KR20240003844A - Method, system, and program for creating work and movement plan for repositioning object through collaboration of multiple manipulator robots - Google Patents

Abstract

The method of establishing a work and movement plan for multiple manipulator robots to rearrange an object through collaboration of the present invention includes calculating an independent object relocation work plan for each robot, collecting the object relocation work plans of the individual robots, and performing the work of the multiple robots. A collaboration planning step in which the collaboration plan of multiple robots is calculated to maximize the number of shifts, a unit task sequence determination step in which the unit task sequence of multiple robots is determined to perform the collaboration plan in the shortest time, and a task sequence in which multiple robots do not collide with each other. In order to execute, it may include a movement planning step of establishing a movement plan for multiple robots and a step of controlling multiple robots based on the established movement plan.

Description

Method, system, and program for establishing work and movement plans for relocating objects through collaboration of multiple manipulator robots {METHOD, SYSTEM, AND PROGRAM FOR CREATING WORK AND MOVEMENT PLAN FOR REPOSITIONING OBJECT THROUGH COLLABORATION OF MULTIPLE MANIPULATOR ROBOTS}

The present invention relates to a method, system, and program for establishing a task and movement plan for relocating an object through the collaboration of multiple manipulator robots, and more specifically, to reach a target object in a limited space where an overhand grip is not permitted. It relates to methods, systems, and programs for establishing work and movement plans for relocating objects through collaboration of manipulator robots.

Recently, the development of autonomous mobile robots that perform autonomous movement and provide certain services has been active, and one of the important roles of these autonomous mobile robots is to extract or remove targets within a specific environment or specific space. .

In other words, there is an increasing need to perform the role of extracting or removing specific targets on behalf of humans in complex environments such as logistics transfer work in factories or dangerous environments such as space exploration work, nuclear waste disposal sites, or the deep sea. .

Accordingly, technologies related to target extraction or target removal within a specific environment or specific space using a manipulator robot are being developed, and Korean Patent No. 10-2338855 (hereinafter referred to as the prior patent) describes a method of relocating obstacles in a limited environment. The technology has been disclosed.

However, the prior patent is a method of relocating obstacles using a single manipulator robot, and is difficult to apply to situations where multiple manipulator robots are used to increase the efficiency of relocation work in a limited environment.

Accordingly, there is a demand for technology to rearrange objects through collaboration between multiple manipulator robots.

This specification provides a method for establishing work and movement plans for rearranging objects through the collaboration of multiple manipulator robots in order to more intuitively, effectively, and quickly relocate obstacles using a plurality of manipulator robots within a specific limited environment or space. The purpose is to provide.

This specification is not limited to the above-mentioned tasks, and other tasks not mentioned will be clearly understood by those skilled in the art from the description below.

In order to achieve the above-described object, a method of establishing a work and movement plan for multiple manipulator robots to relocate an object through collaboration according to an embodiment of the present invention includes calculating an independent object relocation work plan of an individual robot, the individual A collaboration planning step of calculating a collaboration plan of multiple robots by collecting object rearrangement work plans of robots, a unit task sequence determination step of determining a unit task sequence of the plurality of robots in order to perform the collaboration plan in the shortest time, and the plurality of robots. a movement planning step of establishing a movement plan for the plurality of robots to execute the task sequence without conflicting with each other; And it may include controlling the plurality of robots based on the established movement plan.

Additionally, the collaboration planning step may calculate a collaboration plan so that the number of work shifts of the multiple robots is maximized.

In addition, the collaboration planning step includes collecting the work plans of the individual robots and then generating a tree to search for the optimal solution of the collaboration plan. In the process of exploring multiple scenarios in which the multiple robots move objects, the robot's A job costing step that increases the cost of performing work if no work shifts occur; and a goal attainment confirmation step to determine whether all tasks have been performed and the final goal has been achieved.

In addition, in the step of creating the tree, the arrangement state of the object that changes as the unit task belonging to the work plan of the individual robot is performed is defined as one node, and the node that is the next search target is selected to set the search priority. Specify and add it to the search queue, and perform the next unit operation on the search target node with the highest priority in the queue to rearrange the object and create a new child node to increase the depth of the tree, and follow the above method to increase the depth of the tree. By exploring nodes, you can expand the tree until you reach your goal.

In addition, the work cost calculation step is to maximize the number of shifts of the plurality of robots when performing the work plan of the individual robots, without incurring costs when the plurality of robots can perform shifts and not incurring costs when shifts are not performed. Unit operation costs may be incurred.

In addition, the unit task sequence determination step determines the unit tasks of the individual robots for executing the collaboration plan, and executes the unit tasks within the shortest time so that the plurality of robots rearranges the target object within the minimum time. The order of work can be determined to achieve this.

Meanwhile, a robot collaboration system in which multiple manipulator robots perform work and movement plan establishment for object relocation through collaboration includes a work planning unit that calculates an independent object relocation work plan for individual robots, and an object relocation work plan for the individual robots. A collaboration planning unit that collects and calculates the collaboration plan of multiple robots, a unit task sequence determination unit that determines the unit task sequence of multiple robots in order to perform the collaboration plan in the shortest time, and multiple robots execute the task sequence without collision. In order to do this, it may include a movement planning unit that establishes a movement plan for the robot and a control unit that controls the plurality of robots based on the established movement plan.

Additionally, the collaboration planning unit may calculate a collaboration plan to maximize the number of work shifts of the multiple robots.

In addition, the collaboration planning unit includes a process of collecting work plans of the individual robots and then generating a tree for searching for an optimal solution to the collaboration plan; A task cost calculation process that increases task performance costs if task shifts of the plurality of robots do not occur in the process of exploring a plurality of scenarios in which the plurality of robots move objects; And it may further include a goal attainment confirmation process to determine whether all tasks have been performed and the final goal has been achieved.

In addition, the tree creation process defines the arrangement state of an object that changes as a unit task belonging to the work plan of the individual robot is performed as a node, selects the next search target node, and searches by specifying a search priority. Add to the queue, and perform the next unit operation on the search target node with the highest priority in the queue to rearrange the object and create new child nodes to increase the depth of the tree, and remove all nodes in the queue as above. By exploring, you can expand the tree until you reach your goal.

In addition, the work cost calculation process is performed to maximize the number of shifts of the multiple robots in carrying out the work plan of the individual robots, without incurring costs when the multiple robots can perform shifts and not incurring costs when the multiple robots can perform shifts. Unit operation costs may be incurred.

In addition, the unit task sequence determination unit determines the unit tasks of the individual robots for executing the collaboration plan, and executes the unit tasks within the shortest time to achieve a target object relocation state for the multiple robots within the minimum time. You can determine the order of work to achieve it.

The collaboration planning unit may calculate a collaboration plan by selecting a robot that can immediately perform a task among the plurality of robots.

According to the present specification, a work plan and a movement plan of a manipulator robot can be established to simultaneously perform a task for relocating an object.

Additionally, the present invention can minimize the execution time of the relocation plan by performing tasks simultaneously with multiple manipulator robots.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

Figure 1 is a flowchart showing a method of establishing a work and movement plan for multiple manipulator robots to rearrange an object through collaboration according to an embodiment of the present invention.
Figure 2 is an exemplary diagram showing a relocation work plan according to an embodiment of the present invention.
Figure 3 is a flowchart showing a method of calculating a collaboration plan according to an embodiment of the present invention.
Figure 4 is an exemplary diagram showing a tree for searching for an optimal solution according to an embodiment of the present invention.
Figure 5 is an exemplary diagram showing the work sequence of a manipulator robot according to an embodiment of the present invention.
Figure 6 is a block diagram showing the configuration of a robot collaboration system according to an embodiment of the present invention.
Figure 7 is an exemplary diagram showing an experiment set according to an embodiment of the present invention.
Figures 8 to 10 are exemplary diagrams showing experimental results according to an embodiment of the present invention.

The following merely illustrates the principles of the invention. Therefore, those skilled in the art will be able to invent various devices that embody the principles of the present invention and are included in the spirit and scope of the present invention, although not explicitly described or shown herein. In addition, all conditional terms and examples listed herein are, in principle, expressly intended only for the purpose of enabling the concept of the invention to be understood, and should be understood not as limiting to the examples and states specifically listed as such. do.

The above-described purpose, features and advantages will become clearer through the following detailed description in conjunction with the accompanying drawings, and accordingly, those skilled in the art will be able to easily implement the technical idea of the present invention. There will be.

Additionally, in describing the present invention, if it is determined that a detailed description of known technologies related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description will be omitted.

Hereinafter, various embodiments of the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a flowchart showing a method of establishing a work and movement plan for rearranging an object through the collaboration of multiple manipulator robots according to an embodiment of the present invention.

Referring to FIG. 1, the robot collaboration system 100 for establishing a work and movement plan for relocating an object through the collaboration of multiple manipulator robots can calculate an independent object relocation work plan for each robot (S100). Here, the object may be composed of a target, which is the object that the manipulator robot must reach, and an obstacle, which is an object that gets in the way of reaching the target.

Hereinafter, the object that the robot must reach among the objects will be defined and explained as a target.

Specifically, the size and location of targets and objects (or obstacles) can be recognized from images taken of the work space. Here, the image is an image taken from the robot collaboration system 100 or an external imaging device, and the work space may mean a limited space.

In addition, when the manipulator robot can move between any two objects among a plurality of objects without collision, the robot collaboration system 100 uses a traversability graph (hereinafter referred to as a T-graph) connecting any two objects with an edge. ) can be configured. Here, the T-graph is a line that represents the relationship between two objects, and that relationship means the existence of a collision-free path, and the path can be a straight line or a curve.

Next, the robot collaboration system 100 can calculate an independent object relocation work plan for each manipulator robot by calculating the shortest path as the path with the fewest number of obstacles moving until the target is retrieved from the T-graph. Here, the work plan may be a plan indicating the order of objects to be rearranged to reach the target. In this regard, description will be made with additional reference to FIG. 2.

Figure 2 is an exemplary diagram showing a relocation work plan according to an embodiment of the present invention.

Figure 2(a) is an example diagram showing a work space and two manipulator robots, and Figure 2(b) is an example diagram showing a T-graph.

A plurality of obstacles ( , , , , , , , ) and target ( ) is deployed and two manipulator robots ( , ) is placed, the robot collaboration system 100 can configure a T-graph as shown in FIG. 2(b).

And, the robot collaboration system 100 targets ( ), the path with the least number of movement of obstacles is calculated as the shortest path, and the manipulator robot ( )'s relocation work plan ( = , , ) and manipulator robot ( )'s relocation work plan ( = , , ) can be calculated.

Referring again to FIG. 1, the robot collaboration system 100 may collect the object rearrangement work plans of the multiple manipulator robots and calculate the collaboration plan of the multiple manipulator robots (S200). In the collaboration plan calculation step (S200), the robot collaboration system 100 selects one of the uniform-cost search algorithm (Case 1) or the greedy algorithm (Case 2) to plan the collected object rearrangement work. Based on this, the collaboration plan of multiple manipulator robots can be calculated.

First, we will explain the Uniform-cost search (SEARCH) algorithm (Case 1), and then the Greedy algorithm (Case 2).

[Uniform Cost Search Algorithm (Case 1)]

Specifically, in the collaboration plan calculation step (S200), the robot collaboration system 100 may collect the object rearrangement work plans of the multiple manipulator robots and calculate the collaboration plan so that the number of work shifts of the multiple manipulator robots is maximized. In this regard, description will be made with additional reference to FIG. 3.

Referring to FIG. 3, the robot collaboration system 100 can collect work plans of individual manipulator robots and then generate a tree to search for the optimal solution of the collaboration plan (S210).

Specifically, in the tree creation step (S210), the robot collaboration system 100 defines the arrangement state of the object that changes in the work space as one node as it performs the unit task belonging to the work plan of the individual manipulator robot, and searches for the next Select the target node, specify the search priority, add it to the search queue, and perform the next unit operation on the search target node with the highest priority in the queue to rearrange the object and create a new child node. By doing so, you can increase the depth of the tree and expand the tree until the goal is achieved by searching all nodes in the queue using the above method.

In other words, the robot collaboration system 100 generates the arrangement status of objects in the workspace as an initial node, and as a unit task is performed, the arrangement status of the object changed in the workspace as a child node. By creating and repeating this, you can create a tree to search for the optimal solution of the collaborative plan. Here, the tree may represent multiple scenarios in which multiple manipulator robots move obstacles in a workspace.

Additionally, the robot collaboration system 100 may increase task performance costs if task shifts of the manipulator robots do not occur in the process of exploring multiple scenarios in which individual manipulator robots move objects (S220).

Specifically, in performing the work plan of the individual manipulator robot in the task performance cost calculation step (S220), in order to maximize the number of shifts of the robot, no cost is incurred when the robot can perform a shift, and when the robot is not able to perform a shift, no cost is incurred. This may incur operational costs.

That is, when the robot collaboration system 100 expands from an initial node to a child node according to the collected work plans of individual manipulator robots, a task performance cost can be charged on a node-by-node basis depending on whether work shifts between manipulator robots occur. .

Next, the robot collaboration system 100 may determine whether all tasks have been performed in the tree created for search to achieve the final goal and calculate a collaboration plan (S230). Specifically, the robot collaboration system 100 may determine whether the target has been reached by moving the obstacles and calculate a collaboration plan including a work plan and an allocation plan to minimize the cost of performing the work. Here, the allocation plan may be a plan indicating the allocation of a manipulator robot to move the object according to the work plan.

The above-described tree creation step (S210), task performance cost step (S220), and goal attainment confirmation step (S230) will be described with reference to FIG. 4.

Figure 4 is an exemplary diagram showing a tree for searching for an optimal solution according to an embodiment of the present invention.

Referring to Figure 4, the manipulator robots ( , ) work plan ( , ), then perform the Uniform-cost search algorithm to plan the work ( , ), two initial nodes (A, B) can be created from each root node (not shown). Here, the SEARCH algorithm executes a search across all work plans so that the solution relocates the fewest number of objects, and the search can end when the target state in which the target is finally relocated is found. At this time, nodes are defined alphabetically according to the order in which they were created. is a cost function that represents the cost of performing work on that node, for example may represent the cost of performing the task of node G.

And, from the two initial nodes (A, B), the child nodes represent the state of an object after it is relocated in the work space. At this time, the child node on the left through the red arrow is a red manipulator robot ( ), and the child node on the right through the blue arrow is a blue manipulator robot ( ) shows the state in which the object has been rearranged.

The initial node A can be expanded to create node C and node D. Then, Node B, Node C, and Node D with the same task execution cost are in the search queue, and among them, Node B, the oldest, is selected for expansion, creating Node E and Node F. Here, when expanding a node, the node with the lowest task operation cost among the nodes in the search queue list may be selected as a priority.

Next, node C is expanded to create node G and node H, so that node D, node E, node F, node G, and node H are in the queue list. At this time, node G is charged a task performance cost because no switching of the manipulator robot occurs between node C and node G. This happens.

Then, among the nodes with the lowest cost of performing the task, the oldest node D is selected, and node I and node J are added to the frontier list, and among the nodes with the cost of performing the task of 0, node E and node F are sequentially expanded to form nodes K and Node L is created. At this time, the manipulator robot ( ) cannot access the object, so only the correct subtree is created.

Next, after node H is expanded, node M is created, and the target is removed from node M, but target confirmation is performed after the node is selected for expansion.

Then, node I is expanded and node N is created. At this time, the nodes with a task performance cost of 0 are node K and node N. After node O is created, node N is selected for expansion, and after checking whether all tasks have been performed and the final goal has been achieved, the search ends by finding the goal state N. At this time, the nodes highlighted in green in FIG. 5 are the determined collaboration plan, and the robot collaboration system 100 is ( = , , ) work plan and ( = , , ) can determine a collaboration plan that includes an allocation plan.

That is, the robot collaboration system 100 can calculate a collaboration plan so that the number of work shifts for multiple manipulator robots is maximized.

Next, the greedy algorithm (Case 2) will be explained.

[Greed Algorithm (Case 2)]

Specifically, in the collaboration plan calculation step (S200), the robot collaboration system 100 collects the object rearrangement work plans of multiple manipulator robots and selects multiple manipulator robots that can immediately perform the task based on this to calculate the collaboration plan. there is.

Specifically, when the work plans of each manipulator robot are collected, the GREEDY algorithm selects the manipulator robot that can immediately perform the task at each turn. At this time, the selection of the manipulator robot may be selected based on the task performance cost used in the uniform-cost search algorithm. Here, unlike the SEARCH algorithm, which adds up the cost of performing all tasks up to the current node, the GREEDY algorithm only determines whether a switch to the manipulator robot will occur in this turn.

Additionally, when performing the first task, there is no concept of switching of the manipulator robot, so the manipulator robot that is closer to the object may be selected. Afterwards, the manipulator robot that did not perform a relocation operation in the previous turn can reposition the next object.

And, if only one manipulator robot can access the object, the corresponding manipulator robot can be selected. Additionally, if the manipulator robot cannot approach the object, all manipulator robots calculate the distance between the accessible objects, and the manipulator robot with the shortest distance can be selected. In this way, the collaboration plan can be calculated.

Meanwhile, in the collaboration plan calculation step (S200), the robot collaboration system 100 uses the uniform-cost search algorithm (Case 1) or the greedy algorithm in consideration of planning time or execution time depending on the usage environment. You can calculate the collaboration plan of multiple manipulator robots by selecting one of (Case 2).

Referring again to FIG. 1, the robot collaboration system 100 may determine the unit task sequence of multiple manipulator robots in order to perform the calculated collaboration plan in the shortest time (S300).

Specifically, in the unit task sequence determination step (S300), the robot collaboration system 100 identifies the unit tasks of each manipulator robot to implement the calculated collaboration plan, and executes the unit tasks within the shortest time so that a plurality of manipulator robots can The work order can be determined to achieve the target relocation status within the minimum amount of time.

That is, in the unit task sequence determination step (S300), the robot collaboration system 100 may determine the task sequence so that the time (makespan) for all tasks to be completed is minimized. At this time, the unit work of the manipulator robot can be divided into pick, place, and standby.

This will be described with reference to FIG. 5 .

Figure 5 is an exemplary diagram showing the work sequence of a manipulator robot according to an embodiment of the present invention.

Work plan( = , , ) and allocation plan ( = , , ), two manipulator robots ( ) The unit work sequence can be determined as shown in Figure 5(a), and the work plan ( = , , ) and collaboration plan ( = , , ), two manipulator robots ( ) can be determined as shown in Figure 5(b).

Meanwhile, in order to make the most of multiple manipulator robots, manipulator robots that do not perform pinching can be assigned to a standby state.

Next, when the unit task sequence of the manipulator robots is determined, the robot collaboration system 100 may establish a movement plan for the plurality of manipulator robots so that the plurality of manipulator robots can execute the determined task sequence without colliding with each other (S400). Here, if calculated using the SEARCH algorithm in the collaborative planning step (S200), a movement plan is established for all sequenced movements, and if the GREEDY algorithm is used, a movement plan is established after each manipulator robot task pair is determined. can be established.

Additionally, the robot collaboration system 100 can control multiple manipulator robots based on the established movement plan (S500).

In other words, the method of establishing work and movement plans for multiple manipulator robots of the present invention to relocate objects through collaboration includes calculating the entire work plan and allocation plan before execution and immediately assigning them to the relocation task and task after executing all tasks. Includes a method for finding a manipulator robot to become.

Meanwhile, the method for establishing work and movement plans for relocating objects through collaboration described above has been explained based on two manipulator robots, but is not limited to this and can also be applied to three or more manipulator robots.

Next, the configuration of the robot collaboration system 100 will be described with reference to FIG. 6.

Referring to FIG. 6, the robot collaboration system 100 includes all or a work planning unit 110, a collaboration planning unit 120, a unit task ranking unit 130, a movement planning unit 140, and a control unit 150. Some may be included.

The work planning unit 110 may calculate an independent object relocation work plan for each manipulator robot.

Specifically, the work planning unit 110 recognizes the size and location of the target and objects (or obstacles) from images taken of the work space, and allows the manipulator robot to move between any two objects among a plurality of objects without collision. In this case, a traversability graph (hereinafter referred to as a T-graph) can be constructed that connects any two objects with an edge.

In addition, the work planning unit 110 may calculate an independent object relocation work plan for each manipulator robot by calculating the shortest path as the path with the least number of movements of the object until the target is retrieved from the T-graph.

The collaboration planning unit 120 may calculate the collaboration plan of the multiple manipulator robots by collecting the object rearrangement work plans of the multiple manipulator robots.

Specifically, the collaborative planning unit 120 includes a tree creation process 122, a task cost calculation process 124, and a goal reach confirmation process 126, and a uniform-cost search algorithm (Case 1) Alternatively, one of the GREEDY algorithms (Case 2) can be selected to calculate the collaboration plan of multiple manipulator robots based on the collected object rearrangement work plan.

When using the uniform-cost search algorithm (Case 1), the tree creation process 122 can collect the work plans of individual manipulator robots and then generate a tree to search for the optimal solution of the collaborative plan.

Specifically, the tree creation process 122 defines the arrangement state of an object that changes as a unit task belonging to the work plan of an individual manipulator robot is performed as one node, and selects the next search target node to set the search priority. Specify and add it to the search queue, and perform the next unit operation on the search target node with the highest priority in the queue to rearrange the object and create a new child node to increase the depth of the tree, and follow the above method to increase the depth of the tree. By exploring nodes, you can expand the tree until you reach your goal.

The task cost calculation process 124 may increase the task performance cost if the robot's task shift does not occur in the process of exploring multiple scenarios in which the manipulator robot moves an object.

The goal attainment verification process 126 may determine whether all tasks have been performed and the final goal has been achieved.

That is, the collaboration planning unit 120 may calculate a collaboration plan to maximize the number of work shifts.

In addition, when using the GREEDY algorithm (Case 2), the collaboration planning unit 120 collects the object rearrangement work plans of multiple manipulator robots and selects multiple manipulator robots that can immediately perform the task based on this to collaborate. Plans can be calculated.

That is, the collaboration planning unit 120 may calculate a collaboration plan by selecting a manipulator robot that can immediately perform a task among multiple manipulator robots.

The unit task ranking unit 130 may determine the unit task sequence of multiple manipulator robots in order to perform the calculated collaboration plan within the shortest time.

Specifically, the unit task ranking unit 130 identifies the unit tasks of each manipulator robot to implement the calculated collaboration plan, and executes the unit tasks within the shortest time so that multiple manipulator robots can reach their target within the minimum time. The task sequence can be determined to achieve the target's relocation status.

The movement planning unit 140 may establish a movement plan for the manipulator robot so that multiple robots can execute the work sequence without collision.

The control unit 150 can control the plurality of robots based on an established movement plan.

Next, the results of an experiment on the method of establishing a work and movement plan for relocating an object through collaboration between the above-described multiple manipulator robots are explained.

Experiment example

This experiment was conducted in two experimental sets. Specifically, the first set of experiments evaluated the SEARCH algorithm and the GREEDY algorithm without dynamic simulations, as shown in Figure 7(a), and measured objective values, which are the number of switches and switching speed. Since every solution may have a different number of tasks (and therefore a different number of conversion opportunities), showing ratios allows for comparison.

Additionally, other performance-related values are discussed, such as number of relocated objects and replans, success rate within time limit (200 seconds), task planning and movement planning (TAP) time.

The second set of experiments was conducted in a dynamic simulations environment implemented using Unity-ROS integration, as shown in Figure 7(b).

Meanwhile, for reasonable comparison, the DISTANCE algorithm and the RANDOM algorithm were used. The DISTANCE algorithm assigns tasks according to the Euclidean distance between the object and the manipulator robot, if both manipulator robots have access to the object to be relocated, and given the task plan, the first object has access to both manipulator robots. If both manipulator robots can access the object, the manipulator robot closer to the object is selected. Additionally, if only one manipulator robot can access an object, that manipulator robot may be selected. Additionally, if conflict-free motion is not found for both manipulator robots, the action plan can be updated for re-planning, and the selection can be repeated until the target is found.

The RANDOM algorithm works similarly to the DISTANCE algorithm, but can randomly assign tasks if both manipulator robots have access to the object to be relocated.

This experiment uses RRTConnect for movement planning implemented in MoveIt's Open Motion Planning Library, and a system equipped with 32G RAM and AMD Ryzen 7 2700 3.2GHz was used for the experiment.

The dimensions of the workspace (confined space) are 1.1m

first set of experiments

We test the four methods with 30 instances for each N = 12, 16, 20 where object positions are sampled randomly and uniformly, and to ensure a fair comparison, the set of test instances is the same for all methods. Additionally, the task plan and travel planning times for each task are determined to evaluate the average time required to generate a plan for each task.

Referring to Figures 8, 9(a) and 9(b), the method that involves more switching between manipulator robots is much better at finding a solution.

When the instance set is N = 12, the conversion rate of the SEARCH algorithm is 82.2%, which means that the manipulator robot shifts work more than 4 out of 5 times.

An example of task allocation with this ratio is ( = , , , , , ) can be.

For the most complex set of instances, N = 20, SEARCH reduced the work shift rate by up to 70.8% and at least 34.3% compared to the DISTANCE and RANDOM algorithms, respectively, resulting in more work shifts. can parallelize more tasks, resulting in a shorter completion time (makespan) for all tasks, leading to efficient work planning for the manipulator robot.

Although the GREEDY algorithm does not perform as well as the SEARCH algorithm, it shows a higher task conversion rate than the comparison results. The number of shifts in the GREEDY algorithm is greater than that in the SEARCH algorithm, and this result does not contradict the fact that the SEARCH algorithm is optimal. Here, the GREEDY algorithm relocates more objects on average, so the likelihood of task shifts is higher. This is why you need to focus on switching speed.

Additionally, despite the exponential time complexity of the SEARCH algorithm, the actual planning time (TAMP time in Figure 8) is not extremely long.

Compared to the others (see Figure 9(b)), the planning time is up to 21.2% longer (when N = 20), about 18 seconds longer. One reason for the limited planning time is that each node in the search tree has a branch coefficient of 2 (i.e., has at most two children) because each branch is for each manipulator robot performing the task.

Additionally, because the number of objects to be relocated does not exceed 4 on average in the case of the SEARCH algorithm, the search depth is also limited. Therefore, the SEARCH algorithm can be executed efficiently even if the theoretical time complexity is low. The analysis in Figure 9(b) shows that the motion planning time is the shortest in the SEARCH algorithm, and the main reason is that the SEARCH algorithm does not request frequent replanning, resulting in many computationally expensive movement plan queries. Because it doesn't work.

Additionally, the success rate measured the probability of returning a solution within the time limit (200 seconds), and all methods showed high success rates.

In addition, the GREEDY algorithm runs faster than the SEARCH algorithm, but the time difference was not large because the SEARCH algorithm did not replan much in the experiment. If the environment were more dynamic, replanning would occur more frequently and the difference would be expected to increase.

Second set of experiments

In this experiment, we demonstrate the feasibility of executing the method with a real manipulator robot by executing the relocation plan in a dynamic simulation environment. Results for 30 instances with N = 20 are shown in Figures 9(c), 9(d), and Figure 10.

The SEARCH algorithm shows the shortest execution time with the highest number of work shifts. This indicates that the SEARCH algorithm can generate an efficient work plan for execution at runtime.

The reduction in execution time is up to 27.3% and minimum 22.9% compared to the DISTANCE algorithm and the RANDOM algorithm, respectively.

The GREEDY algorithm shows slightly better performance, and the low reduction in execution time (up to 9%) results from short-sighted decision-making in contrast to the myopic SEARCH algorithm.

Furthermore, various embodiments described herein may be implemented in a recording medium readable by a computer or similar device, for example, using software, hardware, or a combination thereof.

According to hardware implementation, the embodiments described herein include application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), and field programmable gate arrays (FPGAs). It may be implemented using at least one of processors, controllers, micro-controllers, microprocessors, and other electrical units for performing functions. In some cases, as described herein, The described embodiments may be implemented as a control module itself.

According to software implementation, embodiments such as procedures and functions described in this specification may be implemented as separate software modules. Each of the software modules may perform one or more functions and operations described herein. Software code can be implemented as a software application written in an appropriate programming language. The software code may be stored in a memory module and executed by a control module.

The above description is merely an illustrative explanation of the technical idea of the present invention, and various modifications, changes, and substitutions can be made by those skilled in the art without departing from the essential characteristics of the present invention. will be.

Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are not intended to limit the technical idea of the present invention, but are for illustrative purposes, and the scope of the technical idea of the present invention is not limited by these embodiments and the attached drawings. . The scope of protection of the present invention should be interpreted in accordance with the claims below, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of rights of the present invention.

Claims (16)

In a method of establishing a work and movement plan for rearranging an object through the collaboration of multiple manipulator robots,
calculating independent object relocation operation plans for individual robots;
A collaboration planning step of calculating a collaboration plan for multiple robots by collecting the object relocation work plans of the individual robots;
A unit task sequence determination step of determining a unit task sequence of the plurality of robots in order to perform the collaboration plan in the shortest time possible;
A movement planning step of establishing a movement plan for the plurality of robots so that the plurality of robots execute the task sequence without colliding with each other; and
A method of establishing a work and movement plan comprising: controlling the plurality of robots based on the established movement plan. According to claim 1,
The collaboration planning step is a method of establishing a work and movement plan, characterized in that the collaboration plan is calculated so that the number of work shifts of the plurality of robots is maximized. According to claim 2,
The collaboration planning stage is,
Collecting the work plans of the individual robots and then generating a tree to search for an optimal solution to the collaboration plan;
A task cost calculation step of increasing task performance costs if task shifts of robots do not occur in the process of exploring a plurality of scenarios in which the plurality of robots move objects; and
A work and movement planning method comprising a goal-reaching verification step to determine whether all work has been performed and the final goal has been achieved. According to claim 3,
The step of creating the tree is,
The arrangement state of the object that changes as the unit task belonging to the work plan of the individual robot is performed is defined as one node,
Select the next search target node, specify search priority, and add it to the search queue.
Increase the depth of the tree by performing the next unit operation on the search target node with the highest priority in the queue, relocating objects and creating new child nodes;
A work and movement plan establishment method characterized by searching all nodes in the queue as described above and expanding the tree until the goal is achieved. According to paragraph 3,
The work cost calculation step is,
In order to maximize the number of shifts of the plurality of robots in carrying out the work plan of the individual robots, no cost is incurred when the plurality of robots can perform shifts, and when shifts are not performed, unit work costs are incurred. What to do and how to plan your move. According to paragraph 2,
The unit operation sequence decision step is,
Identify unit tasks of the individual robots to implement the collaboration plan,
A method for establishing a work and movement plan, characterized in that the task sequence is determined so that the unit tasks are executed within the shortest time so that the plurality of robots can achieve the target object rearrangement state within the minimum time. According to claim 1,
The collaboration planning step is a method of establishing a work and movement plan, characterized in that the collaboration plan is calculated by selecting the individual robot that can immediately perform the task among the plurality of robots. In a robot collaboration system in which multiple manipulator robots perform work and movement planning for relocating objects through collaboration,
a task planning unit that calculates an independent object relocation task plan for each robot;
a collaboration planning unit that calculates a collaboration plan for multiple robots by collecting the object relocation work plans of the individual robots;
a unit task sequence determination unit that determines the unit task sequence of multiple robots in order to perform the collaboration plan in the shortest time;
a movement planning unit that establishes a movement plan for the robots so that multiple robots can execute the work sequence without collision; and
A robot collaboration system including a control unit that controls the plurality of robots based on the established movement plan. According to claim 8,
A robot collaboration system wherein the collaboration planning unit calculates a collaboration plan to maximize the number of work shifts for the plurality of robots. According to clause 9,
The collaboration planning department said,
A process of collecting the work plans of the individual robots and then generating a tree to search for an optimal solution to the collaborative plan;
A task cost calculation process that increases task performance costs if task shifts of robots do not occur in the process of exploring a plurality of scenarios in which the plurality of robots move objects; and
A robot collaboration system further comprising a goal attainment confirmation process to determine whether all tasks have been performed and the final goal has been achieved. According to clause 10,
The tree creation process is,
The arrangement state of the object that changes as the unit task belonging to the work plan of the individual robot is performed is defined as one node,
Select the next search target node, specify search priority, and add it to the search queue.
Increase the depth of the tree by performing the next unit operation on the search target node with the highest priority in the queue, relocating objects and creating new child nodes;
A robot collaboration system characterized by searching all nodes in the queue according to the above method and expanding the tree until the goal is achieved. According to clause 10,
The work cost calculation process is,
In order to maximize the number of shifts of the plurality of robots in carrying out the work plan of the individual robots, no cost is incurred when the plurality of robots can perform shifts, and when shifts are not performed, unit work costs are incurred. A collaborative system that does. According to clause 8,
The unit operation sequence determination unit,
Identify unit tasks of the individual robots to implement the collaboration plan,
A robot collaboration system characterized in that the task sequence is determined so that the unit tasks are executed within the shortest time so that the plurality of robots can achieve the target object rearrangement state within the minimum time. According to clause 8,
A method of establishing a work and movement plan, wherein the collaboration planning unit calculates a collaboration plan by selecting the individual robot that can immediately perform the task among the plurality of robots. A computer-readable recording medium storing a program for performing the method of establishing a work and movement plan according to any one of claims 1 to 7. A program stored in a computer-readable recording medium including program code for executing the work and movement plan establishment method according to any one of claims 1 to 7.