Task fusion and planning method and system for robot multi-task scene

Abstract

The invention relates to the technical field of robots, discloses a robot multi-task scene-oriented task fusion and planning method and system, improves the execution efficiency, safety response capability and decision transparency of a robot in a multi-task concurrent scene, and overcomes the defects of task processing stiffness, resource waste, emergency response lag and the like in the prior art. According to the method, through a multi-task understanding and fusion mechanism driven by a large model, priority quantification, semantic / space / path triple fusion and dynamic re-planning are carried out on a plurality of heterogeneous tasks received by a robot, and an interpretable decision chain is generated; therefore, high-efficiency, high-safety, high-adaptability and high-transparency task execution is realized, and the method is obviously superior to an existing task coverage type or queuing type processing mode.

Description

Technical Field

[0001] This invention relates to the technical field of intelligent robot task management, specifically to an intelligent decision-making method and system based on multi-task fusion and dynamic priority scheduling, which is applicable to scenarios where home service robots handle multiple heterogeneous tasks simultaneously in complex home environments. Background Technology

[0002] With the development of artificial intelligence and service robot technology, home service robots have gradually entered daily life, undertaking various tasks such as item retrieval, calling for help, environmental cleaning, and security inspection. However, existing home robots generally adopt two traditional strategies when handling multiple concurrent tasks: task coverage and task queuing.

[0003] In a task-override mechanism, when a new task arrives, the robot immediately interrupts the currently executing task and switches to the new task, discarding or shelving the original task. While this approach offers rapid response, it lacks assessment of task importance and context, potentially leading to the unexpected termination of critical tasks. For example, if the robot is performing a gas leak detection (a high-safety-level task), and the user issues a command to "find the remote control," a task-override robot might abandon the safety detection and search for the remote control, creating a safety hazard.

[0004] In a task queuing mechanism, the robot queues all tasks in the order they are received and executes them sequentially, regardless of their urgency, spatial location, or semantic relevance. While this method ensures no task is missed, it is inefficient and cannot adapt to dynamically changing home environments. For example, if the robot first receives the task of "tidying the desk in the study" and then receives the instruction to "check if there is a smoke alarm in the kitchen," even if the latter involves personal safety, it must wait until the desk is tidyed before handling it, delaying emergency response.

[0005] To illustrate the shortcomings of existing technologies more specifically, consider the following typical scenario: three things happen simultaneously in a user's home—(1) the kitchen smoke alarm is triggered; (2) a child spills a water glass in the living room and needs to be cleaned up; (3) the user asks the robot to retrieve glasses from the bedroom. If existing home robots adopt a task queuing system, they may retrieve glasses from the bedroom first, then clean up the water stains, and finally respond to the smoke alarm, which seriously violates the principle of safety priority. If a task overlay system is adopted, the task of retrieving glasses may be interrupted on the way to the bedroom, and the task of retrieving glasses may be lost. Then, the task may be interrupted again by the smoke alarm, resulting in frequent task switching, fragmented execution, low overall efficiency, and poor user experience.

[0006] In summary, existing home robots have significant shortcomings in multitasking: they cannot guarantee the priority execution of safety or emergency tasks (i.e., lack of prioritization); tasks with similar content or close proximity must be executed separately, leading to duplicate paths and increased energy consumption (i.e., inability to integrate tasks); they do not consider the robot's path alignment within the physical space (i.e., lack of spatial and path optimization); and they lack high-level task planning, hindering dynamic adjustments (i.e., lack of decoupling between planning and execution). These problems result in robots exhibiting slow response times, resource waste, increased safety risks, and low human-robot trust in real-world home scenarios, making it difficult to meet the complex, dynamic, and highly reliable needs of home services. Therefore, a new technological solution capable of achieving intelligent multitasking integration, dynamic scheduling, and explainable decision-making is urgently needed. Summary of the Invention

[0007] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form to prepare for the more detailed descriptions that follow.

[0008] The purpose of this invention is to solve the above-mentioned problems and to provide a task fusion and planning method and system for robot multi-task scenarios. This significantly improves the robot's execution efficiency, safety response capability and decision transparency in multi-task concurrent scenarios, and effectively overcomes the defects of rigid task processing, resource waste and delayed emergency response in the prior art.

[0009] The technical solution of this invention is as follows: This invention discloses a task fusion and planning method for multi-task scenarios of robots, the method comprising: Step 1: The robot receives the task input; Step 2: Parse the received task to obtain the key attributes of the task and store them in a structured manner; Step 3: Based on the key attributes of the tasks obtained in Step 2, calculate the priority score for each task and sort them. Step 4: Perform task fusion processing based on priority ranking results; Step 5: Generate a task execution plan based on the merged task set to clarify the execution order and logical relationship of each task; Step 6: Inject the task execution plan into the main body model in the form of a structured message; Step 7: The main model generates an action sequence based on the injected task execution plan and the current environmental state; Step 8: The robot executes instructions based on the action sequence output by the main body model.

[0010] According to an embodiment of the task fusion and planning method for robot multi-task scenarios of the present invention, the key attributes of the task in step 2 include: task type, urgency, main responsibilities, task time and task location.

[0011] In one embodiment of the task fusion and planning method for robot multi-task scenarios according to the present invention, the priority score in step 3 is calculated as follows: Priority = w1 Security + W2 Timeliness + W3 Energy consumption +W4 Efficiency + w5 Main responsibilities; Among them, w1 to w5 are the corresponding weights, and the weight values ​​are dynamically configured.

[0012] According to an embodiment of the task fusion and planning method for robot multi-task scenarios of the present invention, the task fusion processing in step 4 further includes: Content fusion: Merge similar or complementary tasks, prioritizing the retention of execution logic for high-priority tasks; Spatial integration: Integrating tasks that occur in the same location or adjacent areas. If multiple tasks exist in the same space, the execution order of the tasks after entering the space is determined according to their priority. Path fusion: Additional execution of tasks that are completed incidentally along the way during path planning. When merging tasks along the way, high-priority tasks are scheduled first, and low-priority tasks are inserted as incidental tasks.

[0013] According to an embodiment of the task fusion and planning method for multi-task scenarios of robots according to the present invention, the generation of the task execution plan in step 5 is based on the fused task set and is derived by reasoning using thought chain, wherein the task execution plan includes a textual execution scheme and a planning rationale explanation.

[0014] According to an embodiment of the task fusion and planning method for robot multi-task scenarios of the present invention, the action sequence output by the main large model in step 8 drives the robot to execute instructions through function calls or model control protocols.

[0015] According to an embodiment of the task fusion and planning method for robot multi-task scenarios of the present invention, the method further includes: Step 9: During the robot's task execution, a dynamic feedback mechanism is triggered based on real-time environmental changes or task completion status to adjust and replan the current task set.

[0016] According to an embodiment of the task fusion and planning method for robot multi-task scenarios of the present invention, the method further includes: Step 10: Record the results of each task fusion and execution as historical data. The historical data is used to train or optimize the intelligence level of future task planning and execution.

[0017] This invention also discloses a task fusion and planning system for multi-task scenarios of robots, the system comprising: The task input module is used by the robot to receive task input; The task parsing module is used to parse the received tasks, obtain the key attributes of the tasks, and store them in a structured manner. The priority evaluation module is used to calculate priority scores and sort tasks based on the key attributes of the tasks obtained from the task parsing module. The task fusion module is used to fuse tasks based on priority ranking results; The task planning and generation module is used to generate task execution plans based on the merged task set, so as to clarify the execution order and logical relationship of each task; The planning injection module is used to inject the task execution plan into the main body model in the form of a structured message; An action sequence generation module is used by the main large model to generate an action sequence based on the injected task execution plan and the current environmental state. The task execution module is used by the robot to execute instructions based on the action sequence output by the main body model.

[0018] According to an embodiment of the task fusion and planning system for multi-task scenarios of robots based on the present invention, the system further includes: The dynamic feedback module is used to trigger a dynamic feedback mechanism during the robot's task execution based on real-time environmental changes or task completion status, thereby adjusting and replanning the current task set.

[0019] According to an embodiment of the task fusion and planning system for multi-task scenarios of robots based on the present invention, the system further includes: The experience learning module is used to record the results of each task fusion and execution as historical data. This historical data is used to train or optimize the intelligence level of future task planning and execution.

[0020] The present invention also discloses an electronic device, the electronic device including a controller, the controller including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes the program stored in the memory, it implements the steps of the task fusion and planning method for robot multi-task scenarios as described above.

[0021] The present invention also discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the task fusion and planning method for robot multi-task scenarios as described above.

[0022] The present invention also discloses a computer program product, which, when executed by a processor, implements the steps of the task fusion and planning method for robot multi-task scenarios as described above.

[0023] Compared with existing technologies, the present invention has the following advantages: The present invention uses a large model-driven multi-task understanding and fusion mechanism to perform priority quantification, semantic / spatial / path triple fusion and dynamic replanning on multiple heterogeneous tasks received by the robot, and generates an interpretable decision chain; thereby achieving high-efficiency, high-security, strong adaptability and high-transparency task execution, which is significantly better than existing task coverage or queuing processing methods.

[0024] In detail, the technical features and corresponding technical effects of the present invention are listed below.

[0025] 1. Calculation of task priority: By comprehensively considering the weights of five dimensions—security, timeliness, energy consumption, efficiency, and primary responsibilities—task priorities are dynamically calculated to ensure that security-related, emergency-related, and core responsibility tasks are prioritized for scheduling and execution, thereby guaranteeing the rationality of system operation and the timeliness of user needs.

[0026] 2. Task Integration: The system integrates content, space, and path at three levels, and uses task priority calculations to determine the execution order and hierarchy after integration. This not only reduces redundant paths and actions but also ensures that important tasks are executed first in multi-task parallel scenarios, achieving a balance between efficiency and security.

[0027] 3. Task planning, generation, and injection: By leveraging Chain-of-Thought (CoT), high-level task planning and rationale are output and injected into the main model as system messages. This decouples task planning from action generation, giving the system greater flexibility and controllability, while ensuring the transparency and interpretability of the planning process.

[0028] 4. Action generation and execution: The main model generates specific action sequences based on high-level planning and controls their execution through function calls and MCP (Model-Control-Policy). This ensures reliable implementation from planning to action, supports modular expansion, and facilitates integration with external execution systems or robotic platforms.

[0029] 5. Dynamic feedback and re-integration: During task execution, dynamic feedback and action adjustments are triggered based on environmental changes (such as path blockage or target loss). This enables the system to respond instantly to anomalies that occur during execution, ensuring the stability and robustness of task execution.

[0030] 6. Experience-based learning mechanism: The system stores historical fusion results, execution paths, user feedback, and efficiency performance, forming an iteratively optimized experience base. Through long-term operation, the system can continuously optimize fusion and scheduling quality, achieving continuous evolution and self-optimization, thereby improving overall performance and user experience.

[0031] 7. Task interruption, insertion, and re-fusion mechanism: When a new task arrives during execution, the system does not suspend or save it independently. Instead, it directly merges the new task with the remaining parts of the currently unfinished tasks into a joint task set. The priority of this joint set is recalculated and a new fusion plan is executed. Subtasks can be adjusted to execute according to the new priority order, but all subtasks belong to the scheduling scope under the same joint task set.

[0032] This mechanism can achieve: seamless integration of new and existing tasks, avoiding complex suspension / resumption state management; timely response of high-efficiency / high-security tasks through instant reordering; maintaining the continuity and consistency of task execution, avoiding task loss or inconsistent states; and simplifying system scheduling logic and improving operational efficiency and reliability. Attached Figure Description

[0033] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0034] Figure 1 A flowchart of an embodiment of the task fusion and planning method for multi-task scenarios of robots according to the present invention is shown.

[0035] Figure 2 A schematic diagram of an embodiment of the task fusion and planning system for multi-task scenarios of robots according to the present invention is shown.

[0036] Figure 3A structural diagram of an electronic device according to the present invention is shown. Detailed Implementation

[0037] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should be noted that the aspects described below with reference to the accompanying drawings and specific embodiments are merely exemplary and should not be construed as limiting the scope of protection of the present invention in any way.

[0038] Figure 1 The flowchart illustrates an embodiment of the task fusion and planning method for multi-tasking scenarios of robots according to the present invention. Please refer to... Figure 1 The implementation steps of the method in this embodiment are described in detail below.

[0039] Step 1: Task input (user command / environment trigger).

[0040] The robot receives voice or text commands from users (such as "go to the kitchen to check the gas"), or tasks automatically triggered by environmental sensors (such as a smoke alarm). These inputs constitute the original set of tasks to be processed.

[0041] Step 2: Task analysis.

[0042] The input task is semantically understood and structurally extracted to identify key attributes such as task type, urgency, main responsibilities, task time and location. This information is stored in a structured manner to provide basic data for subsequent processing (such as priority calculation and task fusion).

[0043] The task type can be categorized as life service, security monitoring, environmental maintenance, etc.; the urgency level can be classified according to the time sensitivity and degree of danger of the task; the main responsibility refers to the core task category of the robot in the design scenario, such as the main responsibility of a tour guide robot being to guide and greet guests, and the main responsibility of a cleaning robot being to clean; the task time is the deadline or suggested execution time of the task; and the task location is the physical location that needs to be reached to execute the task.

[0044] Step 3: Priority calculation.

[0045] A comprehensive priority score is calculated for each task based on a multi-dimensional evaluation. Tasks are then ranked according to their priority scores, with higher-priority tasks given priority consideration. Evaluation dimensions include: safety (whether it involves personal or property safety), timeliness (whether an immediate response is required), energy consumption (energy required for execution), efficiency (whether it can be combined with other tasks), and primary responsibility (whether it falls within the robot's main responsibility scope).

[0046] For example, priority scores are calculated using the following formula: Priority = w1 Security + W2 Timeliness + W3 Energy consumption +W4 Efficiency + w5 Main responsibilities The weights w1 to w5 can be dynamically configured according to the application scenario. For example, in a home security scenario, security weight w1 is higher; in a logistics and delivery scenario, timeliness w2 and efficiency w4 have higher weights.

[0047] Step 4: Task integration.

[0048] Based on the priority ranking results, the tasks are processed in three categories: Content integration: Merge similar or complementary tasks (e.g., merge "finding a pen" and "tidying up the desk" into "tidying up the study"), and prioritize the retention of the execution logic of high-priority tasks; Spatial integration: Integrate tasks that occur in the same location or adjacent areas to reduce repetitive back-and-forth; if multiple tasks exist in the same space, determine the order of task execution after entering the space based on priority (e.g., "clean the living room" and "pick up the toys on the sofa"). Path merging: This involves adding tasks that can be completed incidentally along the route during path planning, thereby reducing energy consumption and time overhead. When merging tasks along the same path, high-priority tasks are scheduled first, while low-priority tasks are inserted as incidental tasks.

[0049] The integration process follows the principle of "high-priority tasks taking the lead" to ensure that critical tasks are not weakened.

[0050] Step 5: Task planning generation.

[0051] Based on the merged task set, a high-level, structured task execution plan is generated (e.g., "go to the kitchen first → then to the living room → finally to the bedroom"), clarifying the execution order and logical relationship of each task, and forming a schedulable action framework.

[0052] The task execution plan is generated based on the merged task set and uses CoT (Chain-of-Thought) reasoning. The task execution plan is not simply a list of tasks, but a structured planning result derived step-by-step, specifically including: Textualized execution plan: Based on the reasoning results of the thought chain, the overall execution order, spatial access order and path arrangement of multiple integrated tasks are clarified to form a high-level task plan that can be executed and invoked. Explanation of Planning Rationale: Synchronously output the planning rationale corresponding to the task execution plan. By explicitly presenting the task priority judgment, fusion decision basis and sorting logic involved in the CoT reasoning process, it explains why the current task combination and execution order are adopted, thereby improving the explainability and decision transparency in multi-task concurrency and fusion scenarios.

[0053] Chain of Thought (CoT) is a large-scale model reasoning method that generates the reasoning process or rationale for each step while simultaneously generating the output. By explicitly displaying the model's thought chain, the rationality, reliability, and interpretability of decisions or plans can be improved. In this invention, the CoT concept is applied to the robot planning and generation process, enabling the system to record the selection criteria and logical reasoning at each step of the action or task generation, thereby assisting the large model in generating more reliable action sequences.

[0054] Step 6: System message injection.

[0055] The generated high-level task plan is injected into the main model in the form of a structured message (such as a system message) as its decision context for subsequent action generation.

[0056] System messages are a type of context injection in large model inference, used to provide constraint information to the main large model.

[0057] Step 7: Processing the main large model.

[0058] By utilizing pre-trained Large Language Models (LLM) or Embodied Intelligence Models, specific action sequences are generated based on the injected task execution plan and the current environmental state.

[0059] Step 8: The robot executes the instructions.

[0060] The action sequences output by the large model drive the robot to perform complex operations, such as navigation, grasping, and voice interaction, through function calls or model-control-policies (MCPs).

[0061] Function call: This refers to the mechanism by which a large model calls external function interfaces to implement specific actions or operations during inference. In robotics scenarios, function calls are used to map abstract instructions generated by a large model into specific control commands that the robot can execute (such as motor rotation, robotic arm movements, voice broadcasts, etc.), driving the robot to actually perform tasks.

[0062] MCP (Model Control Protocol) refers to the communication and control protocol between a large model and the execution system, used to transmit motion sequences or task execution instructions. MCP can standardize the interaction between the output of the large model and the robot control system, enabling the motion generation results to directly drive the robot hardware or middleware execution.

[0063] Step 9: Dynamic feedback and re-integration.

[0064] This is a preferred step. During the robot's task execution, the system continuously acquires information on environmental changes (such as the arrival of new tasks or the appearance of obstacles) or task completion status (such as "remote controller found"). If a change occurs, a dynamic feedback mechanism is triggered, entering the task fusion process in step 4, adjusting and replanning the current task set to achieve closed-loop adaptation.

[0065] The strategies for adjustment and replanning are as follows: New task arrives: If a new task is received at any time, the system will immediately parse the new task and calculate its priority; Merge with the remaining parts of the current task: The new task is directly merged with the remaining parts of the current unfinished task (including sub-steps that need to be completed later) to form a new joint task set; the priority and fusion strategy (content / space / path) of this joint set will be recalculated and a new high-level plan will be generated. Execution order adjustment: Within a set of joint tasks, there may be a new order among the subtasks (for example, subtasks with higher safety or timeliness are placed first). However, the original tasks are not "suspended" and wait to be resumed. Instead, they become part of the new joint tasks and are scheduled and executed by the new plan. Abnormal situation handling: If the path is blocked or the target item is missing, the system will trigger a dynamic feedback mechanism and enter the task fusion process in step 4 to regenerate the plan.

[0066] Step 10: Learning from experience.

[0067] This is an optimal step. Record the results of each task fusion and execution, including historical data such as fusion strategy, path selection, time consumption, energy consumption, and user feedback, and use this data to train or optimize the intelligence level of future task planning and execution.

[0068] Figure 2 The structure of an embodiment of the task fusion and planning system for multi-task scenarios of robots according to the present invention is shown. Figure 2As shown, the system in this embodiment includes: a task input module, a task parsing module, a priority evaluation module, a task fusion module, a task planning generation module, a planning injection module, an action sequence generation module, a task execution module, a dynamic feedback module, and an experience learning module.

[0069] The task input module is used for the robot to receive task input.

[0070] The robot receives voice or text commands from users (such as "go to the kitchen to check the gas"), or tasks automatically triggered by environmental sensors (such as a smoke alarm). These inputs constitute the original set of tasks to be processed.

[0071] The task parsing module is used to parse the received tasks, obtain the key attributes of the tasks, and store them in a structured manner.

[0072] The input task is semantically understood and structurally extracted to identify key attributes such as task type, urgency, main responsibilities, task time and location. This information is stored in a structured manner to provide basic data for subsequent processing (such as priority calculation and task fusion).

[0073] The task type can be categorized as life service, security monitoring, environmental maintenance, etc.; the urgency level can be classified according to the time sensitivity and degree of danger of the task; the main responsibility refers to the core task category of the robot in the design scenario, such as the main responsibility of a tour guide robot being to guide and greet guests, and the main responsibility of a cleaning robot being to clean; the task time is the deadline or suggested execution time of the task; and the task location is the physical location that needs to be reached to execute the task.

[0074] The priority evaluation module is used to calculate priority scores and sort tasks based on the key attributes of the tasks obtained from the task parsing module.

[0075] A comprehensive priority score is calculated for each task based on a multi-dimensional evaluation. Tasks are then ranked according to their priority scores, with higher-priority tasks given priority consideration. Evaluation dimensions include: safety (whether it involves personal or property safety), timeliness (whether an immediate response is required), energy consumption (energy required for execution), efficiency (whether it can be combined with other tasks), and primary responsibility (whether it falls within the robot's main responsibility scope).

[0076] For example, priority scores are calculated using the following formula: Priority = w1 Security + W2 Timeliness + W3 Energy consumption +W4 Efficiency + w5 Main responsibilities The weights w1 to w5 can be dynamically configured according to the application scenario. For example, in a home security scenario, security weight w1 is higher; in a logistics and delivery scenario, timeliness w2 and efficiency w4 have higher weights.

[0077] The task fusion module is used to merge tasks based on priority ranking results.

[0078] Based on the priority ranking results, the tasks are processed in three categories: Content integration: Merge similar or complementary tasks (e.g., merge "finding a pen" and "tidying up the desk" into "tidying up the study"), and prioritize the retention of the execution logic of high-priority tasks; Spatial integration: Integrate tasks that occur in the same location or adjacent areas to reduce repetitive back-and-forth; if multiple tasks exist in the same space, determine the order of task execution after entering the space based on priority (e.g., "clean the living room" and "pick up the toys on the sofa"). Path merging: This involves adding tasks that can be completed incidentally along the route during path planning, thereby reducing energy consumption and time overhead. When merging tasks along the same path, high-priority tasks are scheduled first, while low-priority tasks are inserted as incidental tasks.

[0079] The integration process follows the principle of "high-priority tasks taking the lead" to ensure that critical tasks are not weakened.

[0080] The task planning and generation module is used to generate task execution plans based on the merged task set, so as to clarify the execution order and logical relationship of each task.

[0081] Based on the merged task set, a high-level, structured task execution plan is generated (e.g., "go to the kitchen first → then to the living room → finally to the bedroom"), clarifying the execution order and logical relationship of each task, and forming a schedulable action framework.

[0082] The task execution plan is generated based on the merged task set and uses CoT (Chain-of-Thought) reasoning. The task execution plan is not simply a list of tasks, but a structured planning result derived step-by-step, specifically including: Textualized execution plan: Based on the reasoning results of the thought chain, the overall execution order, spatial access order and path arrangement of multiple integrated tasks are clarified to form a high-level task plan that can be executed and invoked. Explanation of Planning Rationale: Synchronously output the planning rationale corresponding to the task execution plan. By explicitly presenting the task priority judgment, fusion decision basis and sorting logic involved in the CoT reasoning process, it explains why the current task combination and execution order are adopted, thereby improving the explainability and decision transparency in multi-task concurrency and fusion scenarios.

[0083] Chain of Thought (CoT) is a large-scale model reasoning method that generates the reasoning process or rationale for each step while simultaneously generating the output. By explicitly displaying the model's thought chain, the rationality, reliability, and interpretability of decisions or plans can be improved. In this invention, the CoT concept is applied to the robot planning and generation process, enabling the system to record the selection criteria and logical reasoning at each step of the action or task generation, thereby assisting the large model in generating more reliable action sequences.

[0084] The planning injection module is used to inject task execution plans into the main model in the form of structured messages.

[0085] The generated high-level task plan is injected into the main model in the form of a structured message (such as a system message) as its decision context for subsequent action generation.

[0086] System messages are a type of context injection in large model inference, used to provide constraint information to the main large model.

[0087] The action sequence generation module is used by the main model to generate action sequences based on the injected task execution plan and the current environment state.

[0088] By utilizing pre-trained Large Language Models (LLM) or Embodied Intelligence Models, specific action sequences are generated based on the injected task execution plan and the current environmental state.

[0089] The task execution module is used by the robot to execute instructions based on the action sequence output by the main body model.

[0090] The action sequences output by the large model drive the robot to perform complex operations, such as navigation, grasping, and voice interaction, through function calls or model-control-policies (MCPs).

[0091] Function call: This refers to the mechanism by which a large model calls external function interfaces to implement specific actions or operations during inference. In robotics scenarios, function calls are used to map abstract instructions generated by a large model into specific control commands that the robot can execute (such as motor rotation, robotic arm movements, voice broadcasts, etc.), driving the robot to actually perform tasks.

[0092] MCP (Model Control Protocol) refers to the communication and control protocol between a large model and the execution system, used to transmit motion sequences or task execution instructions. MCP can standardize the interaction between the output of the large model and the robot control system, enabling the motion generation results to directly drive the robot hardware or middleware execution.

[0093] The dynamic feedback module is a preferred module used to trigger a dynamic feedback mechanism during the robot's task execution based on real-time environmental changes or task completion status, thereby adjusting and replanning the current task set.

[0094] During the robot's task execution, the system continuously acquires information on environmental changes (such as the arrival of new tasks or the appearance of obstacles) or task completion status (such as "remote controller found"). If a change occurs, a dynamic feedback mechanism is triggered, entering the task fusion module to adjust and replan the current task set, achieving closed-loop adaptation.

[0095] The strategies for adjustment and replanning are as follows: New task arrives: If a new task is received at any time, the system will immediately parse the new task and calculate its priority; Merge with the remaining parts of the current task: The new task is directly merged with the remaining parts of the current unfinished task (including sub-steps that need to be completed later) to form a new joint task set; the priority and fusion strategy (content / space / path) of this joint set will be recalculated and a new high-level plan will be generated. Execution order adjustment: Within a set of joint tasks, there may be a new order among the subtasks (for example, subtasks with higher safety or timeliness are placed first). However, the original tasks are not "suspended" and wait to be resumed. Instead, they become part of the new joint tasks and are scheduled and executed by the new plan. Abnormal situation handling: If the path is blocked or the target item is missing, the system will trigger a dynamic feedback mechanism and enter the task fusion process to regenerate the plan.

[0096] The experience learning module is a preferred module used to record the results of each task fusion and execution as historical data. This historical data is used to train or optimize the intelligence level of future task planning and execution.

[0097] refer to Figure 3 As shown, the present invention also discloses an electronic device, which includes a controller. The controller includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.

[0098] Memory is used to store computer programs; the processor is used to execute the programs stored in memory, enabling functions such as... Figure 1 The steps of the task fusion and planning method embodiment for robot multi-task scenarios are shown.

[0099] Furthermore, this invention also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the following... Figure 1 The steps of the task fusion and planning method embodiment for robot multi-task scenarios are shown.

[0100] Furthermore, this invention also discloses a computer program product, which, when executed by a processor, implements the following: Figure 1 The steps of the task fusion and planning method embodiment for robot multi-task scenarios are shown.

[0101] The following two specific examples illustrate the specific application scenarios of the above method.

[0102] Example 1 (Preferred Example): Task: A: Go find a pen; B: Go to the study and tell Xiao Huang it's time to eat.

[0103] The task fusion output plan is as follows: {"Plan", "Navigate to the study, notify Xiao Huang to go eat, and check along the way for a pen. If a pen is found, grab it and continue notifying the task. If not found, search for the pen in other possible locations.", "Reason", "The two tasks have the potential for fusion in terms of space and content. The pen is likely in the study, so prioritizing the study can avoid extra paths and improve overall efficiency."}

[0104] Execution results: By integrating the pen-retrieval task with the study room notification task in terms of space and path, the robot completes two objectives in one path planning, reducing repeated back-and-forth trips and significantly improving execution efficiency.

[0105] Example 2 (Extended Example): Task: C: Clean the living room; D: Check the kitchen smoke detector.

[0106] The task fusion output plan is as follows: {"Plan", "First, check the smoke alarm in the kitchen. If there is a problem, immediately notify the homeowner. Then, clean the living room.", "Reason", "Checking the smoke alarm is a safety-related task and cannot be merged with other tasks. It should be prioritized to ensure safety. Cleaning the living room is a routine maintenance task and can be done after safety is confirmed."}

[0107] Execution effect: The system identifies security tasks by prioritizing them and places them first to ensure the timeliness of security tasks, while also taking into account the subsequent execution of environmental maintenance tasks, thus achieving a balance between security and efficiency.

[0108] Therefore, compared with existing robot task scheduling and execution, the beneficial effects of this invention are reflected in: 1. Priority Guarantee: Through multi-dimensional weighted calculation, we ensure that safety and emergency tasks are always given priority.

[0109] 2. Multi-level integration: Supports the integration of content, space and path at three levels. Combined with priority information, it minimizes redundant actions and paths and improves execution efficiency.

[0110] 3. Decoupling of planning and execution: A system message injection mechanism is adopted to decouple high-level task planning from low-level action generation, thereby enhancing the system's flexibility and scalability.

[0111] 4. Dynamic interruption and recovery: When a new task appears during execution, the system merges the new task with the remaining part of the current task in real time for unified replanning and scheduling. This can quickly respond to new requirements while maintaining the continuity of existing tasks, avoiding complex suspension / recovery processes.

[0112] 5. Strong adaptability: When environmental anomalies or temporary user instructions occur, the system can immediately re-integrate and re-plan, maintaining the robustness of task scheduling.

[0113] 6. Continuous optimization: By accumulating historical execution data through experience learning mechanisms, the intelligence level of task fusion and scheduling can be improved in the long term, thereby achieving continuous optimization of system performance.

[0114] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0115] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0116] The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein can be implemented or performed using a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The general-purpose processor may be a microprocessor, but in alternatives, it may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors cooperating with a DSP core, or any other such configuration.

[0117] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of both. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor such that the processor can read and write information to / from the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside as discrete components in the user terminal.

[0118] In one or more exemplary embodiments, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functionality may be stored or transmitted as one or more instructions or code on or through a computer-readable medium. A computer-readable medium includes both computer storage media and communication media, encompassing any medium that facilitates the transfer of a computer program from one location to another. A storage medium may be any available medium accessible to a computer. By way of example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and is accessible to a computer. Any connection is also legitimately referred to as a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of a medium. As used in this article, disk and disc include compact discs (CDs), laser discs, optical discs, digital multi-purpose discs (DVDs), floppy disks, and Blu-ray discs. Disks typically reproduce data magnetically, while discs reproduce data optically using lasers. Combinations of these should also be included within the scope of computer-readable media.

[0119] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A task fusion and planning method for multi-task scenarios of robots, characterized in that, The methods include: Step 1: The robot receives the task input; Step 2: Parse the received task to obtain the key attributes of the task and store them in a structured manner; Step 3: Based on the key attributes of the tasks obtained in Step 2, calculate the priority score for each task and sort them. Step 4: Perform task fusion processing based on priority ranking results; Step 5: Generate a task execution plan based on the merged task set to clarify the execution order and logical relationship of each task; Step 6: Inject the task execution plan into the main body model in the form of a structured message; Step 7: The main model generates an action sequence based on the injected task execution plan and the current environmental state; Step 8: The robot executes instructions based on the action sequence output by the main body model; The fusion process in step 4 further includes: Content fusion: Merge similar or complementary tasks, prioritizing the retention of execution logic for high-priority tasks; Spatial integration: Integrating tasks that occur in the same location or adjacent areas. If multiple tasks exist in the same space, the execution order of the tasks after entering the space is determined according to their priority. Path fusion: Additional execution of tasks that are completed incidentally along the way during path planning. When merging tasks along the way, high-priority tasks are scheduled first, and low-priority tasks are inserted as incidental tasks.

2. The task fusion and planning method for robot multi-task scenarios according to claim 1, characterized in that, The key attributes of the task in step 2 include: task type, urgency, main responsibilities, task time, and task location.

3. The task fusion and planning method for multi-task scenarios of robots according to claim 1, characterized in that, The priority score in step 3 is calculated as follows: Priority = w1 Security + W2 Timeliness + W3 Energy consumption +W4 Efficiency + w5 Main responsibilities; Among them, w1 to w5 are the corresponding weights, and the weight values ​​are dynamically configured.

4. The task fusion and planning method for multi-task scenarios of robots according to claim 1, characterized in that, The task execution plan generated in step 5 is based on the fused task set and is derived using thought chain reasoning. The task execution plan includes a textual execution scheme and a rationale for the planning.

5. The task fusion and planning method for multi-task scenarios of robots according to claim 1, characterized in that, The action sequence output by the main large model in step 8 drives the robot to execute instructions through function calls or model control protocols.

6. The task fusion and planning method for robot multi-task scenarios according to claim 1, characterized in that, The method also includes: Step 9: During the robot's task execution, a dynamic feedback mechanism is triggered based on real-time environmental changes or task completion status to adjust and replan the current task set.

7. The task fusion and planning method for robot multi-task scenarios according to claim 1, characterized in that, The method also includes: Step 10: Record the results of each task fusion and execution as historical data. The historical data is used to train or optimize the intelligence level of future task planning and execution.

8. A task fusion and planning system for multi-task scenarios of robots, characterized in that, The system includes: The task input module is used by the robot to receive task input; The task parsing module is used to parse the received tasks, obtain the key attributes of the tasks, and store them in a structured manner. The priority evaluation module is used to calculate priority scores and sort tasks based on the key attributes of the tasks obtained from the task parsing module. The task fusion module is used to fuse tasks based on priority ranking results; The task planning and generation module is used to generate task execution plans based on the merged task set, so as to clarify the execution order and logical relationship of each task; The planning injection module is used to inject the task execution plan into the main body model in the form of a structured message; An action sequence generation module is used by the main large model to generate an action sequence based on the injected task execution plan and the current environmental state. The task execution module is used by the robot to execute instructions based on the action sequence output by the main body model; The task fusion process in the task fusion module further includes: Content fusion: Merge similar or complementary tasks, prioritizing the retention of execution logic for high-priority tasks; Spatial integration: Integrating tasks that occur in the same location or adjacent areas. If multiple tasks exist in the same space, the execution order of the tasks after entering the space is determined according to their priority. Path fusion: Additional execution of tasks that are completed incidentally along the way during path planning. When merging tasks along the way, high-priority tasks are scheduled first, and low-priority tasks are inserted as incidental tasks.

9. The task fusion and planning system for multi-task scenarios of robots according to claim 8, characterized in that, The system also includes: The dynamic feedback module is used to trigger a dynamic feedback mechanism during the robot's task execution based on real-time environmental changes or task completion status, thereby adjusting and replanning the current task set.

10. The task fusion and planning system for multi-task scenarios of robots as described in claim 8, characterized in that, The system also includes: The experience learning module is used to record the results of each task fusion and execution as historical data. This historical data is used to train or optimize the intelligence level of future task planning and execution.

11. An electronic device, characterized in that, The electronic device includes a controller, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus. Memory, used to store computer programs; When a processor executes a program stored in memory, it implements the steps of the task fusion and planning method for robot multi-task scenarios as described in any one of claims 1-7.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the task fusion and planning method for robot multi-task scenarios as described in any one of claims 1-7.

13. A computer program product, characterized in that, When the program is executed by the processor, it implements the steps of the task fusion and planning method for robot multi-task scenarios as described in any one of claims 1-7.

Images