JP2022500722A - Task management method and its system - Google Patents

Abstract

The present invention relates to a task management method and the system that enables task planning for task management by an autonomous system, and in particular, to increase the completion of assigned tasks and / or the possibility of task completion. , It is possible to manage tasks by recovering from a failure, but it is not limited to this. The system generates execution plans based on task goals defined to perform tasks, enabling continuous management of multiple tasks in the event of a failure.

Description

The present invention relates to a task management method and its system, and in particular, a task management method and its system that enable recovery in the event of a failure in order to complete a task or improve the possibility of completing a task. However, it is not limited to this.

The following discussion of the background of the invention is intended to facilitate understanding of the invention. However, it is understood that the discussion here does not admit or acknowledge that any of the materials mentioned are public, publicly known, or part of the general knowledge on the priority date of the application. Should be.

Unlike traditional equipment with a limited range of capabilities, autonomous systems perform more demanding and more complex tasks in an unstructured environment without continuous human intervention. Is possible. Autonomous systems can sense and adapt to the environment, allowing them to work for extended periods of time without human intervention. This leads to cost savings, efficiency and productivity gains.

Autonomous systems generally operate in disciplined situations. When performing a task, the autonomous robot may encounter various contingencies and may not be able to complete the task completely. This is mainly due to the inability to accurately characterize the dynamic environment in which the robot operates and the failure of sensors and actuators when the robot moves in that environment. Furthermore, in the event of a failure, the robot cannot recover quickly and efficiently, and interruptions will inevitably occur.

In a conventional system, analysis of the reason for failure is performed as a response in the event of a failure, but knowledge for identifying the reason for failure and making a recovery plan is required. However, little is known about how autonomous robots respond to the activities of other agents that may or may not cooperate with the execution of tasks, and respond to dynamic environments. Therefore, it may be difficult to extend the analysis of the reason for failure. In addition, such systems typically keep the state of the robot system constant to detect failures and diagnose the cause of the failure, or to collect failure context data that requires complex software analysis and inference. Need to monitor.

Therefore, the present invention attempts to overcome at least some of the above-mentioned drawbacks. In particular, it provides task management methods and systems that enable quick and efficient recovery in the event of a failure, in order to complete assigned tasks and / or increase the likelihood of task completion.

In the present specification, unless otherwise specified, terms such as "comprising" and "consisting of" are interpreted as non-exhaustive and "including". It is not limited to this. "

According to the first embodiment of the present invention, there is provided a task management method for executing a task having a task goal for autonomously managing the assigned task. The method is to receive task information for defining a set of actions having one or more actions to perform a task, to determine a time calculated based on the defined set of actions. Judging whether or not the task goal is achieved by determining whether or not the calculated execution time exceeds the actual execution time, and when the task goal is not achieved, the task goal. To obtain one or more recovery responses associated with, and to obtain a set of updated actions that reflect the one or more recovery responses to manage the task, said one or more recovery responses. It consists of generating an execution plan based on.

Preferably, the task information should include the characteristics of a series of events. The set of actions is associated with a set of events selected to perform the task.

Preferably, the method may execute the generated execution plan in order to continuously execute the series of events in the event of a failure.

One or more sets of instructions associated with a recovery response are acquired and executed to perform the task based on selection criteria determined according to a given situation.

Preferably, the method further develops the expertise based on the observation of the dynamic response in the event of a failure.

Preferably, the method may further rank the recovery response based on the task goal in order to derive the recovery response for managing the failure.

Preferably, the recovery response may consist of one or more of the execution of the alternative mode, the return to the previous non-failure state, and the re-execution of the task.

Preferably, the task goal may consist of confirming previously obtained information from the user or past events in the sequence.

Preferably, the method may further determine the index of localization likelihood by calculating a covariance function based on the following equation.

pf = particle filter (robot, sensor, q, L, np)

Here, q = the covariance of the diffusion noise applied to the particles, L = the covariance of the probability of the sensor model, and np = the number of particles.

It is preferable that the task information further includes the characteristics of a plurality of tasks, and it is possible to execute the plurality of tasks by executing the execution plan. The plurality of tasks may be nested or cascaded.

According to the second embodiment of the present invention, there is provided a task management system for executing a task having a task goal for managing an autonomously assigned task. This system consists of a control module, a memory module, and a task manager. Each module is interconnectable to each other and can operate to communicate with each other. Preferably, a hardware module is provided to perform the task.

i) The control module and the task manager are operational to perform:
Receive task information that defines a set of actions that have one or more actions to perform a task.
Determine the time calculated based on the set of actions defined above.
By determining whether or not the calculated execution time exceeds the actual execution time, it is determined whether or not the task goal has been achieved.
ii) The control module and the memory module are operational to perform the following:
If the task goal is not achieved, one or more return responses associated with the task goal are acquired.
iii) The control module and the task manager are operational to perform:
An execution plan is generated based on the one or more return responses in order to acquire a set of updated actions that reflect the one or more return responses in order to manage the task.

Preferably, the task information should include the characteristics of a series of events.

The task manager is configured to execute the generated execution plan in order to continuously execute the series of events in the event of a failure.

The control module is configured to obtain an appropriate recovery response by applying expertise based on the observation of the dynamic response to the failure.

The control module is configured to rank a plurality of recovery responses based on the task goal and derive the appropriate recovery response.

The control module is configured to review information obtained from the user or from previous events in order to determine if the task goal has been achieved.

The memory module is configured to send the acquired recovery response to the control module, which consists of one or more of executing an alternative set of actions, returning to a previous non-failure state, and re-executing a task. ..

The system further calculates a covariance function based on the following equation to determine an index of the likelihood of localization of the system.

pf = particle filter (robot, sensor, q, L, np)

Here, q = the covariance of the diffusion noise applied to the particles, L = the covariance of the probability of the sensor model, and np = the number of particles.

Here, the covariance function provides an indicator of whether the system is well localized and ensures that its position is correct when the covariance is below a threshold.

It is preferable that the task information further includes the characteristics of a plurality of tasks, and it is possible to execute the plurality of tasks by executing the execution plan. The plurality of tasks may be nested or cascaded.

According to a third embodiment of the present invention, there is provided a computer program product comprising instructions for causing the computer to perform the steps described in the first embodiment of the present invention when the program is executed by the computer.

According to a fourth embodiment of the present invention, a computer-readable medium in which a computer program product according to the third embodiment of the present invention is stored is provided.

The present invention has at least the following advantages.

1) The present invention is a robust autonomous system (eg, robot, physical host) for performing tasks in a dynamic environment that effectively reduces / avoids the state of complete failure when a failure occurs. I will provide a. This has the advantage of reducing the need for human intervention when performing the assigned task.
2) The present invention provides a failure recovery system that minimizes downtime and inaccessibility due to a complete failure condition. As a result, continuous operation for task execution corresponding to the failure mode becomes possible, and the possibility of completing the task when a failure occurs can be improved.
3) The present invention provides a response framework that allows a robot to interact autonomously with the environment in order to manage obstacles encountered during the execution of an assigned task. The failure state of a system has the advantage that it can be managed without diagnosing the cause of the failure or collecting failure context data, which generally requires complex software analysis and inference.

Other embodiments and advantages of the invention will be apparent to those skilled in the art by reference to the following exemplary drawings for various embodiments of the invention.

Embodiments of the present invention will be described with reference to the accompanying drawings, but are illustrated for illustration purposes.

It is a schematic diagram of the task management system which is one Embodiment of this invention.

It is a flow chart of the task management process of the task management system of FIG.

It is a flow diagram which manages a task by repeating the assigned task.

It is a flow diagram that manages tasks by returning to the previous state without obstacles.

It is a flow diagram which manages a task by undertaking an alternative task.

It is a figure which shows the operation process which manages a task by combining the various approaches of FIGS. 3 and 5.

It is a figure which shows the operation process which manages a task by combining the various approaches of FIGS. 3 and 4.

It is a figure which shows the operation process which manages a task by combining various approaches of FIGS. 3, 4, and 5.

Embodiments of the present invention will be described with reference to the accompanying drawings. The terms used herein are used for the purposes of describing only certain embodiments and are not intended to limit the scope of the invention. Moreover, unless otherwise specified, all technical and scientific terms used herein have the same meanings generally understood by those skilled in the art to which the present invention belongs. ..

The use of the singulars "a", "an", and "the" includes both the singular and the plural, unless the context clearly indicates otherwise.

The use of "or" and "/ (slash)" means "and / or" unless otherwise noted. In addition, "including" and "having", other forms of these terms, such as "includes", "included", "has" and "has" The use of terms such as "have" is not limited.

As used herein, the term "failure" means that a system or component is unable to perform the requested function as specified.

With reference to FIGS. 1 to 5, there is a task management system and method for an autonomous system or an autonomous robot that autonomously manages tasks in a dynamic environment in order to adjust an execution plan consisting of task execution information. It is described and can provide disaster recovery information remotely. This task management system and method can incorporate a recovery event or response into an execution plan that effectively mitigates or avoids a completely failed state when a failure occurs.

The task management methods and systems of the present invention have various uses and are applied, for example, to methods of performing autonomous operations in manufacturing and logistics applications. Logistics applications can employ the invention to facilitate the transportation of goods with respect to autonomous systems that efficiently eliminate and / or minimize system failures.

FIG. 1 is a schematic diagram of a task management system 100 according to an embodiment of the present invention. The system 100 includes a control module 102, a task manager 104, and a memory module 106. The control module consists of a task manager 104 and one or more processors (not shown) for controlling and managing the memory module 106.

The control module 102 starts a task by transmitting an instruction set to the task manager 104. The task manager 104 executes a task when it receives an instruction set. After that, the task manager 104 reports the result of the task by sending a report on the task performance to the control module 102. The control module 102 processes a report on task performance and obtains the requested response from the memory module 106.

Each of the control module 102, the task manager 104 and the memory module 106 may be implemented on an autonomous system. Autonomous systems include robots or autonomous vehicles. In this embodiment, the autonomous system is in the form of a robot.

The control module 102, the task manager 104, and the memory module 106 are interconnected via a communication means. The communication means may be a dedicated network or any network suitable for connecting the control module 102, the task manager 104, and the memory module 106 to each other. When the task input is received by the control module 102, an instruction is sent to the task manager 104 and the memory module 106 and processed there. After that, the processed instruction including the recovery response is sent to the task manager 104.

The control module 102 and the task manager 104 are operational to perform the following:
-Receive task information that defines a set of actions to perform a task.
A task may be defined as a sequence of events for executing the task. Depending on the situation, it may be a sequence defined for one or more tasks. If the correct sequence or the best sequence exists, all sequences may be compared to determine it. Depending on the sequence of events, a set of actions may be defined to execute the event. There may be one or more actions in a set of actions.
• Determine the execution time calculated based on a set of defined actions that rely on the task root and the sequence of defined events.
-Determine whether the task goal has been achieved by determining whether the calculated execution time exceeds the actual execution time.
If the actual execution time exceeds the calculated execution time, a failure condition exists and the task goal is not achieved. The task goal may consist of one or more subtask goals.

The control module 102 and the memory module 106 are operational to perform the following:
-Obtain one or more recovery responses when the task goal is not achieved. This one or more recovery responses are associated with the task goal.

The control module 102 and the task manager 104 are operational to perform the following:
• For task management, generate an execution plan based on one or more instruction sets to get a set of actions that reflect the recovery response for the execution of a defined sequence of events.
-In the event of a failure, adapt the recovery response to manage the task and execute the execution plan so that the defined sequence of events is continuously executed.

A task management method according to an embodiment of the present invention will be described with reference to FIG. The task management method consists of the following steps.
-Receive task information that defines a set of actions to perform a task.
A task may be defined as a sequence of events for executing the task. Depending on the situation, it may be a sequence defined for one or more tasks. If the correct sequence or the best sequence exists, all sequences may be compared to determine it. Depending on the sequence of events, a set of actions may be defined to execute the event. A set of actions may consist of one or more actions depending on the assigned task.
-Determine whether the task goal has been achieved by determining whether the calculated execution time exceeds the actual execution time.
If the actual execution time exceeds the calculated execution time, a failure condition exists and the task goal is not achieved. The task goal may consist of one or more subtask goals.
-Obtain one or more recovery responses when the task goal is not achieved. This one or more recovery responses are associated with the task goal.
-For task management, an execution plan is generated based on one or more recovery responses in order to acquire a set of actions updated to reflect one or more recovery responses.

Here, when the execution plan is executed by the task manager 104, the defined sequence of events is executed continuously by adapting the recovery response to manage the task in the event of a failure.

The execution plan is created by the control module 102 for the assigned task. Such tasks may be related to the transportation of goods, the sorting of goods or the shelving of goods. An execution plan consists of a set of one or more individual tasks (or subtasks). Individual tasks may be performed by different entities, such as multiple hardware modules. Individual tasks are grouped together to achieve a common goal.

The execution plan may include the following information.
・ Target position of robot ・ Set of individual tasks to be executed ・ Set of actions to execute execution plan

The control module 102 receives the task input and converts it into a series of actions that satisfy this task. This set of actions is passed to the task manager 104 for execution by a hardware module (not shown). Hardware modules include sensors, actuators, etc. for performing tasks. The control module 102 generates an execution plan for the assigned task. The control module 102 can be configured with various environmental constraints (eg, when physical barriers in a particular area impede the operation), task constraints (eg, when the nature of the task, such as task complexity, affects the operation), and so on. Generate an execution plan according to the constraints. Each individual task is passed to task manager 104 for further processing.

The task manager 104 coordinates the execution of hardware modules when performing tasks. The task manager 104 assigns each individual task for execution by each hardware module. The task manager 104 ensures the reliability of task execution by ensuring that each individual task is executed before moving on to a subsequent individual task.

3, 4, and 5 are high-level flow diagrams illustrating how recovery responses are triggered to manage assigned tasks in an autonomous system, which is an embodiment of the present invention. Is. The recovery response is initiated when the autonomous system encounters a failure condition that cannot handle tasks and / or subtasks. The recovery response includes taking on an alternative mode, repeating the first mode of operation, returning to a fault-free state before the failure state, or a combination thereof, depending on the situation and the task on hand.

FIG. 3 shows an operation process in which an autonomous system is restored by repeating an assigned task in order to achieve the initial goal of the task. This iterative approach can produce different outputs. The autonomous system receives a request to perform task A. The system starts task A in process step 301, and recognizes the success state when task A is completed in step 302. The system proceeds to the next task.

In the process step 303, the system recognizes the failure state when the task A is not executed normally. For example, a failure condition occurs when task A is not executed within a predetermined time, which is the expected execution time of the task. The system retries task A, but may try multiple times. By repeating task A, the possibility of task completion increases. If the system fails to perform the same task after n attempts in step 304, the task is considered a "failed" task. The system monitors the output of the task when task A is completed in step 305. Ideally, repetitive recovery may be suitable for tasks that may produce different output each time the task is executed.

FIG. 4 shows an operating process in which an autonomous system is restored by returning the system to its previous state without failure. By returning to the previous state, the system prepares to take on the next task / subtask. The autonomous system receives a request to perform task A. In the process step 401, the system starts the task A and recognizes the success state when the task A is completed. The system proceeds to the next task.

In the process step 402, the system recognizes the failure state when the task A is not executed normally. For example, a failure condition occurs when task A is not executed within a predetermined time, which is the expected execution time of the task. The system resets and returns to the previous fault-free state. This allows the system to undertake another assigned task (task B). The system monitors the task output of task A. This reset approach may be suitable for tasks where the same result may be obtained each time the task is executed. Ideally, resetting can mitigate the failure conditions that may be encountered when a particular task is performed.

FIG. 5 shows an operational process in which an autonomous system recovers by undertaking an alternative task to accomplish a task. In this case, the alternative task is neither available nor assigned. The autonomous system receives a request to perform task A. In process step 501, the system starts task A and recognizes the success state when task A is completed. The system proceeds to the next task.

In the process step 502, the system recognizes the failure state when the task A is not executed normally. For example, a failure condition occurs when task A is not executed within a predetermined time, which is the expected execution time of the task. The system performs one or more alternative tasks (in this case task B) in an attempt to complete the initial task in an attempt to achieve the purpose of the initial task. This alternative approach may be suitable for tasks that may be accomplished in different ways.

The task management system reduces the labor-intensive work of employees, and realizes efficiency and time saving. Severe process-focused situations, such as organizing books for shelving and creating catalogs, are categorized as simple contexts characterized by causality. In this case, there is a command and control style that sets the corresponding parameters when coordinating the management of tasks. For example, when a problem occurs, the problem can be easily identified, classified, and dealt with appropriately.

An application example of the task management system is an autonomous system for order processing and fulfillment for autonomous delivery of goods. For example, the first attempt to deliver an item may not be successful due to the absence of the recipient. Subsequent delivery attempts will probably guarantee the completion of the task, thereby minimizing service failures.

The task management system can take the form of an autonomous system or robot for navigating in a library or bookstore to find a bookshelf by scanning the RFID tag of a book. In some embodiments, the task may include positioning to a suitable height for the antenna for scanning books on a bookshelf. Repetitive tasks such as navigation through obstacles (eg shelves), positioning, antenna adjustment and / or readjustment are required to complete a series of tasks and are many individual. Includes intermediate tasks.

Hybrid Framework Operational processes such as those shown in FIGS. 3, 4, and 5 provide a basic structure for coordinating appropriate responses for task management.

The task management system and method of the present invention can manage nested tasks by superimposing one task on another. In this way, the tasks may be nested in any number of layers. Tasks may be managed in a cascading manner by continuously taking over the tasks. An execution plan generated by incorporating one or more models to provide a robust system for managing complex tasks.

Hybrid framework 1:
Goals are achieved by completing tasks or subtasks defined by complex tasks. Goals may consist of one or more subtask goals.

It is assumed that the task FA is a task of framework A (FIG. 3). When task FA is submitted to framework B (FIG. 4), the system will try n times before trying another method to achieve the goal. The task FA is managed by a repetitive recovery approach (Fig. 3). You may try to combine different approaches. In this case, the iterative and reset approaches may be combined to perform the task multiple times (or n times) before using another method to achieve the goal.

When the task FA is submitted to framework B (reset approach of FIG. 4), the repeat and reset approaches may be combined in order to execute the task multiple times. The system should try to recover from the previous non-failure state after failing multiple attempts. The task can then be retried using the same method with a higher probability of success.

It is assumed that the task FB is a task of framework B (FIG. 4). This new task FB can be put into framework A (repetitive approach). This allows the system to retry multiple times if both task A and task B fail.

The task FB can be submitted to framework B (reset approach) to cascade another alternative action, task C. The system first attempts to complete the task with task A. If unsuccessful, the system retries to complete the task with task B. If it still fails, proceed to task C until all possible methods have been tried.

The hybrid framework described above is an example of using a basic approach for managing tasks. There are endless combinations that can be used to plan tasks performed by a shelf scan robot consisting of multiple RFID antennas, a telescopic arm for projecting the antenna up and down, and a moving base that moves the telescopic arm structure.

Example 1: Move the antenna upward near the bookshelf.
In some cases, the antenna needs to be aimed at a certain height for accurate scanning. For example, if the robot is near a bookshelf, the books on the bookshelf may interfere with the antenna. In order to detect such obstacles, a sensor (contact sensor) was built into the antenna.

Purpose: Move the antenna to a specific height.
Required action:
1. 1. Move the antenna upwards.
・ Result: Success (height reached) / Failure (obstacle detection)
2. 2.
a) Rotate (keep the antenna away from obstacles)
b) Move the antenna upward.
・ Result: Success (reaching height) / Failure (discovering obstacles)
c) Rotate the back.

The tasks of this example are managed by identifying potential failures before performing the action. The above task is executed on the premise that a simple subtask such as going in and out as shown in steps a and c does not fail and succeeds.

Example 2: Hybrid Framework 1
In this example, action 1 is the default action to achieve the goal. If the system reports a failure, the system will try three times to avoid false alarms from the sensor. Action 2 is an alternative action to action 1. FIG. 6 shows a task management system.

Case 2: Automatic docking When the robot's battery is low or below a preset threshold (eg 20%), an automatic docking task is started.

Purpose: Dock to the docking station and charge action 1. Navigate to the point in front of the docking station (navigation point).
・ Result: Judged as successful 2. Dock-Result: Success (docking) / Failure (not docked correctly)

In this embodiment, the robot cannot produce different dock results if the first attempt fails. Therefore, iterative recovery does not work. However, once the robot's position is reset to the navigation point, the dock task can be performed again.

Example 3: Hybrid framework 2
The robot executes action 1 and action 2 in order. If action 2 fails, the robot state is reset to the state at the end of action 1. This is repeated n times. This situation is shown in FIG.

Case 3: Point-to-point navigation Purpose: Move from one place to another.
Action: Move to the desired location 1. Result: Success (achieve the goal) / Failure (cannot reach the goal)
2. 2. Rotate on the spot 3. Reset obstacle memory

Navigation algorithms may not always be able to find a route if they are close to the minimum space required by the robot. Action 2 may plan different paths by introducing perturbations at the robot's position.

Action 3 is used to clear obstacles in the robot's memory, forcing the robot to consider only currently detected obstacles. This allows the robot to plan routes through previously detected but non-existent obstacles.

Actions 2 and 3 are alternatives for resetting if the navigation task fails and allowing subsequent attempts to produce different results.

Example 4: Hybrid framework 3
FIG. 8 shows a task management framework for a robot to execute action 1. If it fails, action 2 is executed and a retry is performed. If it fails, action 3 is executed and a retry is performed.

The invention may be integrated into an autonomous system such as a robot that realizes the lifting and lowering of telescopic arms that allow antennas to scan books placed at different heights on shelves. The antenna can be blocked if the book protrudes from the shelf and interfere with the scanning task. Also, if the telescopic arm is forcibly moved to collide with a book, the antenna may be damaged.

Example 5
The task management system is implemented in an autonomous robot for moving from one point to another. Autonomous robots can get stuck in a certain place, blocked by nearby obstacles, in local areas where the route to the destination cannot be planned. For example, if the route from the robot's current position to the destination is blocked, the robot cannot plan the route. If a static obstacle is too close, the robot will not be able to find a path.

If the route planning fails, the navigation node automatically signals that the task has failed. The robot may have a built-in mechanism for notifying the failure of the task. The mechanism may include a limit switch on top that turns on when the spring-held top cover hits an object above. Further, a similar limit switch may be added to the lower part, which is turned on when the lower cover held by the spring hits the lower object. This will generate a signal to inform the robot that the task cannot be completed.

It is also possible to perform a recovery operation. The recovery operation is to move the arm several times and try again, then remove the robot and try again. If the arm is successfully moved up and down to scan the RFID tag, the task is considered successful.

In another embodiment of the present invention, the current position of the robot can be effectively changed. Reset tasks such as spinning in place are used to rescue the robot in this situation. After rotating, the position may be updated again, resulting in a slight change in position on the global map. Therefore, it becomes possible to find a route to the destination.

Example 6
Autonomous robots are tasked with checking whether their position is correct when performing some movement. The position information of the robot may not be able to be detected by the robot in an environment where there are few scan matching sensors. Without the correct location information, the robot cannot be detected and may not be useful in determining future actions to move around the environment. The lost detection mechanism stops the navigation in the environment where the robot is lost, and releases the robot only when the robot has the correct position information.

The algorithm for estimating the posture of the robot is expressed by the following equation.

pf = particle filter (robot, sensor, q, L, np)

As a result, the state of the robot is estimated via the scan matching sensor. q is the covariance of the diffusion noise applied to the particles, L is the covariance of the probability of the sensor model, and np is the number of particles.

Therefore, the covariance of the particle filter function can be used to detect whether the robot is sufficiently localized. Robustness of localization can be justified as long as the elements of the covariance matrix are below a certain threshold.

For example, if the covariance is greater than or equal to a threshold of about 1000 times the nominal value, the position task is determined to have failed. This nominal value is the covariance for robots that operate under normal operating conditions without performing disaster recovery.

In other embodiments, machine learning methods may be implemented to automate the process of detecting if the robot is well localized. This is implemented by using the correct position covariance (Ln) as the training data to generate the nominal value and the wrong position covariance (Lw) as the training day to determine the threshold. You may. A properly positioned robot indicates Ln> Lw.

The above shows that the determined threshold can be an absolute value, but it is understood that the threshold depends on various factors including the sensor used in the robot and the environment in which the robot operates. Will be done. Thus, the ratio of Lw to Ln provides an index for the threshold. The value is generally 100 or more.

In another embodiment of the invention, a recovery task is triggered and the robot is reset to a known posture in the environment via past navigation trajectories. If the robot can localize itself to a predefined map, the robot can navigate to the next task.

For tasks managed by "Recovery by Reset" and "Recovery by Alternatives frameworks", the tasks may be performed autonomously or manually. For example, a health monitoring system may be developed and incorporated into a system in which human behavior functions as a recovery task for managing disability conditions. Further, if a specific task fails, the user may provide a course of the next action so that the work flow is not interrupted.

In a further embodiment, the system determines recovery frameworks A, B, C (FIGS. 3, FIG. 4, FIG. 5, respectively, to determine a recovery framework that optimally provides task management to achieve the task objectives. Then) ranking may be adopted. For example, in the framework of "recovery by alternative means" in which task A takes higher priority to achieve its purpose, the system may be configured to try task B only if task A fails. However, if task A is more likely to fail than task B, the execution order of the alternatives can be changed. Therefore, it is possible to improve the system efficiency.

The above embodiments are provided solely for illustration purposes of the present invention, and further modifications and improvements are described herein of the present invention, as will be apparent to those skilled in the art concerned. It will be understood that it is considered to be included in the broad sense of. Further, it is understood that the features of one or more of the described embodiments may be combined to form further embodiments.

Claims (21)

A task management method for executing a task having a task goal, which includes the following steps.
Receiving task information to define a set of actions that have one or more actions to perform a task,
Determining the time calculated based on the set of actions defined above,
To determine whether the task goal is achieved by determining whether the calculated execution time exceeds the actual execution time.
Obtaining one or more recovery responses associated with the task goal if the task goal is not achieved.
To generate an execution plan based on the one or more recovery responses in order to obtain a set of updated actions that reflect the one or more recovery responses to manage the task. The task management method according to claim 1, wherein the task information is composed of characteristics of a series of events. The task management method according to claim 1 or 2, further comprising executing the generated execution plan for the continuous execution of the series of events in the event of a failure. The task management method of claim 1, further comprising developing the expertise based on observing the dynamic response to the failure to obtain the one or more recovery responses. The task management method of claim 4, further comprising ranking a plurality of recovery responses based on the task goals in order to derive said recovery response for managing failures. The task management method according to any one of claims 1 to 5, wherein the recovery response comprises one or more of execution of an alternative set of actions, return to a previous non-failure state, and re-execution of a task. .. The task management system according to claim 1, wherein the task goal is configured to confirm information previously obtained from a user or past events. The task management system according to claim 1, further comprising calculating an index of likelihood of localization by calculating a covariance function based on the following equation.

pf = particle filter (robot, sensor, q, L, np)

q = covariance of diffusion noise applied to the particles, L = covariance of the probability of the sensor model, and np = number of particles. The task management system according to any one of claims 1 to 8, wherein the task information further includes characteristics of a plurality of tasks. The task management system according to claim 9, wherein the plurality of tasks are nested or cascaded. A task management system for executing tasks with task goals.
With the control module
Memory module and
It has a task manager, and the control module, the memory module, and the task manager are interconnected via a communication means.
The control module and the task manager
Receives task information that defines a set of actions that have one or more actions to perform a task.
Determine the time calculated based on the set of actions defined above and
It operates to determine whether or not the task goal has been achieved by determining whether or not the calculated execution time exceeds the actual execution time.
The control module and the memory module
If the task goal is not achieved, it acts to get one or more return responses associated with the task goal.
The control module and the task manager
A task management system that operates to generate an execution plan based on one or more return responses in order to obtain a set of updated actions that reflect one or more return responses to manage the task. The task management system according to claim 11, wherein the task information is composed of characteristics of a series of events. The task management system according to claim 11, wherein the task manager executes the generated execution plan for continuous execution of the series of events when a failure occurs. 11. The task management system of claim 11, wherein the control module applies expertise based on observation of a dynamic response to a failure to obtain an appropriate recovery response. The task management system according to claim 14, wherein the control module ranks a plurality of recovery responses based on the task goal in order to derive the appropriate recovery response. The task management system according to claim 14 or 15, wherein the control module confirms information previously obtained from a user or past events in order to determine whether or not the task goal has been achieved. A claim that the memory module sends the acquired recovery response to the control module, which consists of one or more of execution of an alternative set of actions, return to a previous non-failure state, and re-execution of a task. 11. The task management system according to 11. The task management system according to claim 11, wherein the system further calculates an index of the likelihood of localization of the system by calculating a covariance function based on the following equation.

pf = particle filter (robot, sensor, q, L, np)

q = covariance of diffusion noise applied to the particles, L = covariance of the probability of the sensor model, and np = number of particles. The task management system according to any one of claims 11 to 18, wherein the task information further includes characteristics of a plurality of tasks. A computer program product comprising instructions to cause the computer to perform the method of claim 1-10 when the program is executed by the computer. A computer-readable medium in which the computer program product according to claim 20 is stored.

Images