriptions are relatively simple, and relevant parts can be referred to the descriptions of the method embodiments. The methods, systems, electronic devices, and media described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components indicated as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0133] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A robot-based control method for articulated objects, characterized in that, include: Based on the hinge type of the target hinge object, obtain the initial hinge attributes of the target hinge object; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when hinge motion is triggered. Based on the initial articulation properties, the robot is controlled to perform motion trajectory control on the target articulated object in order to obtain the initial trajectory execution result; Based on the preset ASSG algorithm and the initial trajectory execution result, the initial articulation attribute is corrected to obtain the articulation correction attribute. This corrected attribute is then used to control the robot to repeatedly perform motion trajectory control on the target articulated object. The preset ASSG algorithm is an online reinforcement learning algorithm based on a single trajectory execution result, and the ASSG algorithm itself is a single-step online reinforcement learning algorithm. The initial trajectory execution result includes: trajectory execution completion degree and average execution tension during the motion trajectory control; The method for determining the articulation correction attribute includes: Based on the completion rate of the trajectory execution, the execution distance of the motion trajectory is determined; The motion trajectory execution distance, the average execution tension, and the expected trajectory execution distance for the target articulated object are imported into the preset ASSG algorithm to determine the execution feedback reward; the execution feedback reward is used to characterize the machine's motion trajectory control effect on the target articulated object; Based on the execution feedback reward, the initial articulation attribute is modified to obtain the modified articulation attribute; The method for calculating the execution feedback reward using the preset ASSG algorithm includes the following formula: ； In the formula, R represents the execution feedback reward, l represents the execution distance of the motion trajectory in the trajectory execution result, L represents the expected execution distance for a specific target articulated object, α represents the uniform metric hyperparameter, and F represents the average execution tension; The process of correcting the initial hinge properties includes the following formula: ； ； In the formula, This indicates the hinge correction attribute corresponding to the first hinge attribute. This indicates the hinge correction attribute corresponding to the second hinge attribute. This represents the first hinge property in the initial hinge properties. This represents the second hinge attribute in the initial hinge attributes. and These represent the correction values ​​corresponding to each hinge attribute.

2. The method according to claim 1, characterized in that, The process of obtaining the initial hinge properties of the target hinge-type object includes: Acquire RGB and depth images of the target articulated object; The RGB image is segmented to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface. Based on the first region position and the second region position, the RGB image is aligned with the depth image to determine the hinge type of the target hinged object; The initial hinge attributes are determined based on the hinge type, the first region position, the second region position, and the depth image.

3. The method according to claim 2, characterized in that, The hinge types include: rotary hinges and translational hinges; Determining the initial hinge attributes based on the hinge type, the first region position, the second region position, and the depth image includes: When the hinge type is the rotary hinge, a first hinge attribute and a second hinge attribute are determined based on the first region position, the second region position, and the spatial geometric information in the depth image, and the first hinge attribute and the second hinge attribute are determined as the initial hinge attribute; When the hinge type is the translational hinge, the second hinge attribute is determined based on the position of the first region, the position of the second region, and the spatial geometric information in the depth image, and the second hinge attribute is determined as the initial hinge attribute; The first hinge attribute is the spatial position information of the center point of the edge of the object's surface region, and the second hinge attribute is the hinge motion direction of the object.

4. The method according to claim 1, characterized in that, The robot's robotic arm is equipped with a force sensor at its end, and the method further includes: When the robot controls the motion trajectory of the target articulated object, the force sensor acquires the pulling force in real time. When the pulling force is reduced to zero or exceeds a preset safety threshold, the trajectory control operation of the robot is terminated.

5. The method according to claim 3, characterized in that, The step of controlling the robot to perform motion trajectory control on the target articulated object based on the initial articulation attributes includes: Based on the location of the first region and the initial hinge attributes, a motion execution trajectory is generated; Based on the motion execution trajectory, the robot is controlled to perform motion trajectory control on the target articulated object.

6. A robot-based control system for articulated objects, characterized in that, include: The attribute acquisition module is used to acquire the initial hinge attributes of the target hinge object based on the hinge type of the target hinge object; the initial hinge attributes are used to characterize the motion characteristics of the target hinge object when the hinge motion is triggered. The first trajectory control module is used to control the robot to perform motion trajectory control on the target articulated object based on the initial articulation attributes, so as to obtain the initial trajectory execution result; The second trajectory control module is used to modify the initial articulation attribute based on the preset ASSG algorithm and the initial trajectory execution result to obtain the articulation correction attribute, and control the robot to repeatedly perform motion trajectory control on the target articulated object through the articulation correction attribute; the preset ASSG algorithm is an online reinforcement learning algorithm based on the single trajectory execution result, and the ASSG algorithm is a single-step online reinforcement learning algorithm; The initial trajectory execution result includes: trajectory execution completion degree and average execution tension during the motion trajectory control; The method for determining the articulation correction attribute includes: Based on the completion rate of the trajectory execution, the execution distance of the motion trajectory is determined; The motion trajectory execution distance, the average execution tension, and the expected trajectory execution distance for the target articulated object are imported into the preset ASSG algorithm to determine the execution feedback reward; the execution feedback reward is used to characterize the machine's motion trajectory control effect on the target articulated object; Based on the execution feedback reward, the initial articulation attribute is modified to obtain the modified articulation attribute; The method for calculating the execution feedback reward using the preset ASSG algorithm includes the following formula: ； In the formula, R represents the execution feedback reward, l represents the execution distance of the motion trajectory in the trajectory execution result, L represents the expected execution distance for a specific target articulated object, α represents the uniform metric hyperparameter, and F represents the average execution tension; The process of correcting the initial hinge properties includes the following formula: ； ； In the formula, This indicates the hinge correction attribute corresponding to the first hinge attribute. This indicates the hinge correction attribute corresponding to the second hinge attribute. This represents the first hinge property in the initial hinge properties. This represents the second hinge attribute in the initial hinge attributes. and These represent the correction values ​​corresponding to each hinge attribute.

7. The system according to claim 6, characterized in that, The attribute acquisition module is specifically used for: Acquire RGB and depth images of the target articulated object; The RGB image is segmented to determine the location of the first region of the hinge motion trigger point and the location of the second region on the object surface. Based on the first region position and the second region position, the RGB image is aligned with the depth image to determine the hinge type of the target hinged object; The initial hinge properties are determined based on the hinge type, the first region position, and the second region position.

8. An electronic device, characterized in that, The device includes: a processor, a memory, and a system bus; The processor and the memory are connected via the system bus; The memory is used to store one or more programs, the one or more programs including instructions that, when executed by the processor, cause the processor to perform the robot-based articulated object control method according to any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the robot-based articulated object control method according to any one of claims 1-5.

Images