Method for supporting complex task performance of autonomous behavior robot by using edge and cloud

Abstract

Provided is an edge-based intelligence support method for performing a complex task of an autonomous behavior robot. Provided is a method for performing a task of an autonomous behavior robot, according to an embodiment of the present invention, in which the autonomous behavior robot performs the task by using an ultra-lightweighted artificial intelligence model, an edge server supports the task performance of the autonomous behavior robot by using a lightweighted artificial intelligence model, and a cloud server supports the task performance of the robot by using a huge artificial intelligence model. Accordingly, in addition to distributing and updating, through edge computing, an artificial intelligence model required when the autonomous behavior robot performs various tasks, the task performance itself is supported through an edge or cloud, thereby increasing the task performance efficiency of the autonomous behavior robot to improve service quality.

Description

Method to support complex task execution by autonomous robots using edge and cloud

[0001] The present invention relates to artificial intelligence-based robot control, and more specifically, to an intelligent support method for performing complex tasks of an autonomous robot operating based on artificial intelligence.

[0002] Existing intelligent robot systems are structured such that dedicated intelligence models are mounted on the robot in the form of on-device AI, performing defined tasks in limited locations or environments through this embedded intelligence. While this allows for performance optimized for specific tasks, there are limitations in performing other tasks.

[0003] Furthermore, since the available computing resources for on-device AI-based intelligent robots are limited, the majority are equipped with lightweight AI models optimized for specific tasks. As a result, executing high-performance models such as LLM is difficult, making it challenging to perform new or complex tasks that deviate from pre-learned patterns.

[0004] The present invention has been devised to solve the aforementioned problems, and the objective of the present invention is to provide a method that goes beyond deploying and updating artificial intelligence models required for autonomous robots to perform various tasks via edge computing, and further supports the task execution itself through the edge or cloud.

[0005] A method for performing a task by an autonomous robot according to an embodiment of the present invention for achieving the above objective comprises: a step in which the autonomous robot performs a task using a first artificial intelligence model; a first support step in which an edge server supports the task performance of the autonomous robot using a second artificial intelligence model; and a second support step in which a cloud server supports the task performance of the robot using a third artificial intelligence model.

[0006] The task includes at least one of route driving, object recognition, map updating, given task, natural language understanding and generation, and content understanding and generation, and the task may include at least one of delivery, patrolling, cleaning, and sensing.

[0007] Support may involve performing tasks that the autonomous robot cannot directly perform, or tasks where the accuracy is expected to fall below a standard even if performed directly, and providing the results.

[0008] The task supported in the first support stage is a task with higher complexity, higher difficulty, or requiring a larger-scale artificial intelligence model than the task performed in the execution stage, and the task supported in the second support stage may be a task with higher complexity, higher difficulty, or requiring a larger-scale artificial intelligence model than the task supported in the first support stage.

[0009] The task supported in the first support stage is a task in which the similarity between the input data to be processed and the previous data is less than the first similarity, and the task supported in the second support stage may be a task in which the similarity between the input data to be processed and the previous data is lower than the first similarity, with a second similarity.

[0010] The task supported in the first support stage is a task whose execution result in the execution stage was determined to be a failure, and the task supported in the second support stage may be a task whose execution result in the first support stage was also determined to be a failure.

[0011] The work supported in the first support stage is a work that must be processed in the first time period, and the work supported in the second support stage may be a work that must be processed in the second time period.

[0012] The work supported in the first support stage is a work that must be processed in the first state of the weather conditions, and the work supported in the second support stage may be a work that must be processed in the second state of the weather conditions.

[0013] The work supported in the first support stage is a work that must be processed in a situation where the spatial density is greater than or equal to the first density, and the work supported in the second support stage may be a work that must be processed in a situation where the spatial density is greater than or equal to the second density, which is higher than the first density.

[0014] According to another aspect of the present invention, a system for performing tasks of an autonomous robot is provided, characterized by comprising: an autonomous robot that performs tasks using a first artificial intelligence model; an edge server that supports the performance of tasks of the autonomous robot using a second artificial intelligence model; and a cloud server that supports the performance of tasks of the robot using a third artificial intelligence model.

[0015] According to another aspect of the present invention, a method for supporting the execution of a task by an autonomous robot is provided, characterized by comprising: a first support step in which an edge server supports the execution of a task by an autonomous robot that performs a task using a first artificial intelligence model by using a second artificial intelligence model; and a second support step in which a cloud server supports the execution of a task by the robot by using a third artificial intelligence model.

[0016] According to another aspect of the present invention, a system for supporting the execution of tasks by an autonomous robot is provided, characterized by comprising: an edge server that supports the execution of tasks by an autonomous robot that performs tasks using a first artificial intelligence model using a second artificial intelligence model; and a cloud server that supports the execution of tasks by the robot using a third artificial intelligence model.

[0017] As explained above, according to the embodiments of the present invention, in addition to deploying and updating artificial intelligence models required when an autonomous robot performs various tasks through edge computing, by supporting the task execution itself through the edge or cloud, it is possible to enhance the mission performance of the autonomous robot and improve service quality.

[0018] FIG. 1 is a complex task performance / support system for an autonomous action robot according to one embodiment of the present invention,

[0019] FIG. 2 is a method for performing task substitution of an autonomous action robot based on task characteristics, according to another embodiment of the present invention.

[0020] FIG. 3 is a method for performing task substitution of a data change-based autonomous action robot according to another embodiment of the present invention.

[0021] FIG. 4 is a method for performing task substitution of an autonomous action robot based on task success or failure according to another embodiment of the present invention.

[0022] FIG. 5 is a method for performing task substitution of a time-based autonomous action robot according to another embodiment of the present invention.

[0023] FIG. 6 is a method for performing task substitution of a weather condition-based autonomous action robot according to another embodiment of the present invention.

[0024] FIG. 7 is a method for performing task substitution of an illuminance-based autonomous action robot according to another embodiment of the present invention.

[0025] FIG. 8 is a method for performing task substitution of a spatial density-based autonomous action robot according to another embodiment of the present invention.

[0026] The present invention will be described in more detail below with reference to the drawings.

[0027] An embodiment of the present invention presents an edge / cloud-based intelligence support method for performing complex tasks by an autonomous robot. This technology involves deploying and updating artificial intelligence models required when an autonomous robot performs various tasks via edge computing, or supporting the task execution itself through the edge or cloud.

[0028] FIG. 1 is a diagram illustrating the configuration of a complex task execution / support system for an autonomous robot according to an embodiment of the present invention. In an embodiment of the present invention, the autonomous robot receives support from the edge and the cloud when performing complex tasks.

[0029] A complex task performance / support system for an autonomous robot according to an embodiment of the present invention is configured to include, as illustrated, a cloud server (100), an edge server (200), and an autonomous robot (Device, 300).

[0030] The autonomous robot (300) is a robot system that performs complex tasks through autonomous judgment and action to provide specific services in a designated area. The complex tasks for providing services include path navigation, object recognition, map updating, and given detailed tasks.

[0031] Route navigation, object recognition, and map updates are the fundamental tasks involved in performing specific missions. These specific missions are determined by the type of service and may include item delivery, area patrolling, area cleaning, and area sensing; of course, depending on the service, other tasks may also be included.

[0032] To perform complex tasks through autonomous judgment and action, the autonomous robot (300) utilizes an artificial intelligence model. That is, it performs path navigation, object recognition, map updates, and detailed tasks based on artificial intelligence.

[0033] Meanwhile, since the resources of the autonomous robot (300) are limited in that it is a mobile platform requiring movement, the artificial intelligence model must be an ultra-lightweight model, and the range of tasks that can be performed is also limited.

[0034] Accordingly, to support the performance of complex tasks by the autonomous robot (300), a cloud server (100) and an edge server (200) are operated. In FIG. 1, only one cloud server (100) and one edge server (200) are shown, but this is for convenience of illustration; in reality, multiple servers are operated. In particular, in the case of the edge server (200), multiple edge servers exist within the service area of ​​the autonomous robot (300), so the autonomous robot (300) can receive support for task performance from the nearest edge server (200).

[0035] Support for performing complex tasks includes 1) the deployment / update of the ultra-lightweight model of the autonomous robot (300), and 2) the replacement of tasks to be performed by the autonomous robot (300). 1) is mainly performed by the edge server (200), and 2) is performed by both the massive model of the cloud server (100) and the lightweight model of the edge server (200).

[0036] The distribution of the ultra-lightweight model is necessary when a new task needs to be performed at the beginning of the complex task execution of the autonomous robot (300) or during the complex task execution. An update of the ultra-lightweight model may be necessary when the complexity or difficulty of the task increases during the complex task execution of the autonomous robot (300), when the input data to be processed changes from the previous data, or when the accuracy of the task execution decreases.

[0037] Substitute execution of a task refers to an edge server (200) or a cloud server (100) performing a task that the autonomous robot (300) is supposed to perform on its behalf and returning the result. Substitute execution of a task occurs in various situations, and these are described in detail below with various embodiments.

[0038] FIG. 2 is a diagram illustrating the flow of a method for performing task substitution of an autonomous action robot based on task characteristics according to another embodiment of the present invention.

[0039] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S411). The tasks that can be performed in step S411 are tasks that can be performed with the ultra-lightweight model and resources, such as the aforementioned driving, object recognition, map update, detailed mission (delivery, patrol, cleaning, sensing, etc.).

[0040] Meanwhile, for tasks that the autonomous robot (300) cannot perform directly, or tasks where the accuracy of the performance is expected to be below the standard even if performed directly, the edge server (200) supports the performance of the tasks. For example, this includes the recognition of new objects and high-difficulty map updates, but it does not include simple map updates such as the recognition of objects of learned classes or the addition of obstacles.

[0041] Accordingly, for tasks that are slightly more complex or difficult than tasks that can be performed by the ultra-lightweight model of the autonomous robot (300) or tasks that require a slightly larger scale artificial intelligence model (S412-Y), the lightweight model of the edge server (200) performs the tasks on behalf of the autonomous robot (300) and returns the results of the execution (S413).

[0042] On the other hand, for tasks that are too complex to be performed even by the lightweight model of the edge server (200), that is, tasks requiring a very large-scale artificial intelligence model with very high complexity or difficulty (S414-Y), the large model of the cloud server (100) performs the tasks on behalf of the autonomous robot (300) and returns the results (S415). This may include language understanding and generation, content understanding and generation, and tasks requiring very high computing power.

[0043] FIG. 3 is a diagram illustrating the flow of a method for performing task substitution of an autonomous behavior robot based on data change according to another embodiment of the present invention.

[0044] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S421).

[0045] Meanwhile, for tasks where the similarity between the input data and the previous data that the autonomous robot (300) must process is somewhat low, that is, tasks where it is expected that accurate task execution will be difficult because the input data is somewhat different from the previous one (S422-Y), a lightweight model of the edge server (200) performs the task on behalf of the autonomous robot (300) and returns the result of execution (S423).

[0046] On the other hand, for tasks where the similarity between input data and previous data is very low to the extent that even the lightweight model of the edge server (200) cannot cover them (S424-Y), the large model of the cloud server (100) performs the task on behalf of the autonomous robot (300) and returns the result of the execution (S425).

[0047] FIG. 4 is a diagram illustrating the flow of a method for performing task substitution for an autonomous action robot based on task success or failure according to another embodiment of the present invention.

[0048] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S431).

[0049] Meanwhile, for tasks performed by the ultra-lightweight model of the autonomous robot (300) that are found to be failures (S432-Y), the lightweight model of the edge server (200) performs the tasks of the autonomous robot (300) on its behalf and returns the results of the performance (S433).

[0050] Task execution failure refers to cases where the accuracy output from the model along with the task execution result is below a standard, or where the feedback from the user or the surrounding environment regarding the task execution result is negative.

[0051] On the other hand, for tasks that are found to be a failure when performed by the lightweight model of the edge server (200) (S434-Y), the large model of the cloud server (100) performs the task on behalf of the autonomous robot (300) and returns the result of the execution (S435).

[0052] FIG. 5 is a diagram illustrating the flow of a method for performing task substitution for a time-based autonomous action robot according to another embodiment of the present invention.

[0053] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S441).

[0054] Meanwhile, if the current time zone is the first time zone (S442-Y), the lightweight model of the edge server (200) performs the work of the autonomous robot (300) on its behalf and returns the result of the execution (S443). The first time zone consists of times when many tasks with high complexity and difficulty occur, and is determined according to service attributes.

[0055] Furthermore, if the current time zone is the second time zone (S444-Y), the large model of the cloud server (100) performs the work of the autonomous robot (300) on its behalf and returns the result of the execution (S445). The second time zone consists of times when many tasks with very high complexity and difficulty occur, and this is also determined according to the service attributes.

[0056] FIG. 6 is a diagram illustrating the flow of a method for performing task substitution for a weather-condition-based autonomous action robot according to another embodiment of the present invention.

[0057] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S451).

[0058] Meanwhile, if the current weather conditions are not good (S452-Y), the lightweight model of the edge server (200) performs the work of the autonomous robot (300) on its behalf and returns the result of the work (S453). This is because it may be difficult to perform the work when the weather conditions are not good.

[0059] Furthermore, if the current weather conditions are at a level of deterioration (S454-Y), the large model of the cloud server (100) performs the work of the autonomous robot (300) on its behalf and returns the result of the work (S455). This is because performing the work can be more difficult when the weather is deteriorating.

[0060] FIG. 7 is a diagram illustrating the flow of a method for performing task substitution of an illuminance-based autonomous action robot according to another embodiment of the present invention.

[0061] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S461).

[0062] Meanwhile, if the current illumination is low (S462-Y), the lightweight model of the edge server (200) performs the task of the autonomous robot (300) on its behalf and returns the result of the task (S463). This is because it may be difficult to perform the task when the illumination is low.

[0063] Furthermore, when the current illumination level is very low (S464-Y), the large model of the cloud server (100) performs the work of the autonomous robot (300) on its behalf and returns the result of the work (S465). This is because performing the work is more difficult when the illumination level is very low.

[0064] FIG. 8 is a diagram illustrating the flow of a method for performing task substitution for a spatial density-based autonomous action robot according to another embodiment of the present invention.

[0065] As described above, first, the autonomous robot (300) performs tasks using the ultra-lightweight model mounted on it (S471).

[0066] Meanwhile, if the object density of the space is high (S472-Y), the lightweight model of the edge server (200) performs the task of the autonomous robot (300) on its behalf and returns the result of the task (S473). This is because it may be difficult to perform the task when there are many objects in the space, such as people or vehicles.

[0067] Furthermore, in cases where the object density is very high (S474-Y), a large model of the cloud server (100) performs the work of the autonomous robot (300) on its behalf and returns the result of the work (S475). This is because performing the work can be more difficult when there are a lot of objects in the space.

[0068] Up to now, an edge-based intelligence support method for performing complex tasks by an autonomous robot has been described in detail with reference to preferred embodiments.

[0069] In the above embodiment, beyond deploying and updating the artificial intelligence models required for the autonomous robot to perform various tasks via edge computing, the task execution itself is supported through the edge or cloud, thereby enhancing the mission performance of the autonomous robot and improving service quality.

[0070] Meanwhile, it goes without saying that the technical concept of the present invention may also be applied to a computer-readable recording medium containing a computer program that enables the device and method according to the present embodiment to perform their functions. Furthermore, the technical concept according to various embodiments of the present invention may be implemented in the form of computer-readable code recorded on a computer-readable recording medium. A computer-readable recording medium may be any data storage device that can be read by a computer and store data. For example, a computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, etc. Additionally, computer-readable code or a program stored on a computer-readable recording medium may be transmitted through a network connected between computers.

[0071] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. Various modifications are possible by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention.

Claims

1. A step in which an autonomous robot performs a task using a first artificial intelligence model; A first support step in which an edge server supports the task execution of an autonomous robot using a second artificial intelligence model; A method for performing a task of an autonomous robot, characterized by including a second support step in which a cloud server supports the robot's task performance using a third artificial intelligence model.

2. In Claim 1, The work is, It includes at least one of path navigation, object recognition, map updating, given tasks, natural language understanding and generation, and content understanding and generation, and The mission is, A method for performing a task of an autonomous robot characterized by including at least one of delivery, patrolling, cleaning, and sensing.

3. In Claim 1, Support is, A method for performing a task by an autonomous robot, characterized by performing a task that the autonomous robot cannot directly perform or a task where the accuracy of the performance is expected to be below a standard even if it is directly performed, and providing the result of the performance.

4. In Claim 3, The tasks supported in the first support stage are, It is a task with higher complexity, difficulty, or a larger-scale artificial intelligence model than the task performed during the execution phase, and The tasks supported in the second support phase are, A method for performing a task by an autonomous robot, characterized by the task being more complex, difficult, or requiring a larger-scale artificial intelligence model than the task supported in the first support stage.

5. In Claim 3, The tasks supported in the first support stage are, It is a task in which the similarity between the input data to be processed and the previous data is less than the first similarity, and The tasks supported in the second support phase are, A method for performing a task by an autonomous robot, characterized by the similarity between input data to be processed and previous data being less than a second similarity, which is lower than a first similarity.

6. In Claim 3, The tasks supported in the first support stage are, It is a task whose execution result during the execution phase was determined to be a failure, and The tasks supported in the second support phase are, A method for performing a task by an autonomous robot, characterized in that the result of performing the task in the first support stage is also found to be a failure.

7. In Claim 3, The tasks supported in the first support stage are, It is a task that must be processed during the first time period, and The tasks supported in the second support phase are, A method for performing a task of an autonomous robot characterized by the task being to be processed in a second time period.

8. In Claim 3, The tasks supported in the first support stage are, The weather condition is a task that must be processed in the first state, and The tasks supported in the second support phase are, A method for performing a task of an autonomous robot characterized by the fact that the weather condition is a task that must be processed in a second state.

9. In Claim 3, The tasks supported in the first support stage are, It is a task that must be processed in a situation where the spatial density is at least the first density, and The tasks supported in the second support phase are, A method for performing a task of an autonomous robot, characterized by the fact that the task must be processed in a situation where the spatial density is higher than the first density, or higher than the second density.

10. An autonomous robot performing tasks using the first artificial intelligence model; Edge server supporting task execution of autonomous robots with a second artificial intelligence model; A task execution system for an autonomous robot, characterized by including a cloud server that supports the robot's task execution with a third artificial intelligence model.

11. A first support step in which an edge server supports the execution of a task by an autonomous robot that performs a task with a first artificial intelligence model, using a second artificial intelligence model; A method for supporting task execution of an autonomous robot, characterized by including a second support step in which a cloud server supports the task execution of the robot using a third artificial intelligence model.

12. An edge server that supports the task execution of an autonomous robot performing a task with a first artificial intelligence model using a second artificial intelligence model; A task performance support system for an autonomous robot, characterized by including a cloud server that supports the robot's task performance using a third artificial intelligence model.

Images