Complex task execution method, device and equipment of industrial agent and medium

Abstract

The invention discloses a complex task execution method, device and equipment for an industrial agent and a medium, and relates to the technical field of agent control, and the method comprises the steps: decomposing a to-be-executed task through employing a preset task disassembly rule, and obtaining at least one to-be-executed sub-task, determining a target agent according to the at least one sub to-be-executed task; based on the at least one sub-to-be-executed task and a pre-trained primitive task model library, a target primitive model of the to-be-executed task is determined, the target primitive model comprises at least one sub-primitive model, and the sub-primitive models are in one-to-one correspondence with the sub-to-be-executed tasks; and controlling the target agent to operate based on the target primitive model to complete the to-be-executed task. According to the method, a complex task can be decomposed into a plurality of simple tasks to be processed, the training cost of the model is reduced, the application degree of the primitive model is improved, the primitive model can deal with various and complex use environments, and the flexibility and practicability of an intelligent agent and the use experience of a user are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent body control technology, and in particular to a complex task execution method, device, equipment and medium for an industrial intelligent body. Background Art

[0002] With the development of science and technology, intelligent agents have been applied in industries such as industry and construction to replace humans in completing certain operational tasks, thereby improving production efficiency.

[0003] At present, most intelligent agents use teaching programming, that is, intelligent agents need to acquire the ability to perform tasks through imitation learning. In the process of imitation learning, intelligent agents need a large amount of teaching data and manually labeled information. The more complex the task, the larger the teaching data set required, the longer the training cycle, and the higher the requirements for computer computing resources. The training and use costs of the intelligent agents are high. Secondly, intelligent agents trained by teaching methods cannot perform tasks in new environments or any new tasks that have not been taught. Their popularity and flexibility are low. Summary of the invention

[0004] The present invention provides a complex task execution method, device, equipment and medium for an industrial intelligent body, which can reduce the training cost of the model, improve the applicability of the primitive model, enable it to cope with diverse and complex usage environments, and enhance the flexibility, practicality and user experience of the intelligent body.

[0005] According to one aspect of the present invention, a method for executing complex tasks of an industrial agent is provided, the method comprising:

[0006] Decomposing the task to be executed by using the pre-set task decomposition rules to obtain at least one sub-task to be executed, and determining the target intelligent agent according to the at least one sub-task to be executed;

[0007] Based on at least one sub-task to be executed and a pre-trained primitive task model library, determine a target primitive model of the task to be executed, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model corresponds to the sub-task to be executed one by one;

[0008] Control the target agent to run based on the target primitive model to complete the task to be executed.

[0009] According to another aspect of the present invention, a complex task execution device of an industrial intelligent body is provided, and the complex task execution device of the industrial intelligent body is used to implement the complex task execution method of the industrial intelligent body in any embodiment of the present invention, and the device includes:

[0010] A task decomposition module is used to decompose the task to be executed by using a preset task decomposition rule to obtain at least one sub-task to be executed, and determine a target intelligent agent according to the at least one sub-task to be executed;

[0011] A model determination module, used to determine a target primitive model of a task to be executed based on at least one sub-task to be executed and a pre-trained primitive task model library, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model corresponds to the sub-task to be executed one by one;

[0012] The task execution module is used to control the target agent to run based on the target primitive model to complete the task to be executed.

[0013] According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0014] at least one processor; and a memory communicatively coupled to the at least one processor;

[0015] The memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can execute the complex task execution method of the industrial intelligent body in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for implementing the complex task execution method of the industrial intelligent agent in any embodiment of the present invention when the computer instructions are executed by a processor.

[0017] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the complex task execution method of the industrial intelligent agent according to any embodiment of the present invention is implemented.

[0018] The complex task execution method of the industrial intelligent body of the present invention comprises: using a pre-set task disassembly rule to decompose the task to be executed, obtain at least one sub-task to be executed, and determine the target intelligent body according to the at least one sub-task to be executed; based on the at least one sub-task to be executed and the pre-trained primitive task model library, determine the target primitive model of the task to be executed, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model and the sub-task to be executed correspond one to one; control the target intelligent body to run based on the target primitive model to complete the task to be executed. The technical scheme of the present invention decomposes the task to be executed and divides it into multiple task segments, that is, decomposes the complex task into multiple simple tasks, matches them in the primitive task model library according to the task segments, obtains multiple sub-primitive models, and then splices the sub-primitive models to obtain the target primitive model. The model library only needs to train and update the commonly used primitive models, which can not only reduce the training cost of the model, but also improve the applicability of the primitive model, so that it can cope with diverse and complex usage environments, and improve the flexibility, practicality and user experience of the intelligent body.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 It is a flowchart of a complex task execution method of an industrial intelligent body provided by the present invention;

[0022] Figure 2 It is a flowchart of another complex task execution method of an industrial intelligent agent provided by the present invention;

[0023] Figure 3 It is a structural schematic diagram of a complex task execution device of an industrial intelligent body provided by the present invention;

[0024] Figure 4 It is a structural schematic diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", "initial", "intermediate", "candidate", "target", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Figure 1 This is a flow chart of a complex task execution method for an industrial intelligent body provided by the present invention. The present invention is applicable to low-cost control of intelligent bodies to perform work tasks in new environments, perform new tasks, etc. The method can be performed by a complex task execution device for an industrial intelligent body provided by the present invention. The device can be implemented in the form of hardware and / or software. In a specific embodiment, the device can be integrated into an electronic device. The following embodiments will be described by taking the device integrated into an electronic device as an example. Figure 1 , the method specifically comprises the following steps:

[0028] S101. Decompose the task to be executed by using a preset task decomposition rule to obtain at least one sub-task to be executed, and determine a target intelligent agent according to the at least one sub-task to be executed.

[0029] Task decomposition refers to breaking down a complex task into a series of smaller, easier to manage and execute subtasks. The benefits of decomposition are that it can improve task processing efficiency, reduce errors and ensure that each part of the task receives proper attention. The task to be performed of the present invention can be understood as a relatively complex job that needs to be performed by an agent. The pre-set task decomposition rules can be understood as the decomposition method of complex tasks set in advance, including but not limited to decomposition according to task execution time, decomposition according to task execution location, decomposition according to task type and decomposition according to the connection relationship of tasks. The sub-task to be performed can be understood as the sub-task obtained after decomposing the task to be performed. The target agent can be understood as an agent (including but not limited to a robot) that is more suitable for completing the task to be performed in the working environment, for example, an idle agent closest to the task to be performed, an agent that consumes the least resources to complete the task to be performed, etc. The present invention does not limit this.

[0030] For example, assuming that the task to be executed is a flange screw tightening task, and the task decomposition rule is to abstract common industrial operation skills into multiple iterations of simple tasks, then the sub-task to be executed is a single screw tightening task, and the decomposition result of the flange screw tightening task is n repeated tasks of a single screw tightening task. The specific number of n is related to the detailed parameters of the flange screw tightening task. The target agent can be the agent closest to the first screw tightening task or the agent that consumes the least resources to execute the flange screw tightening task. The advantage of this setting is that complex tasks can be decomposed into multiple repetitive simple tasks, reducing the difficulty of processing the tasks to be executed.

[0031] It is worth noting that the disassembled sub-tasks to be executed may be in the same position or in different positions. For example, if the task to be executed is to tighten screw 1, screw 2, screw 4 and screw 3 in sequence, and the sub-tasks to be executed are to tighten screw 1, tighten screw 2, tighten screw 4 and tighten screw 3, then the sub-tasks to be executed are in different positions. If the task to be executed is to tighten flange screw X, and the sub-task to be executed is to tighten screw X 5 times, then the sub-tasks to be executed are in the same position.

[0032] S102: Determine a target primitive model of the task to be executed based on at least one sub-task to be executed and a pre-trained primitive task model library.

[0033] Among them, the pre-trained primitive task model library can be understood as a set of primitive models of multiple simple tasks that have been trained and stored in advance, the target primitive model can be understood as an algorithm for implementing the task to be executed, and the target primitive model includes at least one sub-primitive model. The sub-primitive model can be understood as an algorithm for implementing the sub-task to be executed, and the sub-primitive model and the sub-task to be executed correspond one to one.

[0034] Specifically, based on at least one sub-task to be executed and a pre-trained primitive task model library, determining the target primitive model of the task to be executed can be understood as determining the sub-primitive model of each sub-task to be executed, and determining the target primitive model based on each sub-primitive model.

[0035] Exemplarily, assuming that the sub-tasks to be executed include task 1, task 2 and task 3, the primitive task model library includes sub-primitive model 1, sub-primitive model 2, sub-primitive model 3, sub-primitive model 4 and sub-primitive model 5, task 1 corresponds to sub-primitive model 4, task 2 corresponds to sub-primitive model 2, and task 3 corresponds to sub-primitive model 1, then the target primitive model is determined to be a model set consisting of sub-primitive model 4, sub-primitive model 2 and sub-primitive model 1. It is worth noting that if the sub-task to be executed is related to multiple sub-primitive models, multiple sub-primitive models can be trained to obtain a sub-primitive model with better adaptability to the sub-task to be executed. The purpose of this setting is to ensure the success rate and quality of task execution. The advantage of this setting is to process complex tasks through a combination of simple tasks and improve the generation efficiency of complex task strategies.

[0036] S103, controlling the target intelligent agent to run based on the target primitive model to complete the task to be executed.

[0037] Specifically, controlling the target intelligent agent to run based on the target primitive model to complete the task to be executed can be understood as controlling the target intelligent agent to run in sequence based on each sub-primitive model in the target primitive model to complete each sub-task to be executed.

[0038] Exemplarily, the target primitive model is a model set consisting of sub-primitive model 4, sub-primitive model 2 and sub-primitive model 1, then the target intelligent agent is controlled to run sub-primitive model 4, sub-primitive model 2 and sub-primitive model 1 in sequence, and the next model is run after a model runs successfully. If the run fails, a warning is issued to prompt the staff to deal with the run abnormality in time. The successful run of sub-primitive model 1 proves that the task to be executed is executed successfully.

[0039] The technical solution of the present invention decomposes the task to be executed into multiple task segments, that is, decomposes a complex task into multiple simple tasks, matches them in the primitive task model library according to the task segments, obtains multiple sub-primitive models, and then splices the sub-primitive models to obtain the target primitive model. The model library only needs to train and update the commonly used primitive models, which can not only reduce the training cost of the model, but also improve the applicability of the primitive model, so that it can cope with diverse and complex usage environments, and improve the flexibility, practicality of the intelligent body and the user experience.

[0040] Figure 2This is a flow chart of another complex task execution method of an industrial intelligent body provided by the present invention. The present invention is applicable to low-cost control of intelligent bodies to perform work tasks in new environments, perform new tasks, etc. The method can be performed by the complex task execution device of the industrial intelligent body provided by the present invention. The device can be implemented in the form of hardware and / or software. In a specific embodiment, the device can be integrated in an electronic device. The following embodiments will be described by taking the device integrated in an electronic device as an example. Figure 2 , the method specifically comprises the following steps:

[0041] S201. Determine at least one sub-task, and the priority and location information of at least one sub-task according to the task requirements of the task to be executed.

[0042] Among them, the task requirement can be understood as the task profile of the task to be performed, which is used to indicate the characteristic information of the task to be performed, for example, screwing screw 1, screw 2, screw 4 and screw 3 in sequence, screwing flange screw X, etc. The priority of the sub-task is used to indicate the execution order of the tasks included in the task requirement, and the location information of the sub-task is used to indicate the execution location of the tasks included in the task requirement, that is, the coordinate information of the location of the screw.

[0043] Specifically, assuming that the task requirement is to tighten screw 1, screw 2, screw 4 and screw 3 in sequence, the sub-tasks are tightening screw 1, tightening screw 2, tightening screw 3 and tightening screw 4, and the priorities of the sub-tasks are tightening screw 1, tightening screw 2, tightening screw 4 and tightening screw 3 in sequence, and the location information of the sub-tasks is the coordinate information of screw 1, screw 2, screw 4 and screw 3 in the working environment. The advantage of this setting is that detailed information of the task to be executed can be quickly obtained.

[0044] S202: Determine a task execution sequence number and a task execution position of at least one sub-work task based on the priority and position information of at least one sub-work task.

[0045] Among them, the task execution sequence number of the sub-work task is used to indicate the execution order of the sub-work task, for example, sorting each sub-work task according to priority (including but not limited to ascending order and descending order), and the task execution position of the sub-work task is the coordinate information of the sub-work task in the working environment.

[0046] The purpose of this step is to sort the sub-tasks according to the execution order so as to determine the position conversion scheme between the sub-tasks.

[0047] For example, assuming that the sub-work tasks are tightening screw 1, tightening screw 2, tightening screw 3 and tightening screw 4, and the priority of each sub-work task is tightening screw 1, tightening screw 2, tightening screw 4 and tightening screw 3, then the task execution sequence number of tightening screw 1 is sequence number 1-1, the task execution sequence number of tightening screw 2 is sequence number 1-2, the task execution sequence number of tightening screw 3 is sequence number 1-4, and the task execution sequence number of tightening screw 4 is sequence number 1-3.

[0048] S203: Determine at least one sub-travel task, the task execution sequence number and the task execution path of at least one sub-travel task based on the task execution sequence number and the task execution position of at least one sub-work task.

[0049] Among them, the sub-travel task is used to indicate the driving mode of the agent between two connected sub-tasks. Specifically, the sub-travel task is used to instruct the target agent to move from the task end of the sub-task to the task starting point of the next sub-task after the sub-task is completed, which is equivalent to moving the agent from the end of the current sub-task to the starting point required by the next sub-task starting point, so that the target agent can perform the next sub-task. The number of sub-travel tasks is the number of sub-tasks minus 1.

[0050] Specifically, after completing a task, the intelligent agent needs to move to the location of the next task before it can start the corresponding task. Assuming that the sub-tasks are screwing screw 1, screwing screw 2, screwing screw 3 and screwing screw 4, the priorities of each sub-task are screwing screw 1, screwing screw 2, screwing screw 4 and screwing screw 3, then the sub-driving tasks include driving task 1 (driving task between sub-task 1 and sub-task 2), driving task 2 (driving task between sub-task 2 and sub-task 4) and driving task 3 (driving task between sub-task 4 and sub-task 3). The task execution path of driving task 1 is the driving path from screw 1 to screw 2, the task execution path of driving task 2 is the driving path from screw 2 to screw 4, and the task execution path of driving task 3 is the driving path from screw 4 to screw 3. The task execution sequence number of driving task 1 is 2-1, the task execution sequence number of driving task 2 is 2-2, and the task execution sequence number of driving task 3 is 2-3.

[0051] The advantage of this setting is that it can determine the movement requirements and the order of each movement requirement when the agent is performing a task, so as to formulate a complete agent operation plan and control the agent to better perform the task.

[0052] S204: Determine a target agent according to at least one sub-task to be performed.

[0053] The target agent is the agent that is most suitable for completing the task to be performed in the working environment, including but not limited to the agent that is closest to the task to be performed and the agent that consumes the least resources to perform the task to be performed.

[0054] Specifically, when the number of sub-work tasks is 1, the candidate intelligent agent closest to the sub-work task is determined as the target intelligent agent, and the candidate intelligent agent is a pre-set intelligent agent for completing the task to be executed; when the number of sub-work tasks is greater than 1, it is determined whether the task to be executed meets the single-agent execution condition; if the task to be executed meets the single-agent execution condition, the target intelligent agent is determined based on the position information of the candidate intelligent agent and the task execution position of the sub-work task with the highest priority in the task execution sequence number; if the task to be executed does not meet the single-agent execution condition, the target intelligent agent is determined based on the position information of the candidate intelligent agent, the task execution position of at least one sub-work task and the task execution sequence number.

[0055] Among them, the single-agent execution condition can be understood as the basis for using one agent to complete the task to be executed, for example, there is only one candidate agent in the working environment, there is only one idle candidate agent in the working environment, the task to be executed includes only one sub-task, etc. The sub-task with the highest priority of the task execution sequence number can be understood as the task that needs to be executed first.

[0056] In one embodiment, the target intelligent agent is determined based on the location information of the candidate intelligent agent, the task execution location and task execution sequence number of at least one sub-work task, including: determining the first candidate sub-work task based on the task execution sequence number of at least one sub-work task, the first candidate sub-work task being the sub-work task with the highest priority in task execution sequence number; determining the candidate intelligent agent with the shortest distance to the task execution location of the first candidate sub-work task as the first target intelligent agent; utilizing the shortest execution time principle to determine the second candidate sub-work task from the remaining sub-work tasks, the remaining sub-work tasks being the sub-work tasks other than the first candidate sub-work task; determining the remaining candidate intelligent agents with the shortest distance to the task execution location of the second candidate sub-work task as the second target intelligent agent, the remaining candidate intelligent agents being the idle intelligent agents among the candidate intelligent agents other than the first target intelligent agent.

[0057] The first target agent will only be one agent. Generally, the second target agent is also an agent. However, in special cases (for example, the working environment is large, there are many candidate agents and many sub-agents to be executed), the second target agent may include multiple agents.

[0058] Taking the second agent as an agent as an example, in order to improve the execution efficiency of the task to be executed, the sub-task to be executed can be divided into two tasks for execution. For example, roughly calculate the execution time required for each sub-task to be executed, and divide it into two parts more evenly. The part with a higher priority is executed by the first target agent, and the part with a lower priority is executed by the second target agent. It is worth noting that the deadline work of the first task is a sub-task, and the starting work of the second task is also a sub-task. That is, when dividing the tasks, the sub-travel task between the two sub-tasks is adaptively deleted. The purpose of this setting is to avoid the first target agent and the second target agent from traveling unnecessary routes and reduce resource consumption.

[0059] Furthermore, the present invention can also adopt other principles such as the principle of closest distance to divide the sub-tasks to be executed into multiple parts for parallel execution. When dividing, it is only necessary to ensure that the sequence of the sub-tasks to be executed does not conflict, and there is no need for a certain intelligent agent to execute related sub-tasks to be executed. The advantage of such a setting is that the intelligent agent can flexibly adapt to the sub-tasks to be executed, improve the efficiency of the task, and save resources.

[0060] S205: Determine the execution order of at least one sub-task to be executed.

[0061] Specifically, determining the execution order of at least one sub-task to be executed can be understood as integrating each sub-work task and sub-driving task to obtain a completed task execution process.

[0062] For example, assuming that the sub-work tasks are locking screw 1, locking screw 2, locking screw 4 and locking screw 3, and the sub-driving tasks are driving task 1 (driving task between screw 1 and screw 2), driving task 2 (driving task between screw 2 and screw 4) and driving task 3 (driving task between screw 4 and screw 3), then the sub-tasks to be executed include 7 (4 sub-work tasks and 3 sub-driving tasks), and the execution order of the complete sub-tasks to be executed is locking screw 1, driving task 1, locking screw 2, driving task 2, locking screw 4, driving task 3 and locking screw 3. Specifically, the execution order of at least one sub-task to be executed can be 1-1, 2-1, 1-2, 2-2, 1-3, 2-3 and 1-4.

[0063] S206: Using the primitive task model library, determine the sub-primitive model of each sub-task to be executed in turn.

[0064] Using the primitive task model library to determine the sub-primitive models of each sub-task to be executed in turn can be understood as determining the algorithm (i.e., sub-primitive model) corresponding to each sub-task to be executed based on the primitive models in the primitive task model library and the operation logic of each primitive model. The sub-primitive model corresponds to the sub-task to be executed one by one, and the determination process is also matched / analyzed one by one.

[0065] For any sub-task to be executed, the sub-element model of the sub-task to be executed is determined by using the primitive task model library, including: based on the characteristic information of the sub-task to be executed, determining whether there is a matching model of the sub-task to be executed in the primitive task model library, that is, whether the model corresponding to the sub-task to be executed can be directly obtained from the primitive task model library; if there is a matching model of the sub-task to be executed in the primitive task model library, then determining the sub-element model of the sub-task to be executed based on the matching result, that is, directly determining the matched model as the sub-element model of the sub-task to be executed; if there is no matching model of the sub-task to be executed in the primitive task model library, obtaining the associated model of the sub-task to be executed from the primitive task model library, and determining the sub-element model of the sub-task to be executed based on the running logic of the associated model.

[0066] The associated model can be understood as a model that cannot directly complete the sub-task to be executed, but has the ability to complete the sub-task to be executed. For example, the sub-task to be executed is to go forward 5 meters and then turn left and drive 3 meters. The primitive task model library only has models for going forward and turning left. The ability of going forward + turning left + going forward can be analyzed according to the forward model and the left turn model, and the model with this ability can be trained as the sub-primitive model of the sub-task to be executed. The purpose of this setting is to quickly determine the sub-primitive model when there is a model that completes the sub-task to be executed in the primitive task model library. When there is no model that completes the sub-task to be executed in the primitive task model library, the sub-primitive model of the sub-task to be executed is trained according to the running logic of the associated model, so as to improve the task execution rate while ensuring the smooth execution of the task.

[0067] S207 . Based on the execution order, combine the sub-element models of the sub-tasks to be executed to obtain a target element model.

[0068] Specifically, combining the sub-element models of each sub-task to be executed based on the execution order can be understood as splicing the sub-element models corresponding to each sub-task to be executed in sequence according to the execution order. The splicing result is the target element model, and the target element model is the sequential arrangement result of each sub-element model.

[0069] For example, assuming that the order of sub-tasks to be executed is Task 1, Task 3 and Task 2, the sub-element model corresponding to Task 1 is Model 4, the sub-element model corresponding to Task 2 is Model 3, and the sub-element model corresponding to Task 3 is Model 6, then the target element model is the sequential combination of Model 4, Model 6 and Model 3.

[0070] S208. Control the target intelligent agent to run based on the target primitive model to complete the task to be executed.

[0071] Controlling the target intelligent agent to run based on the target primitive model and completing the tasks to be executed can be understood as controlling the target intelligent agent to run in sequence based on each sub-primitive model in the target primitive model, running the next model after one model runs successfully, and completing each sub-task to be executed in sequence.

[0072] Optionally, before controlling the target intelligent body to run based on the target primitive model to complete the task to be executed, it also includes: determining the starting driving task of the target intelligent body based on the position information of the target intelligent body and the task execution position of the first sub-task to be executed; determining the starting driving primitive model of the task to be executed based on the starting driving task and the primitive task model library; controlling the target intelligent body to run based on the starting driving primitive model to move the target intelligent body to the task execution position of the first sub-task to be executed, so that the target intelligent body executes the first sub-task to be executed. If the intelligent body encounters an obstacle during driving, a warning message will be issued to prompt the staff to adjust the target intelligent body. The advantage of this setting is that it can ensure that the intelligent body can smoothly execute the first sub-task to be executed and reduce the problem of the task to be executed not being able to proceed smoothly due to driving failures.

[0073] It is worth noting that if the target intelligent agent also needs to return to a fixed position, the stopping and driving task can be determined based on the location information of the last sub-task and the return position of the target intelligent agent after the intelligent agent completes the task to be executed; based on the stopping and driving task and the primitive task model library, the stopping and driving primitive model of the task to be executed can be determined; the stopping and starting driving primitive model of the target intelligent agent is controlled to run, so as to move the target intelligent agent to the parking place, which is convenient for the intelligent agent to perform charging and other tasks, and also convenient for the staff to manage the intelligent agent.

[0074] The technical solution of the present invention decomposes the task to be executed into multiple task segments, that is, decomposes a complex task into multiple simple tasks, matches the task segments in the primitive task model library, obtains multiple sub-primitive models, and then splices the sub-primitive models to obtain the target primitive model. The model library only needs to train and update the commonly used primitive models, which can not only reduce the training cost of the model, but also improve the applicability of the primitive model, so that it can cope with diverse and complex usage environments, and improve the flexibility, practicality and user experience of the intelligent body. Secondly, the present invention splits long-path tasks into short-path tasks, which reduces the requirements of the intelligent body for imitation learning algorithms. Furthermore, the present invention can realize the simplified processing of complex tasks, reduce the cost of model training, improve the reusability of primitive models, realize complex tasks by combination, improve the generation efficiency of complex task strategies, and reduce the difficulty of generating complex task strategies, and has high versatility.

[0075] Figure 3 Schematic diagram of the structure of a complex task execution device of an industrial intelligent body provided by the present invention. Figure 3As shown, the device includes: a task decomposition module 301, a model determination module 302 and a task execution module 303.

[0076] The task decomposition module 301 is used to decompose the task to be executed by using a preset task decomposition rule to obtain at least one sub-task to be executed, and determine the target intelligent agent according to the at least one sub-task to be executed.

[0077] The model determination module 302 is used to determine the target primitive model of the task to be executed based on at least one sub-task to be executed and a pre-trained primitive task model library, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model corresponds to the sub-task to be executed one by one.

[0078] The task execution module 303 is used to control the target agent to run based on the target primitive model to complete the task to be executed.

[0079] Optionally, the sub-tasks to be executed include sub-work tasks and sub-driving tasks. The task decomposition module 301 is specifically used to determine at least one sub-work task, the priority and location information of at least one sub-work task according to the task requirements of the task to be executed; determine the task execution sequence number and task execution location of at least one sub-work task based on the priority and location information of at least one sub-work task; determine the task execution sequence number and task execution path of at least one sub-driving task, at least one sub-driving task based on the task execution sequence number and task execution location of at least one sub-work task, wherein the sub-driving task is used to instruct the target intelligent body to move from the task end point of the sub-work task to the task starting point of the next sub-work task after the sub-work task is completed, and the number of sub-driving tasks is the number of sub-work tasks minus 1.

[0080] Optionally, the task decomposition module 301 is specifically used to determine that when the number of sub-work tasks is 1, the candidate intelligent agent closest to the sub-work task is the target intelligent agent, and the candidate intelligent agent is a pre-set intelligent agent for completing the task to be executed; when the number of sub-work tasks is greater than 1, determine whether the task to be executed meets the single-agent execution condition; if the task to be executed meets the single-agent execution condition, the target intelligent agent is determined based on the position information of the candidate intelligent agent and the task execution position of the sub-work task with the highest priority in the task execution sequence number; if the task to be executed does not meet the single-agent execution condition, the target intelligent agent is determined based on the position information of the candidate intelligent agent, the task execution position of at least one sub-work task and the task execution sequence number.

[0081] Optionally, the task decomposition module 301 is specifically used to determine a first candidate sub-work task based on the task execution sequence number of at least one sub-work task, the first candidate sub-work task being the sub-work task with the highest priority in task execution sequence number; determine the candidate agent with the shortest distance from the task execution position of the first candidate sub-work task as the first target agent; determine the second candidate sub-work task from the remaining sub-work tasks using the shortest execution time principle, the remaining sub-work tasks being the sub-work tasks other than the first candidate sub-work task; determine the remaining candidate agents with the shortest distance from the task execution position of the second candidate sub-work task as the second target agent, the remaining candidate agents being the idle agents among the candidate agents other than the first target agent.

[0082] Optionally, the model determination module 302 is specifically used to determine the execution order of at least one sub-task to be executed; using the primitive task model library, determine the sub-element model of each sub-task to be executed in turn; based on the execution order, combine the sub-element models of each sub-task to be executed to obtain the target primitive model.

[0083] Optionally, for any sub-task to be executed, the model determination module 302 is specifically used to determine whether there is a matching model of the sub-task to be executed in the primitive task model library based on the characteristic information of the sub-task to be executed; if there is a matching model of the sub-task to be executed in the primitive task model library, the sub-element model of the sub-task to be executed is determined based on the matching result; if there is no matching model of the sub-task to be executed in the primitive task model library, the associated model of the sub-task to be executed is obtained from the primitive task model library, and the sub-element model of the sub-task to be executed is determined based on the operating logic of the associated model.

[0084] Optionally, the task execution module 303 is also used to determine the starting driving task of the target intelligent body based on the position information of the target intelligent body and the task execution position of the first sub-task to be executed before controlling the target intelligent body to run based on the target primitive model to complete the task to be executed; determine the starting driving primitive model of the task to be executed based on the starting driving task and the primitive task model library; control the target intelligent body to run based on the starting driving primitive model to move the target intelligent body to the task execution position of the first sub-task to be executed, so that the target intelligent body executes the first sub-task to be executed.

[0085] The complex task execution device of the industrial intelligent body provided by the present invention can execute the complex task execution method of the industrial intelligent body provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0086] Figure 4: is a schematic diagram of the structure of an electronic device provided by the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0087] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (also called random access memory, RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, ROM 12 and RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0088] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0089] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a complex task execution method of an industrial intelligent body.

[0090] In some embodiments, the complex task execution method of the industrial intelligent agent may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the complex task execution method of the industrial intelligent agent described above may be executed. Alternatively, in other embodiments, the processor 11 may be configured to execute the complex task execution method of the industrial intelligent agent in any other appropriate manner (e.g., by means of firmware).

[0091] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0092] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0093] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0094] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0095] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server) or a computing system that includes middleware components (e.g., an application server) or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein) or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0096] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0097] In one embodiment, the present invention further includes a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the complex task execution method of the industrial intelligent body of any embodiment of the present invention.

[0098] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0099] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0100] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A complex task execution method of an industrial agent, characterized in that: include: Decomposing the task to be executed by using a preset task decomposition rule to obtain at least one sub-task to be executed, and determining a target intelligent agent according to the at least one sub-task to be executed; Based on the at least one sub-task to be performed and a pre-trained primitive task model library, determining a target primitive model of the task to be performed, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model corresponds to the sub-task to be performed one by one; The target agent is controlled to run based on the target primitive model to complete the task to be performed.

2. The method according to claim 1, characterized in that The sub-tasks to be executed include sub-work tasks and sub-travel tasks. The pre-set task decomposition rules are used to decompose the tasks to be executed to obtain at least one sub-task to be executed, including: Determine at least one sub-task, the priority and location information of the at least one sub-task according to the task requirements of the task to be performed; Determine the task execution sequence number and task execution position of the at least one sub-task based on the priority and position information of the at least one sub-task; Based on the task execution sequence number and task execution position of the at least one sub-work task, determine at least one sub-driving task, the task execution sequence number and task execution path of the at least one sub-driving task, wherein the sub-driving task is used to instruct the target intelligent agent to move from the task end point of the sub-work task to the task starting point of the next sub-work task after the sub-work task is completed, and the number of the sub-driving tasks is the number of the sub-work tasks minus 1.

3. The method according to claim 2, characterized in that The step of determining a target agent according to the at least one sub-task to be performed comprises: When the number of the sub-tasks is 1, determining the candidate agent closest to the sub-task as the target agent, wherein the candidate agent is a pre-set agent for completing the task to be performed; When the number of the sub-tasks is greater than 1, determining whether the task to be executed meets the single-agent execution condition; If the task to be executed meets the single agent execution condition, the target agent is determined based on the position information of the candidate agent and the task execution position of the sub-task with the highest task execution sequence priority; If the task to be executed does not meet the single-agent execution condition, the target agent is determined based on the position information of the candidate agent, the task execution position of the at least one sub-task and the task execution sequence number.

4. The method according to claim 3, characterized in that The step of determining the target agent based on the position information of the candidate agent, the task execution position and the task execution sequence number of the at least one sub-task comprises: Determine a first candidate sub-work task based on the task execution sequence number of at least one sub-work task, wherein the first candidate sub-work task is the sub-work task with the highest priority of the task execution sequence number; Determine the candidate agent that has the shortest distance from the task execution location of the first candidate sub-task as the first target agent; Determine a second candidate sub-work task from the remaining sub-work tasks by using the shortest execution time principle, wherein the remaining sub-work tasks are sub-work tasks other than the first candidate sub-work task; Determine the remaining candidate agent with the shortest distance to the task execution location of the second candidate sub-work task as the second target agent, wherein the remaining candidate agents are idle agents among the candidate agents except the first target agent.

5. The method according to claim 1, characterized in that The step of determining a target primitive model of the task to be performed based on the at least one sub-task to be performed and a pre-trained primitive task model library includes: Determining an execution order of the at least one sub-task to be executed; Using the primitive task model library, sequentially determine the sub-primitive model of each of the sub-tasks to be executed; Based on the execution order, the sub-element models of the sub-tasks to be executed are combined to obtain the target element model.

6. The method according to claim 5, characterized in that For any sub-task to be executed, the sub-primitive model of the sub-task to be executed is determined by using the primitive task model library, including: Based on the feature information of the sub-task to be executed, determining whether there is a matching model for the sub-task to be executed in the primitive task model library; If a matching model of the sub-task to be executed exists in the primitive task model library, determining the sub-primitive model of the sub-task to be executed based on the matching result; If there is no matching model of the sub-task to be executed in the primitive task model library, the associated model of the sub-task to be executed is obtained from the primitive task model library, and the sub-primitive model of the sub-task to be executed is determined based on the running logic of the associated model.

7. The method according to claim 1, characterized in that Before controlling the target agent to run based on the target primitive model to complete the task to be performed, the method further includes: Determine the starting driving task of the target intelligent body based on the position information of the target intelligent body and the task execution position of the first sub-task to be executed; Determining a starting driving primitive model of the task to be performed based on the starting driving task and the primitive task model library; The target agent is controlled to run based on the starting driving primitive model to move the target agent to a task execution position of the first sub-task to be executed, so that the target agent executes the first sub-task to be executed.

8. A complex task execution device of an industrial intelligent agent, characterized in that: A complex task execution method for an industrial intelligent body according to any one of claims 1 to 7, wherein the complex task execution device of the industrial intelligent body comprises: A task decomposition module is used to decompose the task to be executed by using a preset task decomposition rule to obtain at least one sub-task to be executed, and determine a target intelligent agent according to the at least one sub-task to be executed; A model determination module, used to determine a target primitive model of the task to be executed based on the at least one sub-task to be executed and a pre-trained primitive task model library, wherein the target primitive model includes at least one sub-primitive model, and the sub-primitive model corresponds to the sub-task to be executed in one-to-one correspondence; The task execution module is used to control the target agent to run based on the target primitive model to complete the task to be executed.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the complex task execution method of the industrial intelligent body described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the complex task execution method of an industrial intelligent agent as described in any one of claims 1 to 7 when executed.

Images