Scheduling method and device based on robot, equipment and medium

Abstract

The embodiment of the invention provides a robot-based scheduling method and device, equipment and a medium. The method comprises the following steps: acquiring pose information of a plurality of robots and a scheduling demand of at least one target object needing to interact with the robots; performing path planning according to the pose information of the plurality of robots and the position information of the at least one target object to obtain a plurality of task paths; acquiring a plurality of scheduling plans according to the plurality of task paths and the robot; based on constraints of types of tasks which can be executed by each robot, task types in each scheduling demand, task execution time and a target number of the robots executing the tasks, performing simulation scheduling on the plurality of scheduling plans, and determining a target scheduling plan; and scheduling the robot to execute the task of the at least one target object according to the target scheduling plan. The method is used for achieving the effects of optimizing the scheduling plan of the robot and further improving the task execution efficiency.

Description

Technical Field

[0001] The present application relates to the field of scheduling and planning technology, and in particular to a robot-based scheduling method, device, equipment and medium. Background Art

[0002] With the rapid development of industrial automation, robots are increasingly being used in manufacturing, assembly, logistics, and other fields. To improve production efficiency and ensure the collaborative capabilities of robots in complex tasks, scientific and reasonable task scheduling methods have become a research focus.

[0003] In multi-task, multi-robot scenarios, effectively allocating robot resources, optimizing path planning, reducing task execution time, and ensuring stable and efficient execution are pressing technical challenges. However, existing scheduling methods, which rely on static allocation and traditional algorithms, struggle to ensure the accuracy and reliability of scheduling plans in multi-task, multi-robot scenarios, thus impacting overall task execution efficiency. Summary of the Invention

[0004] The embodiments of the present application provide a robot-based scheduling method, apparatus, equipment, and medium for optimizing the scheduling plan for the robot, thereby improving the efficiency of task execution.

[0005] In a first aspect, an embodiment of the present application provides a robot-based scheduling method, comprising:

[0006] Obtaining position information of multiple robots and scheduling requirements of at least one target object that needs to interact with the robots, each scheduling requirement including position information of the target object, task type, task execution time, and a target number of robots to perform the task;

[0007] Performing path planning based on the posture information of the multiple robots and the position information of the at least one target object to obtain multiple task paths;

[0008] Acquire multiple scheduling plans based on the multiple task paths and the multiple robots, each scheduling plan including a task path for each robot;

[0009] Based on the type of tasks that each robot can perform, the task type in each scheduling requirement, the task execution time, and the target number of robots that perform the tasks, the multiple scheduling plans are simulated and scheduled to determine a target scheduling plan;

[0010] The robot is scheduled to perform a task for the at least one target object according to the target scheduling plan.

[0011] In a possible implementation manner, the method of performing simulated scheduling on the multiple scheduling plans based on the type of tasks executable by each robot, the type of tasks in each scheduling requirement, the task execution time, and the target number of robots for executing tasks, and determining a target scheduling plan includes:

[0012] Performing simulated scheduling on the multiple scheduling plans, and processing the multiple scheduling plans based on the type of tasks executable by each robot, the type of tasks in each scheduling requirement, the task execution time, and the target number of robots for executing tasks, to obtain multiple intermediate scheduling plans; the intermediate scheduling plans are the scheduling plans in the multiple scheduling plans that meet the constraints;

[0013] Determining, from the multiple intermediate scheduling plans, the intermediate scheduling plan with the minimum sum of the path scheduling time and the execution time as the target scheduling plan.

[0014] In a possible implementation manner, the determining, from the multiple intermediate scheduling plans, the intermediate scheduling plan with the minimum sum of the path scheduling time and the execution time as the target scheduling plan includes:

[0015] For each intermediate scheduling plan among the multiple intermediate scheduling plans, respectively performing simulated execution on the plans corresponding to each target object in the intermediate scheduling plan, taking the path scheduling time of the robot that first arrives at the target object as the path scheduling time of the plan, and taking the time from the start of task execution at the target object to the end of task execution as the execution time of the plan;

[0016] Taking the sum of the path scheduling times of the plans corresponding to each target object in the intermediate scheduling plan as the path scheduling time of the intermediate scheduling plan, and taking the sum of the execution times of the plans corresponding to each target object in the intermediate scheduling plan as the execution time of the intermediate scheduling plan; each target object corresponding plan in the intermediate scheduling plan; the plan corresponding to the target object includes the task path of the robot for executing the task of the target object;

[0017] Determining the target scheduling plan with the minimum sum of the path scheduling time and the execution time according to the path scheduling time and the execution time corresponding to each intermediate scheduling plan.

[0018] In a possible implementation manner, during the process of performing simulated execution on the plans corresponding to each target object in the intermediate scheduling plan, if the number of robots involved in the plan exceeds one, the method further includes:

[0019] If the task of the target object allows any number of robots to execute, the execution time of the plan is the time from when the first robot arrives at the target object to start executing the task to the end of the task execution;

[0020] If the task of the target object does not allow any number of robots to execute, the execution time of the plan is the time from when all robots arrive at the target object to start executing the task to the end of the task execution.

[0021] In a possible implementation manner, the method further includes:

[0022] In the case where the task of the target object allows some robots to execute, the execution time is predicted by combining a trained LSTM prediction model and a multi-layer perceptron prediction model;

[0023] Or,

[0024] In the case where the task of the target object does not allow some robots to execute, the execution time is predicted by using a trained multi-layer perceptron prediction model.

[0025] In a possible implementation manner, the pose information of each robot includes the position information of the robot. The path planning based on the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths includes:

[0026] According to the position information of the multiple robots and the position information of the at least one target object, a depth-first search algorithm is used for feasible path planning to obtain the multiple task paths.

[0027] In a possible implementation manner, the obtaining the pose information of the multiple robots includes:

[0028] Obtain the working states of all currently schedulable robots, and the working state of each robot includes a working state or an idle state;

[0029] Obtain the pose information of the multiple robots in the idle state.

[0030] In a possible implementation manner, the simulating and scheduling the multiple scheduling plans, and processing the multiple scheduling plans based on the types of tasks that each robot can execute, the types of tasks in each scheduling requirement, the task execution time, and the constraint of the target number of robots executing the task, includes:

[0031] For multiple paths in each scheduling plan, according to the types of tasks that each robot can execute and the task types in each scheduling requirement, delete the paths where the types of tasks that the robot can execute do not match the task types corresponding to the paths.

[0032] For each scheduling plan, if there is any path in the scheduling plan where the time taken for a robot to execute the task of the target object corresponding to the scheduling requirement exceeds the task execution time in the scheduling requirement, then delete the scheduling plan.

[0033] For each remaining scheduling plan, if the number of robots that can execute the task of the target object corresponding to any scheduling requirement is less than the target number in the scheduling requirement, then delete the scheduling plan.

[0034] In a second aspect, an embodiment of the present application provides a robot-based scheduling device, including:

[0035] An acquisition unit, configured to acquire the pose information of multiple robots and the scheduling requirements of at least one target object that needs to interact with the robots. Each scheduling requirement includes the position information of the target object, the task type, the task execution time, and the target number of robots for executing the task.

[0036] A first processing unit, configured to perform path planning according to the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths.

[0037] A second processing unit, configured to obtain multiple scheduling plans according to the multiple task paths and the multiple robots. Each scheduling plan includes one task path for each robot.

[0038] A determination unit, configured to perform simulated scheduling on the multiple scheduling plans based on the constraints of the types of tasks that each robot can execute, the task types in each scheduling requirement, the task execution time, and the target number of robots for executing the task, and determine a target scheduling plan.

[0039] An execution unit, configured to schedule the robots to execute the tasks of the at least one target object according to the target scheduling plan.

[0040] In a possible implementation manner, the determination unit includes:

[0041] A first determination module, configured to perform simulated scheduling on the multiple scheduling plans, and process the multiple scheduling plans based on the constraints of the types of tasks that each robot can execute, the task types in each scheduling requirement, the task execution time, and the target number of robots for executing the task, to obtain multiple intermediate scheduling plans; the intermediate scheduling plans are the scheduling plans that meet the constraints among the multiple scheduling plans.

[0042] A second determination module, configured to determine, from the multiple intermediate scheduling plans, an intermediate scheduling plan with the minimum sum of path scheduling time and execution time as the target scheduling plan.

[0043] In a possible implementation manner, the second determination module is specifically configured to:

[0044] For each intermediate scheduling plan among the multiple intermediate scheduling plans, respectively simulate the execution of the plans corresponding to each target object in the intermediate scheduling plan, use the path scheduling time of the first robot reaching the target object as the path scheduling time of the plan, and use the time from the start of task execution at the target object to the end of task execution as the execution time of the plan; each intermediate scheduling plan includes the plans corresponding to each target object; the plan corresponding to the target object includes the task path of the robot executing the task of the target object.

[0045] Use the sum of the path scheduling times of the plans corresponding to each target object in the intermediate scheduling plan as the path scheduling time of the intermediate scheduling plan, and use the sum of the execution times of the plans corresponding to each target object in the intermediate scheduling plan as the execution time of the intermediate scheduling plan.

[0046] Determine the target scheduling plan with the minimum sum of path scheduling time and execution time according to the path scheduling time and execution time corresponding to each intermediate scheduling plan.

[0047] In a possible implementation manner, the robot-based scheduling device further includes:

[0048] A third processing unit, configured to, if the task of the target object is executed by any number of robots, use the time from the start of task execution by the first robot reaching the target object to the end of task execution as the execution time of the plan.

[0049] If the task of the target object does not allow any number of robots to execute, use the time from the start of task execution by all robots reaching the target object to the end of task execution as the execution time of the plan.

[0050] In a possible implementation manner, the robot-based scheduling device further includes:

[0051] A prediction unit, configured to, when the task of the target object allows some robots to execute, the execution time is obtained by combining and predicting using a trained LSTM prediction model and a multi-layer perceptron prediction model.

[0052] Or,

[0053] In the case where the tasks of the target object do not allow partial robots to execute, the execution time is predicted by using the trained multi-layer perceptron prediction model.

[0054] In a possible implementation manner, the first processing unit is specifically configured to:

[0055] According to the position information of the multiple robots and the position information of the at least one target object, perform feasible path planning by using the depth-first search algorithm to obtain the multiple task paths.

[0056] In a possible implementation manner, the obtaining unit is partially configured to:

[0057] Obtain the working states of all currently schedulable robots, where the working state of each robot includes a working state or an idle state;

[0058] Obtain the pose information of the multiple robots in the idle state.

[0059] In a possible implementation manner, the first determining module is partially configured to:

[0060] For the multiple paths in each scheduling plan, according to the types of tasks that each robot can execute and the types of tasks in each scheduling requirement, delete the paths where the types of tasks that the robot can execute do not match the types of tasks corresponding to the paths;

[0061] For each scheduling plan, if there is any path in the scheduling plan where the scheduling time for a robot to execute the task of the target object corresponding to the scheduling requirement exceeds the task execution time in the scheduling requirement, then delete the scheduling plan;

[0062] For each remaining scheduling plan, if the number of robots executing the task of the target object corresponding to any scheduling requirement is less than the target number in the scheduling requirement, then delete the scheduling plan.

[0063] In a third aspect, an embodiment of the present application provides a computer device, including: a memory, a processor;

[0064] The memory stores computer execution instructions;

[0065] The processor executes the computer execution instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementation manners of the first aspect.

[0066] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, where computer execution instructions are stored in the computer-readable storage medium, and when the computer execution instructions are executed by a processor, they are used to implement the above first aspect and / or various possible implementation manners of the first aspect.

[0067] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program which, when executed by a processor, implements the above first aspect and / or various possible implementation manners of the first aspect.

[0068] The robot-based scheduling method, device, equipment and medium provided by the embodiments of the present application first obtain the pose information of multiple robots and the scheduling requirements of at least one target object that needs to interact with the robots, then perform path planning based on the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths. Then, based on the multiple task paths and the robots, multiple scheduling plans are obtained. Next, based on the types of tasks that each robot can execute, the types of tasks in each scheduling requirement, the task execution time, and the constraint of the target quantity of the robots executing the tasks, the multiple scheduling plans are simulated and scheduled to determine the target scheduling plan. Finally, the robots are scheduled to execute the tasks of at least one target object according to the target scheduling plan. This approach constrains the scheduling of robots from multiple perspectives including task type, task execution time, and target quantity, ensuring the accuracy, reliability, and efficiency of the scheduling plan formulated in a multi-task and multi-robot scenario, achieving the optimization of the scheduling plan for robots and thus improving the task execution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application and used together with the specification to explain the principles of the present application.

[0070] Figure 1 It is a schematic flowchart of the robot-based scheduling method provided in Embodiment 1 of the present application;

[0071] Figure 2 It is a schematic structural diagram of the robot-based scheduling device provided in Embodiment 3 of the present application;

[0072] Figure 3 It is a schematic structural diagram of the robot-based scheduling device provided in Embodiment 4 of the present application;

[0073] Figure 4 It is a schematic structural diagram of the computer equipment provided by the present application.

[0074] Through the above drawings, the clear embodiments of the present application have been shown, and there will be more detailed descriptions hereinafter. These drawings and text descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0075] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0076] When studying robot-based scheduling methods, the inventors found that how to flexibly formulate accurate and reliable scheduling plans in dynamic scenarios of multiple tasks and multiple robots is a technical problem that needs to be solved urgently in the existing technology. To address this problem, the inventors found that by constraining the scheduling of robots from the perspectives of task type, task execution time, and target number, the accuracy, reliability, and efficiency of the scheduling plans formulated in multi-task, multi-robot scenarios can be ensured, thereby optimizing the scheduling plans for robots and improving the efficiency of task execution.

[0077] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0078] Figure 1 A flowchart of the robot-based scheduling method provided in Example 1 of this application is shown in FIG. Figure 1 As shown, the method includes:

[0079] S101. Obtaining position information of multiple robots and scheduling requirements of at least one target object that needs to interact with the robots, each scheduling requirement includes the position information of the target object, the task type, the task execution time, and the target number of robots that perform the task.

[0080] This step requires obtaining the robot's position information and the scheduling requirements of at least one target object with which the robot will interact. The robot's position information includes its current position in space (i.e., location information) and the posture of its arms. The task execution time in the scheduling requirements refers to the time limit for executing the target object's task.

[0081] It should be understood that the robot in this solution generally refers to a dual-arm robot; the target object includes the equipment, objects or task scenarios that the robot needs to interact with. For example, the target object can be parts to be assembled by the robot, materials to be moved, or equipment to be repaired, etc.

[0082] In a specific implementation, the scheduling requirements of the target object may also include specific requirements related to the operation, such as the type and size of the equipment, the operation steps, the accuracy requirements, etc. When the robot performs the task of the target object, it can determine the details and method of task execution based on the above information.

[0083] It should be understood that in this technical solution, the "target object" is the core focus of the robot's mission execution and the basis for the robot's demand-based scheduling and planning. By obtaining information about the target object's needs, the robot can develop a targeted scheduling plan to ensure that the mission execution achieves the expected mission objectives.

[0084] In a specific implementation, obtaining position information of multiple robots includes:

[0085] Step 1: Get the working status of all robots that can be scheduled currently. The working status of each robot includes working status or idle status.

[0086] Step 2: Obtain the position information of the plurality of robots in an idle state.

[0087] In this implementation, by obtaining the working status of all currently schedulable robots, the robots in the idle state can be dynamically determined, and scheduling plans can be formulated only for the robots in the idle state, thereby optimizing resource allocation in a multi-tasking environment and improving scheduling efficiency.

[0088] S102: Perform path planning based on the posture information of multiple robots and the position information of at least one target object to obtain multiple task paths.

[0089] In this step, for each robot that has obtained posture information, it is necessary to develop multiple paths for it to reach the position of each target object based on the position information in its posture information and the position information of the target object in the scheduling requirements of each target object. Then, multiple paths for each robot to reach the positions of different target objects can be obtained, and multiple task paths can be obtained.

[0090] In a specific implementation, a depth-first search algorithm may be used to perform feasible path planning based on the position information of multiple robots and the position information of at least one target object to obtain multiple task paths.

[0091] The depth-first search algorithm starts from the starting point and goes as deep as possible along one direction (or path) until it encounters an end point or an obstacle, then backtracks to the previous node to continue exploring other paths. This process generates all possible feasible paths.

[0092] Specifically, for each robot, when generating a path for the robot to reach the location of one of the target objects, a depth-first search algorithm is used to perform feasible path planning, which may include the following steps 1 to 4:

[0093] Step 1: Set the robot's current position as the starting point according to the input task requirements (e.g. , )), the target object's position is the end point (such as ( , ,)).

[0094] Step 2: Generate a grid map representing the robot's working environment. The grid map defines whether the robot can move at each location. 0 in the grid map represents a passable area, and 1 represents an obstacle.

[0095] Step 3: Based on the recursive search algorithm, the starting point and the end point are input into the recursive function dfs. Starting from the starting point, each time the node attempts to move in the four directions (up, down, left, and right), it checks whether it can move to the next node. If there is at least one direction from the current node that can move to the next node, the current node is set as the target node, and the current path corresponding to the target node is added to the result list. Otherwise, the recursive search continues until all possible paths corresponding to the starting point and the end point are found.

[0096] When using a recursive search algorithm for path search, if a path is blocked (e.g., encountering an obstacle or having already visited a node), the algorithm backtracks to the previous node and tries another direction. It should be understood that by backtracking, the recursive search algorithm ensures that all possible paths have been traversed.

[0097] Step 4. After the recursive search algorithm is executed, all feasible paths from the starting point to the end point are returned, where each feasible path is a list of coordinate points that represent the order in which the robot moves.

[0098] S103. Acquire multiple scheduling plans based on the multiple task paths and the multiple robots, where each scheduling plan includes a task path for each robot.

[0099] In this step, it is necessary to extract a path from at least one task path corresponding to each robot to form a scheduling plan; this extraction operation is repeated until no new path combinations appear, and all the extracted path combinations are summarized to obtain all the scheduling plans.

[0100] S104. Based on the type of tasks that each robot can perform, the type of tasks in each scheduling requirement, the task execution time, and the constraints of the target number of robots that perform the tasks, multiple scheduling plans are simulated to determine a target scheduling plan.

[0101] In this step, according to the types of executable tasks of each robot and the constraints of the task type, task execution time, and the target number of robots for executing tasks in the scheduling requirements corresponding to each target object, simulation scheduling processing is performed on each scheduling plan, and based on the processing results, a target scheduling plan is determined.

[0102] In a specific implementation manner, the types of executable tasks of each robot can be determined according to the task tags on each robot.

[0103] In a specific implementation manner, the target scheduling plan can be determined in the following manner of steps 1-2:

[0104] Step 1: Perform simulation scheduling on multiple scheduling plans. Based on the types of executable tasks of each robot, the task type, task execution time, and the constraint of the target number of robots for executing tasks in each scheduling requirement, process the multiple scheduling plans to obtain multiple intermediate scheduling plans.

[0105] In this step, simulation scheduling needs to be performed on each scheduling plan. For each scheduling plan, according to the types of executable tasks of each robot in the scheduling plan and the task type in each scheduling requirement, perform preliminary processing on the scheduling plan; then, according to the path scheduling time consumed by each robot in the scheduling plan to execute its corresponding task and the task execution time in each scheduling requirement, further constrain the scheduling plan; finally, based on the number of robots executing the tasks corresponding to any scheduling requirement in the scheduling plan (i.e., the actual number of robots assigned to each target object) and the target number corresponding to each scheduling requirement, obtain multiple intermediate scheduling plans. Among them, the intermediate scheduling plan is a scheduling plan in the multiple scheduling plans that meets the constraints of the task type, task execution time, and the target number of the dual-arm robots for executing tasks in the scheduling requirements.

[0106] Furthermore, performing simulation scheduling on multiple scheduling plans and processing the multiple scheduling plans based on the types of executable tasks of each robot, the task type, task execution time, and the constraint of the target number of robots for executing tasks in each scheduling requirement may include the following steps 1.1-1.3:

[0107] Step 1.1: For multiple paths in each scheduling plan, according to the types of executable tasks of each robot and the task type in each scheduling requirement, delete the paths where the type of executable task of the robot does not match the task type corresponding to the path.

[0108] In this step, for each scheduling plan, it is necessary to constrain each path from the perspective of the task type according to the type of the robot-executable task corresponding to each path in the scheduling plan and the task type in the scheduling requirements of the corresponding each target object, and delete the paths in the scheduling plan where the type of the robot-executable task does not match the task type corresponding to the path.

[0109] Specifically, the following formula can be used to constrain the robot according to the task type in the scheduling requirements:

[0110]

[0111] Wherein, represents whether the robot executes the task of target object j, represents not to execute, represents to execute; represents the task type required by target object ; represents the set of task types included in the task label of the robot ; represents the i-th robot. The above formula means that if the task type in the scheduling requirements of target object is included in the set of task types of the robot , then set the robot in the scheduling plan not to execute the task of target object j (that is, delete the path corresponding to the robot in the scheduling plan), otherwise, retain the path corresponding to the robot in the scheduling plan.

[0112] Optionally, in order to ensure that each robot can only be selected once when facing the tasks of multiple target objects, the following formula can be used to further constrain the robot:

[0113]

[0114] Step 1.2: For each scheduling plan, if there is any path in the scheduling plan where the scheduling time of the robot executing the task of the target object corresponding to the scheduling requirement exceeds the task execution time in the scheduling requirement, then delete the scheduling plan;

[0115] In this step, for each scheduling plan, since each robot in the scheduling plan has a task corresponding to a target object, and each target object corresponds to a scheduling requirement, after calculating the path scheduling time for each robot in the scheduling plan to execute the task of its corresponding target object, by comparing the path scheduling time with the task execution time in the scheduling requirement corresponding to the target object, if the path scheduling time corresponding to any robot in the scheduling plan exceeds the task execution time in the scheduling requirement corresponding to it, then the scheduling plan is deleted.

[0116] Specifically, for each scheduling plan, the following formula can be used to determine the path scheduling time for each robot in the scheduling plan to execute the task of its corresponding target object:

[0117]

[0118] In the above formula, represents the path scheduling time for robot to execute the task of target object j, represents the path length that robot takes to execute the task of target object j, represents the traveling speed of robot .

[0119] The following formula can be used to constrain the robot according to the task execution time in the scheduling requirement:

[0120]

[0121] In the above formula, represents the task execution time in the scheduling requirement of target object j.

[0122] Step 1.3: For each remaining scheduling plan, if the number of robots executing the task of the target object corresponding to any scheduling requirement is less than the target number in the scheduling requirement, then delete the scheduling plan.

[0123] In this step, for each remaining scheduling plan, it is necessary to constrain the scheduling plan from the perspective of the target number according to the number of robots executing the task of the target object corresponding to any scheduling requirement and the target number in the scheduling requirement of the target object, so as to ensure that the number of robots executing the task corresponding to any scheduling requirement in the obtained multiple intermediate scheduling plans is equal to the target number in the scheduling requirement of the target object. Therefore, for each remaining scheduling plan, if there is a situation where the number of robots executing the task corresponding to any scheduling requirement is less than the target number in the scheduling requirement, then delete the scheduling plan; if there is a situation where the number of robots executing the task corresponding to any scheduling requirement is equal to the target number in the scheduling requirement, then determine the scheduling plan as an intermediate scheduling plan.

[0124] Optionally, if the number of robots executing the tasks corresponding to any scheduling requirement in the scheduling plan exceeds the target number in the scheduling requirement, for this scheduling plan, multiple intermediate scheduling plans can be obtained based on this scheduling plan in the following manner:

[0125] Obtain at least one target object in the scheduling plan that does not meet the above target number requirement, and multiple paths corresponding to each target object. For each target object that does not meet the above target number requirement, according to the target number in the scheduling requirement of the target object, perform permutation and combination on the multiple paths of the tasks executed by the target object in the scheduling plan to obtain multiple path combinations; for each scheduling plan, by replacing the multiple paths corresponding to each target object that does not meet the target number requirement in the scheduling plan with any path combination corresponding to each target object, multiple intermediate scheduling plans can be obtained.

[0126] Optionally, the following formula can be used to constrain the scheduling plan from the perspective of the target number:

[0127]

[0128] In the above formula, represents the target number required by the target object and represents the number of robots executing the tasks of the target object .

[0129] Step 2: Determine the intermediate scheduling plan with the minimum sum of path scheduling time and execution time among the multiple intermediate scheduling plans as the target scheduling plan.

[0130] In this step, it is necessary to calculate the path scheduling time and execution time of each intermediate scheduling plan, and use the intermediate scheduling plan with the minimum sum of path scheduling time and execution time as the target scheduling plan.

[0131] Specifically, the target scheduling plan can be determined in the following manner of Steps 2.1 - 2.3:

[0132] Step 2.1: For each intermediate scheduling plan among the multiple intermediate scheduling plans, simulate the execution of the plan corresponding to each target object in the intermediate scheduling plan, use the path scheduling time of the first robot arriving at the target object as the path scheduling time of the plan, and use the time from the start of task execution at the target object to the end of task execution as the execution time of the plan;

[0133] It should be understood that each intermediate scheduling plan includes a plan corresponding to each target object; the plan corresponding to each target object includes the task path of the robot that executes the task of the target object; the path scheduling time of the robot refers to the path scheduling time when it executes the task of the corresponding target object. 。

[0134] Furthermore, according to the number of robots involved in the plan corresponding to the target object, different methods can be adopted to confirm the path scheduling time and execution time of each plan, specifically including the following two cases:

[0135] Case 1: If the number of robots involved in the plan is one, then the path scheduling time of this robot is used as the path scheduling time of the plan; and the directly predicted time obtained by prediction is used as the execution time of the plan.

[0136] Among them, the following method can be used to obtain the directly predicted working time: taking the specific scheduling requirements of the target object and the model of the robot as inputs, using the trained multi-layer perceptron prediction model for prediction, and outputting the time required for the robot to execute the task of the target object; taking the output of the multi-layer perceptron prediction model as the directly predicted time. Exemplarily, the specific scheduling requirements of the target object may include specific requirements related to operations, such as the type, size, operation steps, accuracy requirements, etc. of the equipment.

[0137] Case 2: If the number of robots involved in the scheduling plan exceeds one, then the execution time of the plan can be determined according to whether the task of the target object allows any number of robots to execute. Among them, whether the task of the target object allows any number of robots to execute can be determined according to the operation ability of the robots that execute the target object.

[0138] Case 2.1: If the number of robots involved in the scheduling plan exceeds one, and the task of the target object allows any number of robots to execute, then the path scheduling time of the first robot to reach the target object is used as the path scheduling time of the plan, and the time from when the first robot reaches the target object and starts to execute the task to the end of the task execution is used as the execution time of the plan. The execution time of this plan can be obtained by combining the trained LSTM prediction model and the multi-layer perceptron prediction model.

[0139] It should be understood that after the first robot reaches the target object in Case 2.1, it can directly start to execute the task without waiting for others to arrive.

[0140] For Case 2.1, a detailed introduction will be given below:

[0141] The execution time of the plan in Case 2.1 includes two parts: one is the time for partially executing the task; the other is the combined prediction time.

[0142] Among them, the time for part of the task execution is determined according to the time difference between the arrival time of the first robot at the target object and the arrival time of the last robot at the target object; after all the robots required by the target object arrive at the target object, the working time of the current robot combination predicted on the basis of the executed tasks is determined as the combined prediction time.

[0143] Specifically, the following method can be used to obtain the combined prediction working time: taking the models of multiple robots, the quantity of each model, the prediction results of the completed progress, and the specific scheduling requirements of the target object as inputs, and using the trained multi-layer perceptron prediction model to make predictions, and outputting the time required for all robots to jointly execute the tasks of the target object; taking the output of the multi-layer perceptron prediction model as the combined prediction time.

[0144] It should be understood that the trained multi-layer perceptron prediction models used in Case 2.1 and Case 1 are trained through different training sets, and they are two different models and cannot be used interchangeably.

[0145] The prediction results of the completed progress can be obtained through the following method: First, according to the path scheduling time consumption of each robot, arrange the task paths corresponding to the robots in the order of time consumption; taking the arrangement result, the task scheduling time consumption corresponding to each task path, the specific scheduling requirements of the target object, and the models of each robot as inputs, and using the trained Long Short Term Memory (LSTM) prediction model to make predictions, and the output result is the completion progress of the tasks of the target object during the time from the execution of the tasks by the first robot to the arrival of all robots at the target object; taking the output of the LSTM prediction model as the prediction results of the completed progress.

[0146] It is worth noting that arranging the task paths corresponding to the robots in the order of time consumption is to model the time series relationship between the first-arriving robot and the last-arriving robot, and is used to predict the task progress at each robot arrival stage. And LSTM is a neural network structure suitable for time series data, which can effectively capture the dynamic changes in the data in the sequence. Therefore, the LSTM model is very suitable for modeling the progress changes at each stage of the robot task execution process. It can predict the task execution situation at the next time step through the state of the previous time step (i.e., the completed task progress). The specific dynamic simulation process includes:

[0147] Input features: Input the path scheduling time consumption corresponding to each sorted task path, the specific scheduling requirements of the target object, the models of each robot, etc. into the LSTM model.

[0148] Sequential input: At each time step, the LSTM receives the task progress completed in the current stage and makes a dynamic prediction based on the task execution situation in the previous stage.

[0149] For example, after the first arriving robot executes part of the task, the LSTM can predict the task progress when the next robot arrives based on its progress. Then analyze the robots that arrive later, and on the basis of the current completed progress, the tasks that can be executed, while predicting the task progress when the next robot arrives. And so on, when the last robot arrives, the task progress at this time can be obtained.

[0150] In the method of case 2.1 provided in the embodiments of the present application, by using the LSTM prediction model, it is better to predict the task execution progress of each key stage during the process from partial execution to full completion, and obtain the task completion progress when all robots arrive, effectively improving the prediction accuracy of the execution time of the scheduling plan. In addition, this method allows the robots that arrive first to start executing tasks first without having to wait for all robots to arrive. This can significantly reduce the waiting time of the tasks, improve work efficiency, and maximize the parallelism of task execution and resource utilization.

[0151] Case 2.2: If the number of robots involved in the scheduling plan exceeds one, and the tasks of the target object do not allow any number of robots to execute (that is, the tasks of the target object only allow the target number of robots to execute, where the target number is the target number included in the scheduling requirements of the target object), then update the path scheduling time-consuming of the plan to the path scheduling time-consuming of the last robot arriving at the target object, and use the time from when all robots arrive at the target object to start executing tasks to the end of task execution as the execution time of the plan. The execution time of this scheduling plan can be predicted and combined by using the trained multi-layer perceptron prediction model.

[0152] It should be understood that in order to be able to process different task execution scenarios with a unified algorithm and avoid too much complex branch logic, the execution time of the plan in case 2.2 can also be divided into two parts (the time of partially executing tasks and the combined prediction time), and the time of partially executing tasks is determined to be 0; similarly, the combined prediction time can also be predicted and obtained by using the trained multi-layer perceptron prediction model adopted in case 2.1, where the prediction result of the completed progress input in the multi-layer perceptron prediction model is also 0.

[0153] In the method of case 2.2 provided in the embodiments of the present application, in the case where the tasks of the target object do not allow any number of robots to execute, update the path scheduling time-consuming of the plan to the path scheduling time-consuming of the last robot arriving at the target object to ensure the accuracy and reasonableness of the overall execution time of the tasks of the target object under this scheduling plan.

[0154] It should be noted that, for the methods in Case 1 and Case 2 provided in the embodiments of this application, by introducing two independent multi-layer perceptron prediction models and an LSTM prediction model, the time prediction problems of single-robot tasks and multi-robot tasks are flexibly addressed. The design aims to optimize resource allocation, simplify the algorithm structure, maximize task execution efficiency, and make the system more intelligent and efficient in multi-task and multi-robot collaborative scenarios through progress prediction and flexible handling of partial tasks.

[0155] Step 2.2: Take the sum of the path scheduling durations of the plans corresponding to each target object in the intermediate scheduling plan as the path scheduling duration of the intermediate scheduling plan, and take the sum of the execution times of the plans corresponding to each target object in the intermediate scheduling plan as the execution time of the intermediate scheduling plan.

[0156] Step 2.3: Determine the target scheduling plan with the minimum sum of the path scheduling duration and the execution time according to the path scheduling duration and the execution time corresponding to each intermediate scheduling plan.

[0157] In the above steps, for each intermediate scheduling plan, by determining the path scheduling duration and the execution time of the intermediate scheduling plan according to the path scheduling duration and the execution time of the plans corresponding to each target object included in the intermediate scheduling plan; then, among multiple intermediate scheduling plans, take the intermediate scheduling plan with the minimum sum of the path scheduling duration and the execution time as the target scheduling plan. It should be understood that the target scheduling plan determined by this method is the scheduling plan with the minimum comprehensive cost among multiple intermediate scheduling plans.

[0158] S105: Schedule the robot to execute the tasks of at least one target object according to the target scheduling plan.

[0159] In this step, after obtaining the target scheduling plan, schedule the robot to execute the tasks of at least one target object according to the target scheduling plan.

[0160] In a specific implementation manner, after scheduling the robot to execute the tasks of the target object according to the target scheduling plan, the working state of the robot will be updated; and according to the newly obtained scheduling requirements and the updated working state of each robot, a new target scheduling plan will be formulated for the tasks of the newly obtained target object.

[0161] In this implementation, when a new scheduling requirement is generated or the current task is completed, a new target scheduling plan can be recalculated based on the new scheduling requirement information and the pose information of the new robot, so as to quickly adapt to environmental changes. This dynamic adaptive ability can flexibly cope with the task volume and resource status, ensure the continuity and efficiency of the scheduling task, ensure that the scheduling of new tasks is not affected by the completed tasks, avoid conflicts and resource waste, and ensure the flexibility, robustness, global optimality and efficiency of task execution.

[0162] The scheduling planning method for the robot provided by the embodiment of the present application first obtains the pose information of multiple robots and the scheduling requirements of at least one target object that needs to interact with the robot, then performs path planning according to the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths. Then, according to the multiple task paths and the robots, multiple scheduling plans are obtained. Next, based on the types of tasks that each robot can execute, the types of tasks in each scheduling requirement, the task execution time, and the constraints of the target quantity of the robots executing the tasks, the multiple scheduling plans are simulated and scheduled to determine the target scheduling plan. Finally, the means of scheduling the robot to execute the tasks of at least one target object according to the target scheduling plan restricts the scheduling of the robot from multiple perspectives of task type, task execution time, and target quantity, ensuring the accuracy, reliability, and efficiency of the scheduling plan formulated in the multi-task and multi-robot scenario, achieving the effect of optimizing the scheduling plan for the robot, and further improving the task execution efficiency.

[0163] Embodiment 2 of the present application provides a scheduling method based on a robot, which conducts scientific demonstration through economic benefit calculation and simulation experiments to verify the beneficial effects of the present invention, as follows:

[0164] This experiment was carried out using multiple robots and different types of single-task scenarios, and mainly tested the effects of task allocation, path planning, combined prediction time, and partial task execution.

[0165] In the experiment, three types of target objects were first selected, including an assembly task scenario, an item handling task scenario, and a precision operation task scenario. Each type of task has different requirement contents, such as task type, task execution time limit, task complexity, and the position information of the target object. These tasks require different numbers of robots to cooperate to complete. Among them, 3 robots are required for the handling task, 2 robots are required for the assembly task, and 1 robot is required for the precision operation task.

[0166] By collecting the current position, attitude, and working status information of the robots, combining with the scheduling requirements of the target objects, and using the depth-first search algorithm, multiple task paths are generated for each robot to obtain multiple scheduling plans. Then, based on the types of tasks that each robot can execute, the types of tasks in each scheduling requirement, the task execution time, and the constraint of the number of target robots for task execution, the multiple scheduling plans are processed to obtain multiple intermediate scheduling plans. Next, by comparing the sum of the execution time and the path scheduling time consumption of each intermediate scheduling plan, the intermediate scheduling plan with the minimum sum of the execution time and the path scheduling time consumption is finally determined as the target scheduling plan. Finally, according to the target scheduling plan, the robots are scheduled to execute the tasks of at least one target object.

[0167] In the process of determining the target scheduling plan, different multi-layer perceptron prediction models will be used to predict the direct prediction time of a single robot and the combined prediction time of multiple robots based on the robot models, the specific scheduling requirements of the target objects, and the quantity of each model.

[0168] In this process, it is tested whether the tasks of the target objects allow any number of robots to execute tasks when the number of robots is greater than 1. If any number of robots are allowed to execute tasks, the path scheduling loss will be dynamically updated, and the prediction of the task completion time will be adjusted according to the actual execution situation. Among them, the execution time of some tasks is predicted by the trained LSTM prediction model: after arranging the paths corresponding to each robot in the plan according to the path scheduling time consumption of the robots, the LSTM prediction model uses this data to predict the task progress from the start of the task execution by the first robot to the arrival of all robots at the task scene.

[0169] To verify the advantages of the present invention in task scheduling, a comparative experiment with the traditional scheduling method is designed. The traditional scheduling method relies on static task allocation and scheduling algorithms, requiring all robots to reach the target position and cooperate to complete the tasks before the start of the tasks. The recorded data is shown in Table 1.

[0170] Table 1 Data Record Table

[0171]

[0172] It can be seen from the experimental data that the scheduling strategy of the present invention is significantly superior to the traditional method in terms of task execution time, flexibility, and resource utilization.

[0173] The scheduling time of the present invention is significantly better than the traditional method, especially in the scenario of multi-robot collaborative tasks. For example, in the assembly task scenario, the scheduling times of the traditional method for Robot A1 and Robot A2 are 35 minutes and 34 minutes respectively, while the scheduling times of the present invention are reduced to 28 minutes and 27 minutes, and the optimization ratio reaches about 20%.

[0174] Similarly, in the handling task scenario, the scheduling times of traditional methods are 50 minutes (robot B1) and 48 minutes (robot B2), while the method provided by the present invention reduces the times to 40 minutes and 39 minutes respectively through intelligent scheduling, improving the efficiency by 18% - 20%. This shows that the present invention can effectively reduce the waiting time in multi-robot task execution.

[0175] Traditional methods require all robots to reach the target position synchronously before jointly executing tasks, resulting in a large amount of waiting time. While the present invention, through the innovative design of partial task execution (i.e., any number of robots are allowed to execute the task), allows the robots that arrive first to start the task in advance, greatly reducing the total task time. For example, in the precision operation task scenario, under the traditional method, the scheduling times of robot C1 and robot C2 are 25 minutes and 24 minutes respectively, while the present invention reduces the scheduling times to 20 minutes and 19 minutes respectively by allowing the robots that arrive first to execute the task, with the efficiency improvement reaching about 20%.

[0176] This flexibility of partial task execution, especially in complex multi-robot tasks, can significantly improve the scheduling efficiency and avoid the delay problems caused by synchronous execution common in traditional methods.

[0177] The present invention can optimize the scheduling strategy in real time according to the task execution situation and task requirements of the robots through dynamic task allocation and time prediction models. In contrast, the scheduling of traditional methods is based on static allocation and lacks flexibility and the ability of real-time adjustment. The data in the table clearly shows this. For example, in the handling task scenario, the scheduling times of robot B1 and robot B2 are reduced by 10 minutes and 9 minutes respectively in the present invention, which proves the effectiveness of the dynamic scheduling strategy.

[0178] The multi-layer perceptron prediction model and LSTM prediction model used in the present invention can accurately predict the completion time of tasks and continuously adjust during the task execution process, further reducing the scheduling time.

[0179] The precision operation task scenario in the table has highly complex task requirements. Traditional methods have longer scheduling times when dealing with such complex tasks and cannot fully utilize robot resources. While the present invention can significantly optimize the scheduling time of complex tasks through task complexity modeling and intelligent allocation. In the scenarios of precision operation task C1 and precision operation task C2, the scheduling times of the present invention are reduced by 20% and 20.83% respectively compared with traditional methods, showing the excellent ability of the present invention in dealing with complex tasks.

[0180] In summary, the innovation of the present invention in multi-robot collaborative task scheduling is reflected in the improvement of the accuracy, efficiency, dynamism and flexibility of the scheduling plan. By constraining the scheduling of robots from multiple angles such as task type, task execution time and target number, combined with advanced algorithms such as multi-layer perceptron and LSTM, the scheduling plan of robots is optimized, thereby improving the efficiency of task execution, providing a robot-based scheduling method that is significantly superior to traditional methods.

[0181] Figure 2 This is a schematic diagram of the structure of the robot-based scheduling device provided in Example 3 of the present application, as shown in FIG. Figure 2 As shown, the robot-based scheduling device 20 provided in this embodiment includes:

[0182] An acquisition unit 201 is configured to acquire position information of multiple robots and a scheduling requirement of at least one target object that needs to interact with the robots, where each scheduling requirement includes the position information of the target object, the task type, the task execution time, and the target number of robots that execute the task;

[0183] A first processing unit 202 is configured to perform path planning based on the posture information of the multiple robots and the position information of the at least one target object to obtain multiple task paths;

[0184] A second processing unit 203 is configured to obtain a plurality of scheduling plans based on the plurality of task paths and the plurality of robots, each scheduling plan including a task path for each robot;

[0185] A determination unit 204 is configured to simulate scheduling of the plurality of scheduling plans based on the type of tasks that can be performed by each robot, the type of task in each scheduling requirement, the task execution time, and the target number of robots that can perform the task, and determine a target scheduling plan;

[0186] The execution unit 205 is configured to schedule the robot to execute the task of the at least one target object according to the target scheduling plan.

[0187] The robot-based scheduling device 20 provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effects are similar, and are not described in detail in this embodiment.

[0188] Figure 3 This is a schematic diagram of the structure of the robot-based scheduling device provided in Example 4 of the present application, as shown in FIG. Figure 3 As shown, based on the above embodiment, the robot-based scheduling device 20 provided in this embodiment further includes:

[0189] A third processing unit 206, configured to use the time from when the first robot arrives at the target object to start executing the task to the end of the task execution as the execution time of the plan if the task of the target object allows any number of robots to execute;

[0190] If the task of the target object does not allow any number of robots to execute, use the time from when all robots arrive at the target object to start executing the task to the end of the task execution as the execution time of the plan.

[0191] A prediction unit 207, configured to, when the task of the target object allows some robots to execute, obtain the execution time through combined prediction using a trained LSTM prediction model and a multi-layer perceptron prediction model;

[0192] Or,

[0193] When the task of the target object does not allow some robots to execute, the execution time is obtained through prediction using a trained multi-layer perceptron prediction model.

[0194] In a possible implementation manner, the determination unit 204 includes:

[0195] A first determination module, configured to perform simulated scheduling on the multiple scheduling plans, and process the multiple scheduling plans based on the types of tasks that each robot can execute, the types of tasks in each scheduling requirement, the task execution time, and the constraint of the target number of robots for executing the tasks, to obtain multiple intermediate scheduling plans; the intermediate scheduling plans are the scheduling plans that meet the constraints among the multiple scheduling plans;

[0196] A second determination module, configured to determine, from the multiple intermediate scheduling plans, the intermediate scheduling plan with the minimum sum of the path scheduling time and the execution time as the target scheduling plan.

[0197] In a possible implementation manner, the second determination module is specifically configured to:

[0198] For each intermediate scheduling plan among the multiple intermediate scheduling plans, perform simulated execution on the plan corresponding to each target object in the intermediate scheduling plan, use the path scheduling time of the first robot that arrives at the target object as the path scheduling time of the plan, and use the time from when the task starts to be executed at the target object to the end of the task execution as the execution time of the plan; each intermediate scheduling plan includes the plan corresponding to each target object; the plan corresponding to the target object includes the task path of the robot that executes the task of the target object.

[0199] Take the sum of the path scheduling times of the plans corresponding to each target object in the intermediate scheduling plan as the path scheduling time of the intermediate scheduling plan, and take the sum of the execution times of the plans corresponding to each target object in the intermediate scheduling plan as the execution time of the intermediate scheduling plan;

[0200] Determine the target scheduling plan with the smallest sum of the path scheduling time and the execution time according to the path scheduling time and the execution time corresponding to each intermediate scheduling plan.

[0201] In a possible implementation manner, the first processing unit 202 is specifically configured to:

[0202] Perform feasible path planning by using a depth-first search algorithm according to the position information of the multiple robots and the position information of the at least one target object to obtain the multiple task paths.

[0203] In a possible implementation manner, the obtaining unit 201 is partially configured to:

[0204] Obtain the working status of all currently schedulable robots, and the working status of each robot includes a working status or an idle status;

[0205] Obtain the pose information of the multiple robots in the idle state.

[0206] In a possible implementation manner, the first determining module is partially configured to:

[0207] For the multiple paths in each scheduling plan, delete the paths where the type of task executable by the robot does not match the type of task corresponding to the path according to the type of task executable by each robot and the type of task in each scheduling requirement;

[0208] For each scheduling plan, if there is any path scheduling time for a robot to execute the task of the target object corresponding to the scheduling requirement that exceeds the task execution time in the scheduling requirement, then delete the scheduling plan;

[0209] For each remaining scheduling plan, if the number of robots executing the task of the target object corresponding to any scheduling requirement is less than the target number in the scheduling requirement, then delete the scheduling plan.

[0210] The scheduling device 20 based on robots provided in this embodiment can execute the method provided in the above method embodiment, and its implementation principle and technical effect are similar, which will not be elaborated here in this embodiment.

[0211] Figure 4 It is a schematic structural diagram of a computer device provided for this application. As Figure 4As shown in the figure, the computer device 30 provided in this embodiment includes: at least one processor 301 and a memory 302. Optionally, the device 30 further includes a communication component 303. Among them, the processor 301, the memory 302, and the communication component 303 are connected through a bus 304.

[0212] In a specific implementation process, at least one processor 301 executes the computer-executable instructions stored in the memory 302, so that at least one processor 301 executes the above-mentioned method.

[0213] For the specific implementation process of the processor 301, reference can be made to the above method embodiment. Their implementation principles and technical effects are similar, and will not be elaborated here in this embodiment.

[0214] This application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When the processor executes the computer-executable instructions, the above-mentioned method is implemented.

[0215] This application also provides a computer program product, including a computer program, which implements the above-mentioned method when executed by the processor.

[0216] Finally, it should be noted that: those skilled in the art will easily think of other implementation schemes of the present invention after considering the specification and practicing the invention disclosed herein. The present invention aims to cover any variations, uses, or adaptive changes of the present invention. These variations, uses, or adaptive changes follow the general principles of the present invention and include common general knowledge or conventional technical means in the technical field not disclosed in the present invention. It is not limited to the exact structure described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A robot-based scheduling method, characterized in that, Including: Obtain the pose information of multiple robots and the scheduling requirements of at least one target object to be interacted with the robots. Each scheduling requirement includes the position information of the target object, the task type, the task execution time, and the target number of robots for executing the task; Perform path planning based on the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths; Obtain multiple scheduling plans according to the multiple task paths and the multiple robots. Each scheduling plan includes a task path of each robot; Based on the type of tasks that each robot can execute, the task type in each scheduling requirement, the task execution time, and the constraint of the target number of robots for executing the task, perform simulated scheduling on the multiple scheduling plans to determine the target scheduling plan; Schedule the robots to execute the tasks of the at least one target object according to the target scheduling plan.

2. The method according to claim 1, characterized in that, The performing simulated scheduling on the multiple scheduling plans based on the type of tasks that each robot can execute, the task type in each scheduling requirement, the task execution time, and the constraint of the target number of robots for executing the task to determine the target scheduling plan includes: Perform simulated scheduling on the multiple scheduling plans, and process the multiple scheduling plans based on the type of tasks that each robot can execute, the task type in each scheduling requirement, the task execution time, and the constraint of the target number of robots for executing the task to obtain multiple intermediate scheduling plans; the intermediate scheduling plan is the scheduling plan that satisfies the constraint among the multiple scheduling plans; Determine the intermediate scheduling plan with the minimum sum of the path scheduling time and the execution time from the multiple intermediate scheduling plans as the target scheduling plan.

3. The method according to claim 2, wherein The determining the intermediate scheduling plan with the minimum sum of the path scheduling time and the execution time from the multiple intermediate scheduling plans as the target scheduling plan includes: For each intermediate scheduling plan among the multiple intermediate scheduling plans, respectively perform simulated execution on the plans corresponding to each target object in the intermediate scheduling plan, take the path scheduling time of the robot that first arrives at the target object as the path scheduling time of the plan, and take the time from the start of executing the task at the target object to the end of the task execution as the execution time of the plan; the intermediate scheduling plan includes the plans corresponding to each target object; the plan corresponding to the target object includes the task path of the robot for executing the task of the target object; Take the sum of the path scheduling times of the plans corresponding to each target object in the intermediate scheduling plan as the path scheduling time of the intermediate scheduling plan, and take the sum of the execution times of the plans corresponding to each target object in the intermediate scheduling plan as the execution time of the intermediate scheduling plan; Determine the target scheduling plan with the minimum sum of the path scheduling time and the execution time according to the path scheduling time and the execution time corresponding to each intermediate scheduling plan.

4. The method according to claim 3, wherein During the process of simulating the execution of the plan corresponding to each target object in the intermediate scheduling plan, if the number of robots involved in the plan exceeds one, the method further includes: If the task of the target object allows any number of robots to execute, the execution time of the plan is the time from when the first robot arrives at the target object to start executing the task to the end of the task execution; If the task of the target object does not allow any number of robots to execute, the execution time of the plan is the time from when all robots arrive at the target object to start executing the task to the end of the task execution.

5. The method according to claim 4, characterized in that, The method further includes: In the case where the task of the target object allows some robots to execute, the execution time is predicted by combining a trained LSTM prediction model and a multi-layer perceptron prediction model; Or, In the case where the task of the target object does not allow some robots to execute, the execution time is predicted by a trained multi-layer perceptron prediction model.

6. The method according to any one of claims 1 to 5, characterized in that, The pose information of each robot includes the position information of the robot. The path planning based on the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths includes: According to the position information of the multiple robots and the position information of the at least one target object, a depth-first search algorithm is used for feasible path planning to obtain the multiple task paths.

7. The method according to any one of claims 1 to 5, characterized in that, The obtaining the pose information of the multiple robots includes: Obtaining the working status of all currently schedulable robots, and the working status of each robot includes a working status or an idle status; Obtaining the pose information of the multiple robots in the idle state.

8. The method according to any one of claims 2 to 5, characterized in that The simulating and scheduling the multiple scheduling plans, and processing the multiple scheduling plans based on the type of tasks that each robot can execute, the type of tasks in each scheduling requirement, the task execution time, and the constraint of the target number of robots executing the tasks includes: For multiple paths in each scheduling plan, according to the type of tasks that each robot can execute and the type of tasks in each scheduling requirement, delete the paths where the type of tasks that the robot can execute does not match the type of tasks corresponding to the path; For each scheduling plan, if there is any path scheduling time-consuming for a robot to execute the task of the target object corresponding to the scheduling requirement that exceeds the task execution time in the scheduling requirement, delete the scheduling plan; For each remaining scheduling plan, if the number of robots executing the task of the target object corresponding to any scheduling requirement is less than the target number in the scheduling requirement, delete the scheduling plan.

9. A robot-based scheduling device, characterized in that, Includes: An obtaining unit, configured to obtain the pose information of the multiple robots and the scheduling requirements of at least one target object that needs to interact with the robots, and each scheduling requirement includes the position information of the target object, the type of task, the task execution time, and the target number of robots executing the task; A first processing unit for performing path planning based on the pose information of the multiple robots and the position information of the at least one target object to obtain multiple task paths; A second processing unit for obtaining multiple scheduling plans according to the multiple task paths and the multiple robots, each scheduling plan including one task path for each robot; A determination unit for performing simulated scheduling on the multiple scheduling plans based on the types of tasks that each robot can execute, the task types in each scheduling requirement, the task execution time, and the constraint of the target number of robots executing the tasks, and determining a target scheduling plan; A processing unit for scheduling the robots to execute the tasks of the at least one target object according to the target scheduling plan.

10. A computer device, characterized in that, Comprising: A memory and a processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor executes the method according to any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, Computer-executable instructions are stored in the computer-readable storage medium, and when the computer-executable instructions are executed by the processor, they are used to implement the method according to any one of claims 1-8.

Images