Robot arm decision-making and control method and system based on task tree algorithm and storage medium
Through the combination of task tree algorithm and decision tree, the robot arm can quickly adapt to different production line tasks, reduce production line switching delays, reduce costs and support multi-task execution, solving the problems of expensive assignment of robot arm tasks and long production line replacement time in the prior art.
Patent Information
- Application Number
- CN202510753349.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-06
- Publication Date
- 2025-07-22
AI Technical Summary
In the prior art, robot arms are assigned too expensive in automated intelligent manufacturing plants and the production line is replaced for a long time, resulting in high production costs and loss of production capacity.
The robot arm decision-making and control method based on the task tree algorithm is adopted, object information is obtained through the robotic arm vision system, 2D attribute features are coded and generated, and the task tree algorithm is used to record positions and output action sequences through the decision tree to realize action recognition and execution.
Improves adaptability of robotic arms, reduces production line switching delay time, reduces production costs, and supports multitasking execution and collaboration with human workers.
Smart Images
Figure BDA0005438015090000011 
Figure BDA0005438015090000021 
Figure BDA0005438015090000041
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of tunnel lighting, and in particular, to a robot arm decision-making and control method, system, and storage medium based on a task tree algorithm. Background Art
[0002] Robot arms are widely used in product assembly in automated intelligent manufacturing factories. Generally, tasks such as product packaging, testing, and assembly are performed by multiple robot arms deployed on the production line, and each robot arm is specifically assigned to perform one task. This method achieves the goal of intelligent manufacturing, but it is still too expensive and does not fully utilize the potential of the robot arm (i.e., the ability to repeatedly perform tasks at high speed and high precision). The practice of assigning one task to each robotic arm that is prevalent in manufacturing factories results in high production costs. In addition, the time required to replace the fixtures of each robotic arm on the production line results in a loss of production capacity. Summary of the Invention
[0003] The object of the present invention is to provide a robot arm decision-making and control method, system, and storage medium based on a task tree algorithm to solve one or more technical problems existing in the prior art and at least provide a beneficial alternative or create conditions.
[0004] The present invention adopts the following technical solutions to achieve the above object of the invention:
[0005] The present invention provides a robot arm decision-making and control method based on a task tree algorithm, including:
[0006] Obtaining object information through a robotic arm vision system;
[0007] Encoding to implement the information fusion process and generating 2D attribute features from the object information;
[0008] Importing the 2D attribute features as motion data into the task tree algorithm, so that the task tree algorithm automatically records the position based on the imported motion data as the basis for action execution;
[0009] Importing the pre-set target actions and tasks into the task tree algorithm through a decision tree, so that the task tree algorithm performs action recognition and outputs an action sequence;
[0010] Based on the output action sequence, the robotic arm completes the corresponding actions.
[0011] Further, the method for obtaining object information through a robotic arm vision system includes:
[0012] Collecting an image of the object through a camera installed on the robotic arm;
[0013] Preprocess the acquired object image using the Gaussian filtering method;
[0014] Analyze the preprocessed image to identify the key features of the object;
[0015] Identify the pose of the object by detecting the key feature points of the object and calculating the relative positional relationship of these feature points.
[0016] Furthermore, the formula for preprocessing the acquired object image using the Gaussian filtering method is as follows:
[0017]
[0018]
[0019] where σ is the standard deviation of the Gaussian distribution, controlling the smoothness of the filter; i 2 +j 2 is the square of the Euclidean distance from the current pixel (x + i, y + j) to the central pixel (x, y); is the normalization coefficient, ensuring that the total weight of the Gaussian kernel is 1; I(x, y) is the pixel value of the original image at the coordinate (x, y); I'(x, y) is the new pixel value of the filtered image at the coordinate (x, y); G(i, j) is the weight value of the Gaussian filter at the offset (i, j); i, j are the horizontal and vertical offsets relative to the central pixel (x, y).
[0020] Furthermore, the method for encoding and implementing the information fusion process to generate 2D attribute features of object information includes:
[0021] Extract the key features and pose features of the identified object;
[0022] Combine and encode the extracted features into a structured 2D attribute feature, represented as a matrix, with the formula as follows:
[0023] motion = [Category1, Orientation1 Category2, Orientation2……]
[0024] where: Category represents the object category identifier, Orientation represents the pose feature, and motion represents the 2D attribute feature of the object;
[0025] Define a motion data structure for storing the information in the feature vector and the parameters related to motion;
[0026] Encapsulate the encoded feature vector into a motion data object and add the parameters required for the robot motion;
[0027] Convert motion data into instructions or data formats directly used by the robot control system.
[0028] Furthermore, a method for enabling the task tree algorithm to automatically record positions based on the imported motion data as the basis for action execution includes:
[0029] Conduct a detailed analysis of the tasks of the robot arm to clarify the execution process and logical relationships of the tasks;
[0030] Define each node in the task tree according to the task logic. The node types include sequential nodes, selection nodes, conditional nodes, and action nodes. Among them, the sequential node represents executing child nodes in sequence, the selection node represents selectively executing child nodes according to conditions, the conditional node is used to judge whether a specific condition is met, and the action node represents specific robot actions;
[0031] Organize the defined nodes into a tree structure according to the task logic;
[0032] Import the transformed motion data into the task tree algorithm for preprocessing;
[0033] In the task tree algorithm, set up a data structure to store the recorded position information. At the start of task execution, initialize this data structure to prepare for recording positions;
[0034] According to the execution process and logic of the task tree, set the update logic for position recording in the corresponding action nodes or conditional nodes;
[0035] In subsequent action nodes, plan the motion trajectory of the robot based on the recorded position information and the position information of the placement area, and use it as the basis for action execution.
[0036] Furthermore, a method for importing the pre-set target actions and tasks into the task tree algorithm through a decision tree, enabling the task tree algorithm to perform action recognition and output an action sequence includes:
[0037] When the target actions and task information of the decision tree are imported into the task tree, the results of the decision tree are mapped to the action nodes and control nodes of the task tree, decomposing complex tasks into multiple subtasks and action units;
[0038] Dynamically adjust the structure or parameters of the task tree according to the target actions and task information provided by the decision tree;
[0039] Traverse the task tree through the depth-first search method. Starting from the starting node of the task tree, select an unvisited adjacent node for access and mark it as visited;
[0040] For each visited node, repeat the above steps until no further in-depth exploration is possible;
[0041] When unable to proceed further, backtrack to the previous node and attempt other unvisited adjacent nodes;
[0042] Repeat the above steps until all task tree nodes have been visited, and finally identify all action nodes;
[0043] Collect the action names of all action nodes that meet the conditions, and at the same time optimize the action sequence using heuristic rules or algorithms, and output the action sequence.
[0044] Furthermore, a method for enabling the robotic arm to complete corresponding actions based on the output action sequence includes:
[0045] Parse the action instructions in the action sequence and extract the specific parameters of each action.
[0046] Ensure that the robotic arm controller has been initialized and the robotic arm is in a ready state to receive and execute action instructions;
[0047] Traverse the action sequence, and according to the type and parameters of each action instruction, call the corresponding robotic arm control function to execute the specific action.
[0048] The present invention provides a robotic arm decision-making and control system based on a task tree algorithm, including:
[0049] An acquisition unit for acquiring object information through the robotic arm vision system;
[0050] An encoding unit for encoding to implement the information fusion process and generating 2D attribute features from the object information;
[0051] An import unit for importing the 2D attribute features as motion data into the task tree algorithm, enabling the task tree algorithm to automatically record positions based on the imported motion data as the basis for action execution;
[0052] An identification unit for importing the pre-set target actions and tasks into the task tree algorithm through a decision tree, enabling the task tree algorithm to perform action identification and output an action sequence;
[0053] An output unit for enabling the robotic arm to complete corresponding actions based on the output action sequence.
[0054] The present invention provides a robotic arm decision-making and control system based on a task tree algorithm, including a memory and a processor;
[0055] The memory is used for storing instructions;
[0056] The processor is used for operating according to the instructions to execute the steps of the above method.
[0057] The present invention provides a computer-readable storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, the steps of the above method are implemented.
[0058] The beneficial effects of the present invention are as follows:
[0059] By combining the conditional judgment of the decision tree and the structured execution model of the task tree, the 2D feature attributes of visual recognition are accurately mapped to the motion data, enabling the robot system to reach a new height in terms of accuracy.
[0060] Through the decision-making logic in the task tree algorithm, when it receives object information on different production lines, it can quickly find a suitable action sequence from the corresponding decision branches, making the robotic arm more adaptable, reducing the delay time of algorithm operations and decision-making, realizing rapid switching between production lines, enabling the robotic arm to not only complete tasks on one production line but also assist tasks on other production lines. In addition, the proposed algorithm enables the robotic arm to cooperate with other robotic arms and human production line workers. Compared with traditional robotic arms that perform single tasks, multi-task robotic arms are more adaptable because they have the following advantages: rapid switching between production lines, quick replacement of maintenance and repair tools, significant reduction in downtime, and the ability to use multiple end effectors. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 FIG. is a flowchart of a method for decision-making and control of a robotic arm based on a task tree algorithm according to an embodiment of the present invention;
[0062] Figure 2 FIG. is a flowchart of the conversion relationship between a decision tree and a task tree in a method for decision-making and control of a robotic arm based on a task tree algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0063] The present invention will be further described below in conjunction with specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.
[0064] As Figures 1 to 2 , a method for decision-making and control of a robotic arm based on a task tree algorithm provided by the present invention includes the following steps:
[0065] S100, obtaining object information through a robotic arm vision system, specifically including:
[0066] S101, image acquisition: The vision system first acquires images of the object through a high-resolution camera. These cameras are usually installed on the end effector of the robotic arm or at fixed positions in the working area, and can capture the appearance information of the object from multiple angles.
[0067] S102, Image preprocessing: The acquired images usually contain noise and unnecessary background information. To improve the recognition accuracy, the Gaussian filtering method is used to preprocess the images.
[0068] Gaussian filtering is a linear filter based on the Gaussian function, and its core is a Gaussian kernel. The weights of the Gaussian kernel are distributed according to the Gaussian distribution, with the weight of the central pixel being the largest and the weights of the surrounding pixels gradually decreasing. This characteristic enables Gaussian filtering to better retain the edge information of the image while smoothing the image.
[0069] The output I′(x,y) of Gaussian filtering can be expressed as:
[0070]
[0071] I(x,y): The pixel value (such as brightness or color value) of the original image at the coordinate (x,y).
[0072] I'(x,y): The new pixel value of the filtered image at the coordinate (x,y).
[0073] G(i,j): The weight value of the Gaussian filter at the offset (i,j).
[0074] i,j: The horizontal and vertical offsets relative to the central pixel (x,y) (in pixels). In actual calculations, the ranges of i and j usually take finite values (such as -3σ to 3σ) because the Gaussian function approaches zero at a distance.
[0075] Among them, G(i,j) is the Gaussian kernel, defined as:
[0076]
[0077] σ: The standard deviation of the Gaussian distribution, which controls the smoothness of the filter.
[0078] i 2 +j 2 : The square of the Euclidean distance from the current pixel (x+i,y+j) to the central pixel (x,y).
[0079] Normalization coefficient, ensuring that the total weight of the Gaussian kernel is 1.
[0080] S103, Feature recognition: The preprocessed images are further analyzed to identify the key features of the object, such as color features, shape features, and spatial position features.
[0081] S104, Pose Recognition: By detecting the key feature points (such as corner points and edge points) of an object and calculating the relative positional relationships of these feature points, the pose of the object is deduced.
[0082] S200, Implement Information Fusion Process by Encoding, Generate 2D Attribute Features from Object Information, and Import the 2D Attribute Features into the Task Tree Algorithm as Motion Data, Specifically Including:
[0083] S201, Feature Extraction: Use a Convolutional Neural Network (CNN) or other classification algorithms to extract the object and generate a class identifier. Here, the pre-trained YOLOv8 model is used. Use a more efficient convolutional neural network structure, increase the number of layers to capture more complex features, use depthwise separable convolutions and dilated convolutions to reduce the model's parameter quantity and computational amount, and simultaneously perform multi-scale key feature extraction. Use the PANet structure to effectively integrate features at different levels through cross-layer fusion, and adopt an improved loss function to improve accuracy and reduce false detections. Finally, generate a class identifier.
[0084] S202, Pose Extraction: Map the rotation angle and direction of the object into 2D features, and use the correspondence between 2D image points and 3D space points to obtain the pose features of the object through a geometric transformation model. For example, use two components of Euler angles (such as pitch angle and yaw angle) to represent the pose.
[0085] S203, Feature Encoding: Combine and encode the extracted features into structured 2D attribute features. These features can be represented as a 2D feature vector or matrix. For example:
[0086] motion = [Category1,Orientation1 Category2,Orientation2……]
[0087] Where: Category is the object class identifier, and Orientation is the pose feature.
[0088] S204, Data Conversion: Define a motion data structure to store the information in the feature vector and the parameters related to motion. Package the encoded feature vector into a motion data object and add the parameters required for the robot's motion, such as speed, acceleration, and gripping force, etc. Finally, convert the motion data into an instruction or data format that can be directly used by the robot control system. These motion data contain the key information required for the robot arm to complete the task. By using these attribute features as input data and importing them into the task tree algorithm. The reference code is as follows:
[0089]
[0090]
[0091] S300 enables the task tree algorithm to automatically record positions based on the imported motion data as the basis for action execution, specifically including:
[0092] S301, Analyze task logic: First, conduct a detailed analysis of the tasks of the robot arm to clarify the execution process and logical relationships of the tasks.
[0093] S302, Define task nodes: According to the task logic, define each node in the task tree. Node types usually include sequence nodes, selection nodes, conditional nodes, and action nodes. A sequence node indicates that child nodes are executed in sequence, a selection node indicates that child nodes are selectively executed according to conditions, a conditional node is used to determine whether a specific condition is met, and an action node represents a specific robot action.
[0094] S303, Build a tree structure: Organize the defined nodes into a tree structure according to the task logic.
[0095] S304, Import data: Import the transformed motion data into the task tree algorithm for preprocessing to ensure the accuracy and consistency of the data.
[0096] S305, Initialize position recording: In the task tree algorithm, set up a data structure to store the recorded position information. At the start of task execution, initialize this data structure to prepare for recording positions.
[0097] S306, Position update strategy: According to the execution process and logic of the task tree, set the update logic for position recording in the corresponding action nodes or conditional nodes.
[0098] S307, Build the basis for action execution: In subsequent action nodes, plan the motion trajectory of the robot based on the recorded position information and the position information of the placement area, and use it as the basis for action execution.
[0099] S400, Import the pre-set target actions and tasks into the task tree algorithm through a decision tree, specifically including:
[0100] S401, Define the decision tree structure: A decision tree is used to make decisions based on conditions, while a task tree is used to execute a series of actions.
[0101] S402, Import target actions and tasks: Import the pre-set target actions and tasks in the decision tree into the task tree. This can be achieved by defining actions and tasks as functions or objects.
[0102] Reference code for decision tree construction:
[0103]
[0104] S500 enables the task tree algorithm to perform action recognition and output an action sequence, specifically including:
[0105] S501, after the target action and task information from the decision tree module are imported into the task tree, the results of the decision tree are mapped to the action nodes and control nodes of the task tree. The complex task is decomposed into multiple subtasks and action units.
[0106] S502, dynamically adjust the structure or parameters of the task tree according to the target action and task information provided by the decision tree. This may include adding, deleting, or modifying nodes, as well as adjusting the execution conditions of the nodes.
[0107] S503, the task tree traverses the task tree through the depth-first search method. Starting from the starting node of the task tree, select an unvisited adjacent node to visit and mark it as visited. For each visited node, repeat the above steps until it is impossible to continue going deeper. When it is impossible to continue going deeper, backtrack to the previous node and try other unvisited adjacent nodes. Repeat the above steps until all task tree nodes are visited. Finally, identify all action nodes.
[0108] S504, collect the action names of all action nodes that meet the conditions, and at the same time optimize the action sequence using heuristic rules or algorithms, and finally output the action sequence.
[0109] The depth search pseudocode is as follows:
[0110]
[0111]
[0112] The code for traversing the task tree and collecting the action sequence is as follows:
[0113]
[0114] S600 enables the robotic arm to complete corresponding actions based on the output action sequence, specifically including:
[0115] S601, parse the action sequence: The output action sequence is usually a list containing a series of action instructions, and each action instruction may contain parameters such as action type, target position, speed, acceleration, etc. First, these instructions need to be parsed to extract the specific parameters of each action.
[0116] S602, initialize the robotic arm controller: Ensure that the robotic arm controller has been initialized and the robotic arm is in a ready state to receive and execute instructions.
[0117] S603, execute the action sequence: traverse the action sequence, call the corresponding robot arm control function according to the type and parameters of each action instruction, and execute the specific action.
[0118]
[0119]
[0120] The present invention provides a robot arm decision and control system based on a task tree algorithm, comprising:
[0121] An acquisition unit, used for acquiring object information through a robot arm vision system;
[0122] The encoding unit is used to encode and realize the information fusion process, and generate 2D attribute features from the object information;
[0123] An import unit, used to import 2D attribute features as motion data into the task tree algorithm, so that the task tree algorithm automatically records the position according to the imported motion data as the basis for action execution;
[0124] The recognition unit is used to import the target actions and tasks that have been set in advance into the task tree algorithm through the decision tree, so that the task tree algorithm can perform action recognition and output an action sequence;
[0125] The output unit is used to enable the robot arm to complete corresponding actions based on the output action sequence.
[0126] The robot arm decision and control system based on the task tree algorithm provided in the present application may also include: a memory and a processor; the memory is used to store instructions;
[0127] The processor is used to operate according to the instructions to execute the steps of the aforementioned robot arm decision-making and control method based on the task tree algorithm.
[0128] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned robot arm decision-making and control method based on a task tree algorithm.
[0129] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0130] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in multiple blocks.
[0131] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in multiple blocks.
[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in multiple blocks.
[0133] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.
Claims
1. A robot arm decision-making and control method based on a task tree algorithm, characterized in that Including: Obtain object information through the robotic arm vision system; Encode to implement the information fusion process, and generate 2D attribute features from the object information; Import the 2D attribute features as motion data into the task tree algorithm, so that the task tree algorithm automatically records the position according to the imported motion data, serving as the basis for action execution; Import the pre-set target actions and tasks into the task tree algorithm through a decision tree, so that the task tree algorithm performs action recognition and outputs an action sequence; Based on the output action sequence, the robotic arm completes the corresponding actions.
2. The robot arm decision-making and control method based on the task tree algorithm according to claim 1, wherein, The method for obtaining object information through the robotic arm vision system includes: Collect images of the object through a camera installed on the robotic arm; Preprocess the collected object images using the Gaussian filtering method; Analyze the preprocessed images to identify the key features of the object; Identify the posture of the object by detecting the key feature points of the object and calculating the relative position relationship of these feature points.
3. A robot arm decision-making and control method based on a task tree algorithm according to claim 2, characterized in that, The formula for preprocessing the collected object images using the Gaussian filtering method is as follows: where σ is the standard deviation of the Gaussian distribution, controlling the smoothness of the filter; i 2 +j 2 is the square of the Euclidean distance from the current pixel (x + i, y + j) to the central pixel (x, y); is the normalization coefficient to ensure that the total weight of the Gaussian kernel is 1; I(x, y) is the pixel value of the original image at the coordinate (x, y); I'(x, y) is the new pixel value of the filtered image at the coordinate (x, y); G(i, j) is the weight value of the Gaussian filter at the offset (i, j); i, j are the horizontal and vertical offsets relative to the central pixel (x, y).
4. A robot arm decision-making and control method based on a task tree algorithm according to claim 1, characterized in that, The method for encoding to implement the information fusion process and generating 2D attribute features from the object information includes: Extract the key features and posture features of the identified object; Combine and encode the extracted features into structured 2D attribute features, represented as a matrix, and the formula is as follows: motion = [Category1,Orientation1 Category2,Orientation2……] Where: Category represents the object category identifier, Orientation represents the posture feature, and motion represents the 2D attribute feature of the object; define a motion data structure for storing the information in the feature vector and the parameters related to the motion; Package the encoded feature vector into a motion data object and add the parameters required for the robot motion; Convert the motion data into an instruction or data format directly used by the robot control system.
5. A method for robot arm decision-making and control based on a task tree algorithm according to claim 1, characterized in that, The method for enabling the task tree algorithm to automatically record the position according to the imported motion data, serving as the basis for action execution, includes: Conduct a detailed analysis of the tasks of the robot arm to clarify the execution process and logical relationship of the tasks; According to the task logic, define each node in the task tree. The node types include sequence nodes, selection nodes, conditional nodes, and action nodes. Among them, the sequence node represents executing the sub-nodes in sequence, the selection node represents selectively executing the sub-nodes according to conditions, the conditional node is used to judge whether a specific condition is met, and the action node represents a specific robot action; Organize the defined nodes into a tree structure according to the task logic; Import the transformed motion data into the task tree algorithm for preprocessing; In the task tree algorithm, set up a data structure for storing the recorded position information. At the start of task execution, initialize this data structure to prepare for recording the position; According to the execution process and logic of the task tree, set the update logic for position recording in the corresponding action nodes or conditional nodes; In the subsequent action nodes, plan the motion trajectory of the robot according to the recorded position information and the position information of the placement area, and use it as the basis for action execution.
6. A robot arm decision and control method based on a task tree algorithm according to claim 1, characterized in that A method for importing pre-set target actions and tasks into a task tree algorithm through a decision tree, enabling the task tree algorithm to perform action recognition and output an action sequence, includes: After the target actions and task information of the decision tree are imported into the task tree, the results of the decision tree are mapped to the action nodes and control nodes of the task tree, decomposing complex tasks into multiple subtasks and action units; Dynamically adjust the structure or parameters of the task tree according to the target actions and task information provided by the decision tree; Traverse the task tree through the depth-first search method, starting from the starting node of the task tree, select an unvisited adjacent node for access, and mark it as visited; For each visited node, repeat the above steps until it is impossible to continue going deeper; When it is impossible to continue going deeper, backtrack to the previous node and try other unvisited adjacent nodes; Repeat the above steps until all task tree nodes are visited, and finally identify all action nodes; Collect the action names of all action nodes that meet the conditions, and at the same time optimize the action sequence using heuristic rules or algorithms, and output the action sequence.
7. A robot arm decision and control method based on a task tree algorithm according to claim 1, characterized in that, A method for enabling a robotic arm to complete corresponding actions based on the output action sequence, includes: Parse the action instructions in the action sequence and extract the specific parameters of each action. Enable the robotic arm controller to be initialized, and the robotic arm is in a ready state to receive and execute action instructions; Traverse the action sequence, and according to the type and parameters of each action instruction, call the corresponding robotic arm control function to execute the specific action.
8. A robot arm decision-making and control system based on a task tree algorithm, characterized in that, Includes: An acquisition unit for acquiring object information through the robotic arm vision system; An encoding unit for encoding to implement the information fusion process and generating 2D attribute features from the object information; An import unit for importing the 2D attribute features as motion data into the task tree algorithm, enabling the task tree algorithm to automatically record positions based on the imported motion data as the basis for action execution; An identification unit for importing pre-set target actions and tasks into the task tree algorithm through a decision tree, enabling the task tree algorithm to perform action recognition and output an action sequence; An output unit for enabling the robotic arm to complete corresponding actions based on the output action sequence.
9. A robot arm decision-making and control system based on a task tree algorithm, characterized in that, Includes a memory and a processor; The memory is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1-7.
10. 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 steps of the method according to any one of claims 1-7.