Instruction generation type collaborative robot decision processing method and system and health care robot

Abstract

The application belongs to the technical field of artificial intelligence, and provides an instruction generation type collaborative robot decision processing method, system and a health care robot, and the technical scheme is: receiving a to-be-executed task instruction, decoupling and splitting the to-be-executed task into a series of atomic tasks; fusing multi-dimensional dynamic information, calculating a dynamic trigger factor of each subtask in real time, constructing a dynamic trigger judgment matrix of each subtask based on the dynamic trigger factor of each subtask; and determining a subtask trigger decision result according to the dynamic trigger judgment matrix of each subtask and a decision judgment condition. When facing complex health care needs, the most suitable health care scheme and instruction are dynamically generated, and the individualization and efficiency of the service are ensured.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of artificial intelligence, and particularly relates to an instruction generation type collaborative robot decision processing method and system and a health care robot. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] The current mainstream health care robot system development mode mainly includes a standardized function kit mode, a closed integration mode and a limited customization mode. The standardized function kit mode can quickly deploy through pre-installed basic function modules, but lacks personalized adaptation capability and cannot meet the deep needs of diversified health care scenes. The closed integration mode can ensure system stability, but the expandability is severely limited and it is difficult to reconstruct the service process according to the special needs of institutions or individuals. The limited customization mode allows partial parameter adjustment, but still has significant shortcomings in complex task processing, multi-device collaboration and dynamic strategy generation, especially when it involves cross-domain service integration. The system response lag and insufficient execution accuracy are particularly prominent. SUMMARY

[0004] In order to solve at least one of the technical problems in the background art, the first aspect of the present application provides an instruction generation type collaborative robot decision processing method, which dynamically generates the most suitable health care scheme and instruction when facing complex health care needs, ensuring the individualization and efficiency of the service.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] The first aspect of the present application provides an instruction generation type collaborative robot decision processing method, comprising the following steps:

[0007] Receiving a to-be-executed task instruction, and decoupling and splitting the to-be-executed task into a series of atomic tasks;

[0008] Fusing multi-dimensional dynamic information, calculating the dynamic trigger factor of each subtask in real time, and constructing a dynamic trigger judgment matrix of each subtask based on the dynamic trigger factor of each subtask;

[0009] Determining the subtask trigger decision result according to the dynamic trigger judgment matrix of each subtask and the decision judgment condition.

[0010] Further, the to-be-executed task is decoupled and split into a series of atomic tasks according to a predefined workflow, including:

[0011] The to-be-executed task is taken as a parent task, corresponding splitting basis is defined according to actual rehabilitation training, the parent task is split according to the set splitting basis, first-level subtasks are obtained, it is judged whether the first-level subtasks meet the set splitting condition, if yes, the first-level subtasks are continuously split to obtain second-level subtasks, and the process is repeated until the set splitting condition is not met, and the subtasks of the current level are the final atomic tasks.

[0012] Further, the dynamic trigger factor of each subtask includes a semantic weight, a preset priority, a user state correlation and a historical behavior correlation.

[0013] Further, the dynamic trigger determination matrix is:

[0014] ,

[0015] ,

[0016] ,

[0017] ,

[0018] ,

[0019] ,

[0020] ,

[0021] wherein, is a dynamic trigger determination matrix, is a semantic weight, is a preset priority, is a user state correlation, is a historical behavior correlation, represents a current user input, represents a historical recognized intention, represents a user input of the current round, represents a user-defined emergency task, represents a user-defined high-priority task, represents a user-defined regular task, represents a user-defined low-priority task, represents a relationship density between a user state behavior and a current behavior, represents a gamma correlation, and an empirical value is adopted, represents a current user state behavior, represents a main behavior in a task scenario, represents a user-defined regular behavior, representing the last update behavior, representing the current input behavior of the user, representing the main learning factor, representing the secondary learning factor.

[0022] Further, the determination of the sub-task triggering decision result according to the dynamic triggering judgment matrix and the judgment condition of each sub-task comprises:

[0023] If the dot product value of the triggering matrix is greater than the set highest threshold value, the triggering type is active triggering;

[0024] If the dot product value of the triggering matrix is less than the set lowest threshold value, the triggering type is conditional triggering;

[0025] If the dot product value of the triggering matrix is between the highest and lowest threshold values, the order of execution is determined by the large model, so as to achieve conditional triggering.

[0026] Further, the method further comprises obtaining the execution state and effect data of the triggered sub-task, and correcting the dynamic triggering judgment matrix according to the execution state and effect data of the triggered sub-task.

[0027] Further, the decision judgment condition is represented as:

[0028]

[0029] Among them, representing the type of triggering, representing active triggering, representing conditional triggering, representing optimal order triggering, representing the highest active mode activation threshold value defined by the user.

[0030] Further, if the sub-task execution fails or the condition changes, the dynamic retry, skip, triggering replacement or remedial sub-task.

[0031] The second aspect of the present application provides an instruction generation type collaborative robot decision processing system, which dynamically generates the most suitable health care scheme and instruction when facing complex health care needs, and ensures the individualization and efficiency of the service.

[0032] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0033] The second aspect of the present application provides an instruction generation type collaborative robot decision processing system, which comprises:

[0034] The task decoupling module is used for receiving a to-be-executed task instruction, and decoupling and splitting the to-be-executed task into a series of atomic tasks;

[0035] ​A dynamic trigger judgment module is configured to fuse multi-dimensional dynamic information, calculate a dynamic trigger factor of each subtask in real time, and construct a dynamic trigger judgment matrix of each subtask based on the dynamic trigger factor of each subtask.

[0036] A decision generation module is configured to determine a subtask trigger decision result according to the dynamic trigger judgment matrix of each subtask and a decision judgment condition.

[0037] A third aspect of the present application provides a health-care robot, comprising a memory, a processor and a user interface.

[0038] The memory is configured to store a computer program.

[0039] The user interface is configured to interact with a user.

[0040] The processor is configured to read the computer program in the memory, and the processor implements the instruction generation type collaborative robot decision processing method when executing the computer program.

[0041] Compared with the prior art, the present application has the following beneficial effects:

[0042] The present application decouples complex to-be-executed tasks in combination with a self-defined workflow, and can dynamically generate a corresponding decision strategy in combination with a generated dynamic trigger judgment matrix, so as to cope with complex health-care scenarios, enhance the adaptability and flexibility of the system, and significantly improve the service efficiency and personalization level.

[0043] The advantages of the additional aspects of the present application will be partially given in the following description, partially become obvious from the following description, or be learned through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0044] The drawings accompanying the specification of the present application form a part thereof, serve to provide further understanding of the present application, and together with the description of the illustrative embodiments of the present application and the explanation thereof serve to explain the present application, and do not constitute an improper limitation of the present application.

[0045] Figure 1 is a flow chart of the instruction generation type collaborative robot decision processing method provided by the embodiment of the present application;

[0046] Figure 2 is a block diagram of the instruction generation type collaborative robot decision processing system provided by the embodiment of the present application;

[0047] Figure 3 is a structural schematic diagram of the health-care robot provided by the embodiment of the present application. DETAILED DESCRIPTION

[0048] The present application will be further described below in combination with the drawings and embodiments.

[0049] It should be noted that the following detailed description is illustrative only, and is intended to provide further description in order to provide a fuller enabling and comprehensive disclosure of the exemplary embodiments according to the present application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the application pertains.

[0050] It is also important to note that the use of the term 'illustrative' does not mean that the exemplary embodiments according to the present application are preferred or advantageous over other exemplary embodiments. It is to be understood that the phraseology and terminology employed herein are by way of description and illustration, and should not be construed to limit the scope of the exemplary embodiments according to the present application. As used herein, the singular form "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, steps, operations, elements, components and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components and / or groups thereof.

[0051] With the continuous development of health and wellness field, AI, big data, and robot services will certainly provide stronger empowerment for the health and wellness industry. The health and wellness collaborative robot system is an innovative achievement in the future service field. The present application deeply integrates three key technologies of workflow customization, complex problem decoupling, and AIGC instruction generation, providing customized and intelligent all-weather accompanying experience. Through advanced voice recognition technology and robot interconnection architecture, the system can accurately understand and execute the personalized needs of users, providing just the right support in rehabilitation training, daily life care, and emotional communication, further improving the interactive experience and service quality, and creating a more warm, safe, and efficient intelligent health and wellness environment for users.

[0052] The current mainstream health and wellness robot system development modes mainly include standardized function kit mode, closed integration mode, and limited customization mode. The standardized function kit mode quickly deploys through pre-installed basic function modules, but lacks personalized adaptation ability and cannot meet the deep needs of diversified health and wellness scenarios. The closed integration mode can ensure system stability, but its expandability is severely limited and it is difficult to reconstruct service processes according to the special needs of institutions or individuals. The limited customization mode allows partial parameter adjustment, but still has significant shortcomings in complex task processing, multi-device collaboration, and dynamic strategy generation. Especially when it comes to cross-field service integration, the system response lag and execution accuracy are particularly prominent.

[0053] The key of the present application technology lies in deeply integrating workflow customization, complex problem decoupling technology and AIGC instruction generation, to provide users with an unprecedented intelligent experience. Through user-defined workflow, the system can flexibly cope with diversified accompanying needs, while the built-in complex problem decoupling technology ensures efficient handling of complex scenarios. The system supports real-time voice recognition, by accessing the SDK document of the corresponding robot, based on the AIGC instruction generation technology, the workflow is accurately converted into executable instructions, to complete the robot operation task, and improve the operation convenience. In addition, the system can interconnect different robots for cooperation, to ensure all-around accompanying service. This integration not only enhances the adaptability and flexibility of the system, but also significantly improves the service efficiency and personalization level.

[0054] Please refer to Figure 1 , Figure 1 An instruction generation type collaborative robot decision processing method flow chart provided by an embodiment of the present application is shown, and the instruction generation type collaborative robot decision processing method specifically includes the following steps:

[0055] Step 1: receiving a task instruction to be executed, and decoupling and splitting the task to be executed into a series of atomic tasks;

[0056] Specifically, the following steps are included:

[0057] Step 101, receiving a task instruction to be executed, and analyzing to obtain the corresponding task to be executed;

[0058] In this embodiment, the received task instruction includes but is not limited to natural language interaction, preset workflow triggering or other system events;

[0059] When the received task instruction is natural language interaction, the input instruction can be a voice command, text input and multi-round dialogue; natural language processing (NLP) technology can be used to analyze and understand the user's natural language, and the content of the task to be executed is parsed;

[0060] The preset workflow trigger can be a timing task, a condition trigger and a chain action, etc.;

[0061] Other system events, such as hardware signals, etc.;

[0062] It should be noted that no matter which task instruction is used, the corresponding analysis method is used to analyze the corresponding instruction to obtain the corresponding task to be executed;

[0063] Step 102, decoupling and splitting the task to be executed into a series of atomic tasks according to the predefined workflow;

[0064] In this embodiment, first, the to-be-executed task is decoupled and split into a series of atomic tasks according to the predefined workflow and the self-defined task architecture. When the task architecture is formulated, the to-be-executed task is taken as a parent task, the corresponding splitting basis is defined according to the actual rehabilitation training, the parent task is split according to the set splitting basis, the first-level child task is obtained, it is judged whether the first-level child task meets the set splitting condition, if yes, the first-level child task is continuously split to obtain the second-level child task, and the like is repeated until the set splitting condition is not met, and the child task of the current level is the final atomic task.

[0065] It should be noted that the set splitting basis can be defined according to the actual rehabilitation training characteristics, for example, according to the body part, the rehabilitation target, the type and attribute of the training, the difficulty level and the like;

[0066] Body part: the rehabilitation target is for a specific body part (such as upper limbs, lower limbs, core, hands) or functional system (such as motor system, sensory system, balance system, cardiopulmonary system, speech / swallowing system, cognitive system). For example, to improve lower limb function, the subtasks are quadriceps strength training, ankle dorsiflexion range of motion training, and gait balance training.

[0067] Functional target: the final purpose of rehabilitation is to restore or improve specific functional activity ability (such as walking, transfer, dressing, eating, communication). For example, to improve self-care ability, the subtasks are independent completion of bed-to-chair transfer training, independent dressing training (upper limb function + coordination), and independent eating training (hand fine motor + coordination).

[0068] Training type: rehabilitation training includes various types, each type has its specific target and method. For example, according to the “upper limb rehabilitation training” task, the body part splitting basis is decoupled into: arm lifting movement, arm flexion and extension movement, hand grip training, real-time monitoring of heart rate and blood pressure, and the like.

[0069] Difficulty level: the basis is: rehabilitation training needs to follow the principle of gradual progression, from easy to difficult, from simple to complex, from low intensity to high intensity, and from stable environment to unstable environment.

[0070] It should be noted that the predefined workflow is flexibly customized according to the specific needs of the user based on the robot in the health care activities and operations; the workflow can be easily created and adjusted through a graphical interface or natural language instructions, to ensure that the robot can efficiently and accurately execute various tasks. Through this self-defined function, a workflow that meets the user's needs can be designed, improving the flexibility and adaptability of the robot. The workflow self-defined function is not only suitable for simple tasks, but also can cope with complex health care scenarios, so that the robot can execute a series of preset steps and actions.

[0071] Step 2: fuse the obtained multi-dimensional dynamic information, and calculate the dynamic trigger factor of each sub-task in real time, and construct a dynamic trigger judgment matrix for each sub-task based on the dynamic trigger factor of each sub-task;

[0072] After splitting a series of atomic tasks, the sub-tasks are not executed according to a predefined static sequence. In this embodiment, multi-dimensional dynamic information is fused, the dynamic trigger factor of each sub-task is calculated in real time according to the user-defined workflow, and the trigger judgment basis for each sub-task is generated. These trigger bases are used to form a dynamic trigger judgment matrix, and the dynamic trigger judgment matrix is used to intelligently determine when to trigger which sub-task (or sub-tasks) and the order of triggering.

[0073] Specifically, the following steps are included:

[0074] Step 201, obtaining multi-dimensional dynamic information;

[0075] In this embodiment, the multi-dimensional information obtained from various sensors, monitoring systems, message queues, databases, APIs, etc. is used to affect the task trigger decision over time. Specifically, the historical recognition intention, task priority, user state correlation and historical behavior correlation data are included.

[0076] Step 202, fuse multi-dimensional dynamic information, and calculate the dynamic trigger factor of each sub-task in real time;

[0077] In this embodiment, the dynamic trigger factor of each sub-task includes semantic weight, preset priority, user state correlation and historical behavior correlation;

[0078] Specifically, the semantic weight is represented as:

[0079] ,

[0080] wherein, represents the current user input representation, represents the historical recognition intention, such as the intention to perform the health care training last time, represents the user input in this round;

[0081] It can be understood that if the user input in this round is a voice instruction, the voice instruction is converted into text, and the text is vectorized after embedding to obtain the current user input representation;

[0082] The preset priority is represented as:

[0083] ,

[0084] wherein, represents a user-defined emergency task, represents a user-defined high-priority task, represents a user-defined regular task, represents a low-priority task;

[0085] user state relevance represents a relationship density between the user state behavior and the current behavior,

[0086] ,

[0087] ,

[0088] wherein, represents a relationship density between the user state behavior and the current behavior, represents a gamma relevance, with an empirical value, represents the current user state behavior, represents a main behavior in the task scenario, represents a user-defined regular behavior;

[0089] historical behavior relevance represents a relationship density between the user state behavior and the current behavior,

[0090] ,

[0091] ,

[0092] wherein, represents the last updated behavior, represents the user's current input behavior, represents a main learning factor, represents a secondary learning factor.

[0093] Step 203, constructing a dynamic trigger decision matrix for each subtask based on the dynamic trigger factor of each subtask;

[0094] dynamic trigger decision matrix represents a relationship density between the user state behavior and the current behavior,

[0095] ,

[0096] wherein, is a semantic weight, is a preset priority, is a user state relevance, is a historical behavior relevance.

[0097] Based on the dynamic trigger decision matrix, the triggered subtasks and the trigger order of the subtasks can be determined; this embodiment makes decisions based on the comprehensive weight of multi-dimensional information, rather than simply sequential execution or conditional branching.

[0098] Step 3: Determine the sub-task trigger decision result according to the dynamic trigger decision matrix and judgment conditions of each sub-task;

[0099] If the dot product value of the trigger matrix is greater than the set highest threshold, the trigger type is active trigger;

[0100] If the dot product value of the trigger matrix is less than the set lowest threshold, the trigger type is conditional trigger;

[0101] If the dot product value of the trigger matrix is between the highest and lowest thresholds, the order of execution is determined by the large model, so as to achieve conditional trigger.

[0102] The formula is:

[0103] ,

[0104] Among them, represents the type of trigger, represents active trigger, represents conditional trigger, represents optimal order trigger, represents the highest active mode activation threshold defined by the user;

[0105] For example: active trigger: the current user instruction only inputs "start training", but the system real-time monitors that the user's heart rate data abnormally fluctuates (user state), and is associated with the historical report that the user's upper limb strength recovery is slow and the upper limb training set in the rehabilitation stage has high priority (preset task priority), so the arm lifting movement sub-task closely related to upper limb recovery will be actively triggered.

[0106] Conditional trigger: during the training process, when the heart rate and blood pressure real-time monitoring sub-task detects that the user's breathing frequency exceeds the preset safety threshold, the system will immediately pause the current upper limb training sub-task, and trigger the execution of the breathing adjustment auxiliary sub-task (which is another atomic task after decomposition) according to the preset safety rule priority.

[0107] Optimal order trigger: for a "morning routine" task, it may include sub-tasks such as washing, taking medicine, and simple exercises. The system will dynamically adjust and trigger the optimal execution order of sub-tasks according to the user's historical behavior preferences (such as being used to washing first), the current environment (such as whether the bathroom is occupied), and the priority of the medicine reminder.

[0108] Step 4: Obtain the execution state and effect data of the triggered sub-task, and correct the dynamic trigger decision matrix according to the execution state and effect data of the triggered sub-task;

[0109] In this embodiment, the execution state and effect data of the triggered subtask includes whether it is completed, completion quality, user feedback, physiological index change and other related data, which is fed back to step 2 to update the dynamic trigger judgment matrix, such as adjusting the task weight and updating the user state, forming a dynamic closed-loop control link.

[0110] Further, if the subtask execution fails or the condition changes, the system can dynamically decide to retry, skip or trigger alternative / remedial subtasks.

[0111] The present application can perform well in various health care scenarios through AIGC instruction generation technology. Whether it is daily care, rehabilitation training or emotional companionship, it can provide high-quality services. This technology not only improves the interactivity and intelligence of robots, but also enables robots to handle complex instructions and tasks. In the face of complex health care needs, AIGC instruction generation technology can dynamically generate the most suitable health care plan and instructions according to the specific circumstances of the user, ensuring personalized and efficient service.

[0112] Please refer to Figure 2 , Figure 2 The present application provides an instruction generation type collaborative robot decision processing system block diagram, which comprises:

[0113] The task decoupling module 201 is used for receiving a to-be-executed task instruction, and decoupling and splitting the to-be-executed task into a series of atomic tasks.

[0114] The dynamic trigger judgment module 202 is used for fusing multi-dimensional dynamic information, calculating the dynamic trigger factor of each subtask in real time, and generating the trigger judgment basis for each subtask, and constructing the dynamic trigger judgment matrix of each subtask based on the trigger judgment basis of each subtask.

[0115] The decision generation module 203 is used for determining the subtask trigger decision result according to the dynamic trigger judgment matrix of each subtask and the decision judgment condition.

[0116] In the task decoupling module, the to-be-executed task is decoupled and split into a series of atomic tasks according to a predefined workflow, including:

[0117] The to-be-executed task is taken as a parent task, the corresponding splitting basis is defined according to the actual rehabilitation training, the parent task is split according to the set splitting basis, the first-level subtask is obtained, it is judged whether the first-level subtask meets the set splitting condition, if yes, the first-level subtask is continuously split to obtain the second-level subtask, and so on, until the set splitting condition is not met, and the subtask of the current level is the final atomic task.

[0118] In the dynamic trigger judgment module, the trigger judgment of each subtask is based on semantic weight, preset priority, user state relevance and historical behavior relevance.

[0119] The dynamic trigger judgment matrix is:

[0120] ,

[0121] ,

[0122] ,

[0123] ,

[0124] ,

[0125] ,

[0126] ,

[0127] wherein, is the dynamic trigger judgment matrix, is the semantic weight, is the preset priority, is the user state relevance, is the historical behavior relevance, represents the current user input, represents the historical recognized intent, represents the user input in this round, represents the user-defined urgent task, represents the user-defined high-priority task, represents the user-defined regular task, represents the user-defined low-priority task, represents the relationship density between the user state behavior and the current behavior, represents the gamma relevance, which is an empirical value, represents the current user state behavior, represents the main behavior in the task scenario, represents the user-defined regular behavior, represents the behavior updated last time, represents the current user input behavior, represents the main learning factor, represents the secondary learning factor.

[0128] In the decision generation module, the subtask trigger decision result is determined according to the dynamic trigger judgment matrix and the judgment condition of each subtask, which includes:

[0129] If the dot product value according to the matrix is greater than the set highest threshold value, the trigger type is active trigger;

[0130] If the dot product value according to the matrix is less than the set lowest threshold value, the trigger type is conditional trigger;

[0131] If the dot product value according to the matrix is between the highest and lowest threshold values, the order of execution is determined by the large model, so as to achieve conditional trigger.

[0132] The decision-making judgment condition is represented as:

[0133] ,

[0134] Among them, represents the type of trigger, represents active trigger, represents conditional trigger, represents optimal order trigger, represents the highest active mode activation threshold value defined by the user.

[0135] It should be noted that the specific implementation mode of the instruction generation type collaborative robot decision processing system of the embodiment of the application is similar to the specific implementation mode of the instruction generation type collaborative robot decision processing method of the embodiment of the application. For details, please refer to the description in the method part. In order to reduce redundancy, this part will not be repeated here.

[0136] Please refer to Figure 3 , Figure 3 A kangling robot block diagram provided by an embodiment of the application is shown, the kangling robot includes a memory 302, a processor 301 and a user interface 303;

[0137] The memory 302 is used to store a computer program;

[0138] The user interface 303 is used to realize interaction with the user;

[0139] The processor 301 is used to read the computer program in the memory 302, and the processor 301 executes the computer program to realize the following steps:

[0140] Receiving a to-be-executed task instruction, decoupling and splitting the to-be-executed task into a series of atomic tasks;

[0141] Fusing multi-dimensional dynamic information, calculating the dynamic trigger factor of each subtask in real time, and generating the trigger judgment basis for each subtask, constructing the dynamic trigger judgment matrix of each subtask based on the trigger judgment basis of each subtask;

[0142] The subtask trigger decision result is determined based on the dynamic trigger judgment matrix and decision judgment conditions of each subtask.

[0143] It should be noted that the specific implementation method of the embodiment of the present invention is similar to the specific implementation method of the instruction-generating collaborative robot decision-making processing method of the embodiment of the present invention. They belong to the same inventive concept, solve the same technical problems, and achieve the same technical effects. Please refer to the description of the method part for details, and the similarities will not be repeated here.

[0144] Among them, Figure 3 In the present invention, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits such as one or more processors represented by processor 301 and memory represented by memory 302. The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore not further described herein. The bus interface provides the interface. Processor 301 is responsible for managing the bus architecture and general processing, while memory 302 can store data used by processor 301 when performing operations.

[0145] The processor 301 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor 301 may also adopt a multi-core architecture.

[0146] It should be noted that the division of units in the embodiments of the present application is schematic and is merely a logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0147] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can 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 can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) that contain computer-usable program code.

[0148] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions specified in the flowchart block or blocks. Figure 1 The flowchart and / or block diagram in the variation of the present application can describe any device, system or computer program (including software and / or firmware) implementing processes, algorithms, methodologies, etc. in flowchart form and / or block diagram form. Figure 1 The flowchart and / or block diagram in the variation of the present application can describe any device, system or computer program (including software and / or firmware) implementing processes, algorithms, methodologies, etc. in flowchart form and / or block diagram form.

[0149] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions specified in the flowchart block or blocks. Figure 1 The flowchart and / or block diagram in the variation of the present application can describe any device, system or computer program (including software and / or firmware) implementing processes, algorithms, methodologies, etc. in flowchart form and / or block diagram form. Figure 1 The flowchart and / or block diagram in the variation of the present application can describe any device, system or computer program (including software and / or firmware) implementing processes, algorithms, methodologies, etc. in flowchart form and / or block diagram form.

[0150] Obviously, numerous modifications and variations are possible in accordance with the teachings of the present application without departing from the scope of the present application. Accordingly, while the present application has been described above with particularity, and reference being made to the drawings attached, it is to be understood that the application is not limited to the embodiments disclosed, but is instead to be understood to cover all alternatives falling within the scope of the application and equivalents thereof.

Claims

1. A command-generating collaborative robot decision process method, characterized by, The method comprises the following steps: Receiving a to-be-executed task instruction, and decoupling and splitting the to-be-executed task into a series of atomic tasks; Fusing multi-dimensional dynamic information, and calculating a dynamic trigger factor of each subtask in real time, and constructing a dynamic trigger judgment matrix of each subtask based on the dynamic trigger factor of each subtask; The dynamic trigger factor of each subtask comprises semantic weight, preset priority, user state correlation and historical behavior correlation; and the dynamic trigger judgment matrix is: ， ， ， ， ， ， ， wherein, is a dynamic trigger decision matrix, is a semantic weight, is a preset priority, is a user state relevance, is a historical behavior relevance, represents a representation of a current user input, represents a historical recognized intent, represents a user input of a current round, represents a user defined urgent task, represents a user defined high priority task, represents a user defined regular task, represents a user defined low priority task, represents a relationship density of a user state behavior and a current behavior, represents a gamma relevance, with an empirical value, represents a current user state behavior, represents a primary behavior in a task scenario, represents a user defined regular behavior, represents a behavior of a last update, represents a user current input behavior, represents a primary learning factor, represents a secondary learning factor; According to the dynamic trigger judgment matrix of each subtask and a decision judgment condition, a subtask trigger decision result is determined, which comprises: if the dot product value of the dynamic trigger judgment matrix is greater than a set highest threshold value, the trigger type is active trigger; if the dot product value of the dynamic trigger judgment matrix is less than a set lowest threshold value, the trigger type is conditional trigger; and if the dot product value of the dynamic trigger judgment matrix is between the highest and lowest threshold values, the order of execution is determined through a large model, so that conditional trigger is achieved.

2. The instruction generating collaborative robotic decision-making process method of claim 1, wherein, Decoupling and splitting the to-be-executed task into a series of atomic tasks according to a predefined workflow, which comprises: Taking the to-be-executed task as a parent task, defining a corresponding splitting basis according to actual rehabilitation training, splitting the parent task according to the set splitting basis to obtain a first-level subtask, judging whether the first-level subtask meets a set splitting condition, if yes, continuing to split the first-level subtask to obtain a second-level subtask, and so on, until the set splitting condition is not met, and the subtask of the current level is the final atomic task.

3. The instruction generating co-bot decision-making process method of claim 1, wherein, The method further comprises obtaining execution state and effect data of the triggered subtask, and correcting the dynamic trigger judgment matrix according to the execution state and effect data of the triggered subtask.

4. The instruction generating co-bot decision-making process method of claim 1, wherein, The decision judgment condition is represented as: ， wherein, represents the type of trigger, represents a proactive trigger, represents a conditional trigger, represents an optimal order trigger, represents a user-defined highest proactive mode activation threshold.

5. The instruction generating co-bot decision-making process method of claim 1, wherein, If the subtask execution fails or the condition changes, the subtask is dynamically retried, skipped, triggered to be replaced or remedied.

6. A command-generating collaborative robotic decision processing system, characterized by It comprises: A task decoupling module for receiving a to-be-executed task instruction and decoupling and splitting the to-be-executed task into a series of atomic tasks; A dynamic trigger judgment module for fusing multi-dimensional dynamic information, calculating a dynamic trigger factor of each subtask in real time, and constructing a dynamic trigger judgment matrix of each subtask based on the dynamic trigger factor of each subtask; the dynamic trigger factor of each subtask comprises semantic weight, preset priority, user state correlation and historical behavior correlation; and the dynamic trigger judgment matrix is: ， ， ， ， ， ， ， wherein, is a dynamic trigger decision matrix, is a semantic weight, is a preset priority, is a user state relevance, is a historical behavior relevance, represents a representation of a current user input, represents a historical recognized intent, represents a user input of a current round, represents a user defined urgent task, represents a user defined high priority task, represents a user defined regular task, represents a user defined low priority task, represents a relationship density of a user state behavior to a current behavior, represents a gamma relevance, with an empirical value, represents a current user state behavior, represents a primary behavior in a task scenario, represents a user defined regular behavior, represents a behavior of a last update, represents a user current input behavior, represents a primary learning factor, represents a secondary learning factor; A decision generation module for determining a subtask trigger decision result according to the dynamic trigger judgment matrix of each subtask and a decision judgment condition, which comprises: if the dot product value of the dynamic trigger judgment matrix is greater than a set highest threshold value, the trigger type is active trigger; if the dot product value of the dynamic trigger judgment matrix is less than a set lowest threshold value, the trigger type is conditional trigger; and if the dot product value of the dynamic trigger judgment matrix is between the highest and lowest threshold values, the order of execution is determined through a large model, so that conditional trigger is achieved.

7. A health-care robot, characterized by, It comprises a memory, a processor and a user interface; The memory is used for storing a computer program; The user interface is used for realizing interaction with a user; and The processor is used for executing the computer program. The processor is configured to read a computer program in the memory, and when the processor executes the computer program, an instruction generation type collaborative robot decision processing method according to any one of claims 1-5 is implemented.

Images