Mobile transport robot transportation task rolling assignment method based on forward-looking pre-positioning

Abstract

The application discloses a mobile transport robot transportation task rolling allocation method based on forward pre-positioning, relates to the mobile transport robot task scheduling field, and aims at the problems of space-time mismatch between current transportation demand and candidate mobile transport robots, idle of part of the mobile transport robots and slow response to short-term future demand in a dynamic multi-site operation scene, acquires a current system state, respectively determines unloading allocation results of the load mobile transport robots, pick-up matching results of current positive demand tasks and forward pre-positioning allocation results of unmatching mobile transport robots, updates a to-be-executed transportation intention sequence of each mobile transport robot based on the allocation results and historical uncompleted tasks, and realizes rolling allocation and dynamic adjustment of the transportation tasks, so that the empty running level is reduced, and the utilization rate and transportation response efficiency of the mobile transport robots are improved.

Description

Technical Field

[0001] This invention relates to the field of mobile handling robot task scheduling technology, and in particular to a method for rolling allocation of mobile handling robot transportation tasks based on forward-looking pre-positioning. Background Technology

[0002] In intelligent manufacturing, warehousing and logistics, sorting and distribution, and other dynamic multi-site operation scenarios, objects to be transferred often need to be transferred multiple times between multiple service stations. As tasks arrive dynamically, system status changes in real time, and the degree of resource collaboration and coupling increases, the task allocation method of mobile handling robots in these scenarios directly affects the overall logistics efficiency, resource utilization level, and system responsiveness.

[0003] Flexible workshops are a typical form of the aforementioned dynamic multi-site operation scenario. In flexible workshops, workpieces typically need to be transferred multiple times between processing equipment, input buffers, output buffers, and various service stations. With the application of multi-variety, small-batch production models, internal transportation tasks in the workshop exhibit characteristics such as dynamic arrival, real-time status changes, and tight resource coupling. Automated guided vehicles (AGVs), as typical mobile handling robots in flexible workshops, directly impact logistics efficiency and equipment utilization through their task allocation methods.

[0004] In existing technologies, several solutions have been researched to address the scheduling problem of flexible job shops integrating multiple AGVs. For example, Chinese patent document CN113671910A discloses a method, device, and medium for scheduling flexible job shops integrating multiple AGVs, which obtains a scheduling scheme by jointly scheduling machines, workpieces, and multiple AGVs. Heger and Voss proposed a dynamic priority-based AGV dispatching method in their paper "Dynamic priority based dispatching of AGVs in flexible job shops" to analyze the scheduling and dispatching problem of flexible job shops that includes AGV transportation time. Lin et al. established a network model and provided a corresponding solution method for the AGV dispatching problem in manufacturing systems in their paper "Network model and effective evolutionary approach for AGV dispatching in manufacturing system". The above solutions can realize the allocation between mobile handling robots and transportation tasks to a certain extent, but they mostly focus on the currently explicit transportation needs or take the overall scheduling scheme as the optimization goal, lacking a coordinated consideration of the utilization of mismatched robots and potential transportation needs.

[0005] In addition, some existing technologies attempt to pre-assign vehicles by predicting equipment completion times. For example, Chinese patent document CN114613710A discloses a method, system, and computer device for pre-assignment of vehicles based on completion time. This method predicts machine completion times and dispatches vehicles to machines nearing completion in advance, thereby reducing material waiting time. This type of solution reflects the idea of ​​pre-assigning vehicles to meet future needs, but its application scenarios are mainly geared towards specific material handling systems. It has not yet addressed the unified rolling allocation problem between unloading decisions for cargo robots, current demand task allocation, and forward-looking pre-positioning of unmatched robots in dynamic multi-site operation scenarios.

[0006] Therefore, the existing technology still has the following shortcomings: First, when allocating tasks only around the current demand, some candidate mobile handling robots may be left idle in the current decision round without being assigned any tasks, increasing empty runs and waiting; Second, it does not make full use of potential tasks that have not yet formed actual transportation demand at the current decision time but are expected to be transformed into transportation demand in a short period of time; Third, existing methods usually lack a mechanism to connect and update the current round allocation results with historical unfinished tasks, making it difficult to support rolling decision-making and continuous execution in dynamic environments.

[0007] Therefore, there is an urgent need for a rolling allocation method for mobile handling robot transportation tasks based on forward-looking pre-positioning. This method can simultaneously complete the selection of unloading targets for loaded mobile handling robots, matching of current demand tasks, and forward-looking pre-positioning of unmatched mobile handling robots. It can also update the sequence of pending transportation intentions for each mobile handling robot by combining historical unfinished tasks, so as to improve the utilization rate of mobile handling robots and enhance the responsiveness to future transportation needs. Summary of the Invention

[0008] To address the above issues, this invention proposes a rolling allocation method for mobile transport robots based on forward-looking pre-positioning. This invention obtains the allocation result of mobile transport robots by simultaneously considering the selection of unloading targets for loaded mobile transport robots, matching of current demand tasks, and forward-looking pre-positioning of unmatched mobile transport robots. Based on the allocation result and historical unfinished tasks, the sequence of transport intentions to be executed for each mobile transport robot is updated, and the rolling allocation and dynamic adjustment of transport tasks are carried out to achieve coordinated allocation of current and potential demands, reduce empty runs, and improve the overall transport efficiency of the system.

[0009] A method for rolling allocation of transportation tasks for mobile handling robots based on forward-looking pre-positioning is applied to a dynamic multi-site operation system comprising several service stations and several mobile handling robots, including the following rolling execution steps:

[0010] S1, obtain the status information of the dynamic multi-site operation system at the current decision moment, and obtain the current set of candidate mobile handling robots participating in transportation allocation, the current set of transportation tasks in demand, and the set of potential transportation tasks; the candidate set of mobile handling robots includes the set of loaded mobile handling robots and the set of unloaded mobile handling robots.

[0011] S2, using the unloading comprehensive cost function to match the set of cargo machines with the information of optional unloading sites, to obtain the unloading allocation result at the current decision moment;

[0012] S3, perform task capacity constraint matching on the current demand transportation task set and the candidate mobile handling robot set to obtain the current demand task matching result at the current decision time; and take the unmatched mobile handling robots in the candidate mobile handling robot set as unmatched mobile handling robots.

[0013] S4, match the unmatched mobile transport robots with the set of potential transport tasks, and obtain the prospective pre-positioning allocation result of the unmatched mobile transport robots at the current decision time;

[0014] S5. Based on the unloading allocation results, the current demand task matching results, and the forward-looking pre-positioning allocation results, the transportation allocation results for the current decision moment are formed.

[0015] S6, based on the unfinished tasks left over from the previous decision time and the transportation allocation results at the current decision time, update the sequence of transportation intentions to be executed for each mobile handling robot; the mobile handling robot executes the task according to the sequence of transportation intentions to be executed.

[0016] Preferably, the unloading allocation result at the current decision moment is obtained by matching the set of cargo machines with the information of the optional unloading sites using the unloading comprehensive cost function, as follows.

[0017] Construct a set of candidate unloading sites for each mobile handling robot in the cargo handling robot set;

[0018] For each candidate unloading site in the candidate unloading site set, construct the comprehensive unloading cost function, expressed as:

[0019] ;

[0020] in, This represents the overall cost function for unloading. Indicates candidate uninstallation sites The corresponding processing equipment input buffer area and the moving and transporting machine Waiting load for workpieces of the same type and stage currently under load; This indicates the pending load for other types of workpieces or workpieces at other stages; Indicates candidate uninstallation sites The remaining processing time for the workpiece currently being processed by the corresponding equipment; Indicates mobile transport robot Move from current location to candidate uninstallation site The estimated cost of driving; and Preset weighting coefficients;

[0021] Each mobile handling robot in the cargo handling robot set selects the candidate unloading station with the smallest value of the unloading comprehensive cost function to obtain the unloading allocation result at the current decision time.

[0022] Preferably, the task capacity constraint is specifically implemented as follows:

[0023] Downstream availability is constructed and expressed as:

[0024] ;

[0025] in, Indicates downstream availability; Indicates task Minimum expected waiting cost to proceed to optional downstream equipment after the completion of the current transport;

[0026] Construct tasks based on downstream availability The overall demand value is expressed as:

[0027] ;

[0028] in, This represents the overall demand value; , and Indicates the preset weighting coefficient; Indicates task The queue pressure value; Indicates the final process quantity;

[0029] Based on task Determining the overall demand value The task capacity constraint is expressed as:

[0030] ;

[0031] in, Indicates task capacity constraints; This represents the normalized overall demand value; This indicates the total number of current candidate mobile handling robots; Indicates the current number of tasks that are actually waiting; This indicates taking the minimum value; This indicates taking the maximum value.

[0032] Preferably, the task The queue pressure value is expressed as:

[0033] ;

[0034] in, Indicates task The normalized actual number of waiting items at the current decision-making moment; This represents the age stress value of the task after normalization; This represents the future impact of the task after normalization;

[0035] The future impact of the task is represented as follows:

[0036] ;

[0037] in, This indicates that the current system can subsequently flow to tasks. The set of upstream related workpieces for the corresponding site; Indicates the process stage index; Indicates workpiece Stage of the process; Indicates the quantity of the corresponding workpiece; Indicates workpiece Arrive at the mission The estimated remaining time at the corresponding location. This is the preset attenuation coefficient.

[0038] Preferably, the picking task matching under the task capacity constraint is based on the task-side preference relationship and the mobile handling robot-side preference relationship.

[0039] The task-side preference relationship is as follows: Task Candidate mobile handling robots are sorted from smallest to largest according to their estimated arrival time to form a task-side preference list; when multiple mobile handling robots have the same estimated arrival time, the currently idle mobile handling robot is selected first.

[0040] The preferred relationship of the mobile transport robot is as follows: for candidate tasks near the mobile transport robot, priority is given to candidate tasks with larger overall demand values; for candidate tasks that are farther away, priority is given to candidate tasks with lower travel costs after the mobile transport robot is released to reach the task station.

[0041] Preferably, the estimated arrival time is expressed as:

[0042] ;

[0043] in, Indicates mobile transport robot Arrive at the mission The estimated arrival time; Indicates mobile transport robot The estimated release time for the current pickup task can begin. This indicates the release station corresponding to the release time; Indicates the distance between stations; Indicates the cost per unit of travel time; Indicates task The target site location.

[0044] Preferably, in step S4, the unmatched mobile handling robots are matched with the set of potential transportation tasks to obtain the forward-looking pre-positioning allocation result of the unmatched mobile handling robots at the current decision time, as follows:

[0045] Perform a validity check on the look-ahead pre-positioning intentions left in the sequence of transport intentions to be executed by the mobile transport robot at the previous decision moment;

[0046] Calculate the estimated arrival time of the unmatched mobile transport robot to the site corresponding to the potential task; then calculate the time difference between the mobile transport robot and the potential task by dividing the estimated arrival time by the estimated readiness time.

[0047] For unmatched mobile transport robots, construct a feasible set of prospective pre-positioning tasks; represented as:

[0048] ;

[0049] in, This represents a set of feasible forward-looking pre-positioning tasks; Represents a set of potential transportation tasks; Indicates the actual number of people waiting; Indicates the estimated readiness time; Indicates the early tolerance threshold; Indicates mobile transport robot With potential tasks The time difference between them; Indicates the hysteresis tolerance threshold;

[0050] Under a single capacity constraint with a task capacity constraint value of 1, bilateral matching is performed between the unmatched mobile transport robot and the set of feasible forward-looking pre-positioning tasks to obtain the forward-looking pre-positioning allocation results.

[0051] Preferably, the validity check of the forward-looking pre-positioning intentions left over from the previous decision-making moment specifically includes:

[0052] If a potential task corresponding to a certain forward positioning intention has not yet been transformed into a current active demand task, and its estimated readiness time is still valid, and the time difference between the corresponding mobile transport robot and the potential task is still within the preset feasible window, then the forward positioning intention is retained; otherwise, the forward positioning intention is cleared. If the same potential task is held by multiple mobile transport robots at the same time, only the one with the earliest estimated arrival time and priority to the empty mobile transport robot is retained, and the other duplicate forward positioning intentions are cleared.

[0053] Preferably, under a single capacity constraint with a task capacity constraint value of 1, a two-sided matching is performed between the unmatched mobile transport robot and the set of feasible forward-looking pre-positioned tasks, and the matching is based on the potential task-side preference relationship and the mobile transport robot-side preference relationship;

[0054] The potential task-side preference relationship is as follows: each candidate mobile handling robot is sorted from smallest to largest according to its expected arrival time; when the expected arrival times are the same, the unloaded mobile handling robot is selected first.

[0055] The preferred relationship of the mobile handling robot is as follows: sorting the potential tasks by the absolute value of the time difference between the mobile handling robot and the potential task from smallest to largest; when the absolute values ​​of the time differences of multiple potential tasks are the same, the task with the smaller expected arrival time is selected first; in the case of further equivalence, the potential task with the higher demand evaluation value is selected first.

[0056] Preferably, the process of updating the sequence of unfinished tasks left over from the previous decision time and the transportation allocation result at the current decision time for each mobile handling robot satisfies the following constraints: the length of the sequence of unfinished transportation intentions for an empty mobile handling robot does not exceed 1; the length of the sequence of unfinished transportation intentions for a loaded mobile handling robot is 1 or 2, wherein the first intention is an unloading intention, and the second intention is a picking intention or a forward pre-positioning intention; forward pre-positioning allocation is only executed for mobile handling robots that are not matched at the current decision time.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] (1) By simultaneously considering the selection of unloading targets for the cargo-carrying mobile handling robot, the matching of current demand tasks, and the forward-looking pre-positioning of unmatched mobile handling robots, this invention achieves coordinated allocation of current and potential demands, reduces empty runs, and improves the overall transportation efficiency of the system. Without increasing additional transportation resources, it is beneficial to improve the utilization rate of mobile handling robots and the efficiency of transportation task allocation, and enhances the system's adaptability to future changes in transportation demand.

[0059] (2) By introducing time feasibility constraints on potential transportation tasks, this invention enables unmatched mobile handling robots to move to future demand locations in advance, thereby improving the responsiveness to upcoming transportation tasks and reducing waiting time.

[0060] (3) By connecting and updating the current decision results with the historical unfinished tasks, a rolling updated sequence of transportation intentions is formed, enabling the system to achieve continuous decision-making and stable execution in a dynamic environment. Attached Figure Description

[0061] The present invention will now be described in further detail with reference to the accompanying drawings;

[0062] Figure 1 This is a schematic diagram of a flexible workshop transportation scenario structure based on the mobile handling robot transportation task rolling allocation method based on forward-looking pre-positioning, according to an embodiment of the present invention.

[0063] Figure 2 This is a flowchart illustrating the steps of the mobile handling robot's rolling task allocation method based on forward-looking pre-positioning, according to an embodiment of the present invention.

[0064] Figure 3 This is a flowchart illustrating the rolling allocation method for transportation tasks of a mobile handling robot based on forward-looking pre-positioning, according to an embodiment of the present invention.

[0065] Figure 4 This is a schematic diagram of the unloading target determination process of a cargo AGV using a forward-looking pre-positioning mobile handling robot transportation task rolling allocation method according to an embodiment of the present invention.

[0066] Figure 5 This is a schematic diagram of the combined allocation process of positive demand task matching and forward pre-positioning in the mobile handling robot transportation task rolling allocation method based on forward pre-positioning according to an embodiment of the present invention.

[0067] Figure 6 This is a schematic diagram of the transportation intention sequence update process of the mobile handling robot transportation task rolling allocation method based on look-ahead pre-positioning in an embodiment of the present invention. Detailed Implementation

[0068] The present invention will be further described below through specific embodiments.

[0069] This method is applicable to the rolling allocation of transportation tasks in dynamic multi-site operation scenarios. Dynamic multi-site operation scenarios include flexible workshops, intelligent warehousing, logistics sorting, and production line material distribution scenarios; mobile handling robots include automated guided vehicles (AGVs), autonomous mobile robots (AMRs), and other autonomous mobile devices capable of performing object transfer tasks between multiple service sites.

[0070] The following uses an AGV transportation system in a flexible workshop as an example to illustrate this method. In this embodiment, the workpiece can be regarded as the object to be transferred, and the processing equipment, input buffer area, output buffer area, and various loading and unloading handover positions can be uniformly regarded as service stations. AGV is a specific implementation of a mobile handling robot. Figure 1 The flexible workshop transportation scenario structure used in this embodiment is shown.

[0071] In this embodiment, the flexible workshop includes multiple processing devices, an input buffer area, an output buffer area, several service stations, and multiple AGVs. Each service station is used to perform loading, unloading, or workpiece transfer operations, and each AGV can move within the workshop's internal road network and complete workpiece transportation tasks between different service stations.

[0072] In this embodiment, the set consisting of all AGVs is denoted as . Record the first The system state at each decision moment is:

[0073] ;

[0074] in, This represents the set of status information related to the current workshop and AGV transportation allocation. Indicates the current system time. A set of status information. It includes at least a set of empty AGVs, a set of loaded AGVs, a set of candidate AGVs, a set of currently demanded transportation tasks, and a set of potential transportation tasks. It may also include information such as workpiece processing status, input buffer status, service station status, and the current location status of AGVs.

[0075] It should be noted that, in this embodiment, the empty AGV set, the cargo AGV set, and the candidate AGV set correspond to the empty mobile handling robot set, the cargo mobile handling robot set, and the candidate mobile handling robot set, respectively; the workpiece transportation task corresponds to the transportation task; and the service station corresponds to various object transfer nodes in the dynamic multi-site operation scenario.

[0076] Record No. The set of empty AGVs, the set of loaded AGVs, and the set of candidate AGVs at each decision moment are respectively , , ,in, This represents the set of candidate AGVs that can currently participate in pickup task allocation or forward deployment allocation. Let the set of currently demanding transportation tasks be denoted as . Let the set of potential transportation tasks be denoted as .

[0077] In this embodiment, a current demand transportation task refers to a task that already has actual transportation demand at the current decision-making time and can be immediately initiated for pickup and allocation. A potential transportation task refers to a task that has not yet formed actual pickup demand at the current decision-making time, but whose transportation demand can be inferred to form within the predicted time window based on the current status of the workpiece, the process route, and the estimated completion time. For potential task candidates... If its expected readiness time satisfy Then include it in the potential transportation task set. .in, The preset prediction time window length.

[0078] Potential transportation tasks may include tasks for workpieces that are being unloaded at a service station, are waiting for processing in the input buffer, are being processed on equipment, or are in other situations where subsequent transportation demand can be inferred. When the dynamic multi-site operation scenario is a flexible workshop, potential transportation tasks include workpiece transfer tasks that are in the processing stage, the waiting stage in the input buffer, the loading and unloading handover stage, or the workpiece transfer tasks that have not yet formed a current positive demand but are expected to form a transportation demand within a preset time range.

[0079] The method in this embodiment is executed once at each decision point, and its overall flow is as follows: Figure 2 and Figure 3 As shown, it includes S1 to S6.

[0080] S1: Obtain the status information of the dynamic multi-site operation system at the current decision moment, and obtain the set of candidate mobile handling robots currently participating in transportation allocation, the set of current demand transportation tasks, and the set of potential transportation tasks.

[0081] First, obtain the system state at the current decision-making moment. and from the state information set The system identifies AGVs that are currently in an idle state and forms a set. Identify AGVs currently in a loaded state to form a set. And determine the set of candidate AGVs currently participating in transportation allocation. Simultaneously, extract the set of currently demanding transportation tasks from the current system state. and potential transportation task set .

[0082] In this embodiment, the system status information also includes the workpiece processing status, input buffer status, service station status, and AGV current position status.

[0083] S2, using the unloading comprehensive cost function to match the set of cargo machines with the information of optional unloading sites, obtains the unloading allocation result at the current decision moment.

[0084] For each cargo-carrying AGV, determine the corresponding unloading target. For any cargo-carrying AGV... First, based on the type of workpiece corresponding to its current load, the current process stage, the range of subsequent processing equipment, and the acceptability of the service station, a set of candidate unloading stations for the AGV is constructed. .

[0085] For any candidate uninstallation site Construct the unloading comprehensive cost function, expressed as:

[0086] ;

[0087] in, Indicates candidate uninstallation sites The corresponding processing equipment input buffer area and AGV The waiting load for workpieces of the same type and stage under the current load. The waiting load for other types or stages of workpieces can be obtained by statistically analyzing the processing time and quantity of each waiting workpiece in the input buffer area of ​​the corresponding device at the candidate site. Indicates candidate uninstallation sites The remaining processing time for the workpiece currently being processed by the corresponding equipment; Indicates AGV Move from current location to candidate uninstallation site The estimated cost of driving; and These are the preset weighting coefficients.

[0088] In this embodiment, the expected driving cost can be expressed as:

[0089] ;

[0090] in, Indicates AGV With candidate uninstall sites Distance between stations This represents the cost per unit of travel time.

[0091] In this embodiment, the cargo AGV The uninstallation target is determined as follows:

[0092] ;

[0093] When the overall cost of multiple candidate unloading sites is the same, the selection can be further made according to the shorter processing time of the corresponding equipment or the preset site priority rules.

[0094] Therefore, the unloading allocation result at the current decision moment is obtained:

[0095] ;

[0096] Figure 4 The process of determining the unloading target for a cargo AGV is illustrated.

[0097] S3, perform task capacity constraint matching on the current demand transportation task set and the candidate mobile handling robot set to obtain the current demand task matching result at the current decision time; and take the unmatched mobile handling robots in the candidate mobile handling robot set as unmatched mobile handling robots.

[0098] Targeting the current demand for transportation tasks With candidate AGV set Perform pickup task matching under capacity constraints. For any currently demanding transportation task:

[0099] ;

[0100] in, Indicates the site corresponding to the task. Indicates the workpiece type. Indicates a stage in the process.

[0101] First, the statistical task The actual number of waiting times at the current decision-making moment is denoted as . This indicates the number of similar tasks that are currently in a waiting state for pickup.

[0102] Calculate the age pressure value of the task based on the time it takes for the corresponding workpiece to enter the current waiting position:

[0103] ;

[0104] in, Representation and Task The corresponding set of currently waiting workpieces, Indicates workpiece Duration of stay at the current waiting position, These are the preset weighting coefficients.

[0105] To characterize the impact of future upstream workpieces on the current task site demand, the future impact of the task is constructed:

[0106] ;

[0107] in, This indicates that the current system can subsequently flow to tasks. The set of upstream related workpieces corresponding to the site, Indicates upstream workpiece The current stage of the process. Indicates the quantity of the corresponding workpiece. Indicates workpiece Arrive at the mission The estimated remaining time at the corresponding location. This is the preset attenuation coefficient.

[0108] To standardize the units of measurement, in this embodiment, the above components are normalized:

[0109] , , ;

[0110] Based on this, construct the task. Queue pressure value:

[0111] ;

[0112] in, , , These are the preset weighting coefficients.

[0113] To reflect the continuity of tasks in subsequent processes after the current pickup is completed, downstream availability is constructed:

[0114] ;

[0115] in, Indicates task The minimum expected waiting cost to proceed to optional downstream equipment after the current transport is completed. Preferably, The remaining processing time of the corresponding candidate equipment and the queuing load of the input buffer can be jointly determined. For final operations with no subsequent processing equipment to choose from, a final operation indicator is defined:

[0116] ;

[0117] Therefore, construct the task Overall demand value:

[0118] ;

[0119] in, , , These are the preset weighting coefficients.

[0120] Furthermore, by normalizing the task demand values ​​in the current set of tasks with active demand, we obtain:

[0121] ;

[0122] Based on the normalized task requirement value, combined with the current total number of candidate AGVs and the current actual number of tasks waiting. Define the task Capacity constraints:

[0123] ;

[0124] Based on this, task-side preference relationships and AGV-side preference relationships are constructed respectively.

[0125] Regarding task-side preference relationships, for any candidate AGV Calculate its arrival time at the task Estimated arrival time:

[0126] ;

[0127] in, Indicates AGV The estimated release time for the current pickup task can begin. This indicates the release station corresponding to the release time. Indicates the distance between stations. This represents the cost per unit of travel time.

[0128] In this embodiment, when the AGV is in an unloaded state, take ,and Its current position; when the AGV is in the unloading execution state. Given its remaining uninstallation time, and This is the current unloading station; when the AGV is loaded but has not yet finished unloading, The sum of the remaining driving time and the unloading time required to complete the current unloading, and The target site for unloading is determined in step S2.

[0129] Task according to Candidate AGVs are sorted from smallest to largest to form a task-side preference list; when multiple AGVs have the same estimated arrival time, the currently idle AGV is selected first.

[0130] Regarding AGV-side preference relationships, in this embodiment, it is preferable to base them on task requirement values. The cost of the AGV reaching the task station after release is also considered. Preferably, for candidate tasks near the AGV, tasks with higher demand values ​​are prioritized; for candidate tasks further away, tasks with lower arrival costs are prioritized, thus balancing task urgency and AGV response cost.

[0131] Finally, under the task capacity constraint Next, based on the task-side preference relationship and the AGV-side preference relationship, a capacity-constrained bilateral matching is performed to obtain the matching result of the current positive demand task. , Figure 5 The process of matching positive demand tasks is illustrated.

[0132] S4, match the unmatched mobile transport robots with the set of potential transport tasks, and obtain the prospective pre-positioning allocation result of the unmatched mobile transport robots at the current decision time;

[0133] For AGVs in S3 that are not matched with a currently needed task, further lookahead pre-positioning allocation is performed. Let the set of unmatched AGVs be:

[0134] ;

[0135] Before executing a new forward deployment allocation, preferably, the validity of forward deployment intentions left over from the previous decision time is first checked. If the potential task corresponding to a certain forward deployment intention has not yet been transformed into a currently needed task, its estimated readiness time is still valid, and the time difference between the corresponding AGV and the potential task is still within a preset feasible window, then the forward deployment intention is retained; otherwise, the forward deployment intention is cleared. If the same potential task is held by multiple AGVs simultaneously with forward deployment intentions, only the one with the earliest estimated arrival time and priority to the empty AGV is retained, and the remaining duplicate forward deployment intentions are cleared.

[0136] Furthermore, for any unmatched AGV and any potential task Calculate the estimated arrival time of the AGV to the station corresponding to the potential task:

[0137] ;

[0138] in, Indicates AGV The anticipated release time for forward-looking pre-positioning can begin. Indicates the corresponding release site location. Indicates potential task The target site location.

[0139] Further, calculate AGV With potential tasks Time difference between them:

[0140] ;

[0141] In this embodiment, an advance tolerance threshold is set. and hysteresis tolerance threshold When satisfied At that time, it was believed that AGV With potential tasks Matching in terms of time is feasible.

[0142] Meanwhile, to avoid repeatedly including tasks that have already become current demand in the forward deployment, in this embodiment, it is preferable to only include potential tasks with a current actual waiting quantity of zero and an effective expected readiness time as forward deployment candidate tasks. Preferably, it may further be required that the corresponding potential task has a positive demand evaluation value.

[0143] Therefore, for any unmatched AGV The set of feasible look-ahead pre-positioning tasks is defined as follows:

[0144] ;

[0145] In this embodiment, a single capacity allocation is used for the look-ahead pre-deployment task, meaning that each potential task is matched with at most one AGV. Therefore, its capacity constraint is set to... ;

[0146] For potential task-side preference relationships, it is preferable to follow the expected arrival time of each candidate AGV. Sort by size from smallest to largest; when estimated arrival times are the same, prioritize empty AGVs.

[0147] For AGV side preference relationships, the preferred approach is based on the absolute value of the time difference. The tasks are sorted from smallest to largest to prioritize those that match the expected readiness time of the potential tasks more closely. When the absolute time difference of multiple potential tasks is the same, the task with the smaller expected arrival time is prioritized. In cases where the time difference is even more equal, the potential task with the higher demand evaluation value can be prioritized.

[0148] Finally, under the single capacity constraint, a two-sided matching process is performed between the unmatched AGVs and the set of feasible forward-looking pre-positioning tasks to obtain the forward-looking pre-positioning allocation results. , Figure 5 It also illustrates the joint allocation process of positive demand task matching and forward-looking pre-positioning.

[0149] S5, based on the unloading allocation results, the current demand task matching results, and the forward pre-positioning allocation results, forms the transportation allocation result for the current decision moment.

[0150] The unloading allocation results obtained in S2, the positive demand task matching results obtained in S3, and the forward pre-positioning allocation results obtained in S4 are integrated to form the transportation allocation results at the current decision moment.

[0151] In this embodiment, the loading class intent set is represented as:

[0152] ;

[0153] Accordingly, the transportation allocation result at the current decision moment can be expressed as:

[0154] ;

[0155] in, This is used to reflect the unloading, picking, or forward positioning tasks assigned to each AGV at the current decision-making moment.

[0156] S6, based on the unfinished tasks left over from the previous decision time and the transportation allocation results at the current decision time, update the sequence of transportation intentions to be executed for each mobile handling robot; the mobile handling robot executes the task according to the sequence of transportation intentions to be executed.

[0157] Based on the unfinished tasks left over from the previous decision time and the transportation allocation results at the current decision time, update the sequence of transportation intentions to be executed for each AGV.

[0158] AGV At the moment of decision The subsequent sequence of transport intentions to be executed is as follows Then the set of all AGVs' unexecuted transport intention sequences can be represented as:

[0159] ;

[0160] In this embodiment, S6 includes the following processing rules.

[0161] First, delete the first historical intent that has actually been completed from the historical intent sequence of the previous decision moment. Here, "completed" means that the corresponding loading or unloading operation has been completed and caused a change in the AGV's execution status, not just that the AGV has reached the target station.

[0162] Secondly, for AGVs that have not yet entered the loading or unloading execution phase, their first and subsequent unexecuted intentions are updated based on the transportation allocation results recalculated at the current decision moment. In other words, the transportation allocation result for this round is not a differential increment relative to the historical plan, but rather a regenerated allocation output for the current round based on the current system state; the final sequence of unexecuted transportation intentions... It is obtained by updating the current round's allocation output and the historical unfinished intentions under the constraints of the execution state.

[0163] Furthermore, for AGVs that have entered the loading or unloading execution phase, their current primary intent remains locked, and only their subsequent unexecuted intents are refreshed, thereby avoiding the need to roll back and modify the currently executing loading or unloading targets.

[0164] The sequence of transport intentions to be executed satisfies the following constraints:

[0165] When AGV When the AGV is unloaded, ;

[0166] When AGV When used as a cargo AGV, ;

[0167] In this context, the primary intention of a loaded AGV is an unloading intention, and the secondary intention, if present, is either a pickup intention or a look-ahead pre-positioning intention. For an unloaded AGV, the sole intention, if present, is either a pickup intention or a look-ahead pre-positioning intention. Look-ahead pre-positioning allocation is only executed for AGVs that are not matched at the current decision time, to prevent the same AGV from simultaneously receiving both a pickup intention and a look-ahead pre-positioning intention at the same decision time.

[0168] Figure 6 The process of updating the transport intention sequence is illustrated.

[0169] In this embodiment, S1 to S6 are re-executed based on the current system state at each decision moment, and the current decision result is updated with the historical unfinished tasks to form a sequence of transportation intentions to be executed that is dynamically adjusted according to the changes in the system state, thereby realizing the rolling allocation and continuous adjustment of transportation tasks.

[0170] Through the above implementation methods, this method can simultaneously handle the selection of unloading targets for cargo AGVs, matching of current demand tasks, and forward pre-positioning of unmatched AGVs, thereby achieving coordinated allocation of current and potential demands. At the same time, by linking and updating the current decision results with historical unfinished tasks, the system can achieve continuous decision-making and stable execution in a dynamic environment, and enhance its responsiveness to short-term future transportation demands.

[0171] It should be noted that the above embodiments use an AGV transportation system in a flexible workshop as an example to illustrate this method, but the method is not limited to this specific scenario. For mobile handling robot systems with object transfer capabilities in warehousing, sorting, distribution, and other dynamic multi-site operation scenarios, as long as there is a coexistence of current and potential transportation needs, dynamic changes in robot status, and collaborative operation of multiple service stations, the method of this invention can be used to achieve rolling allocation and forward-looking pre-positioning of transportation tasks.

[0172] To further verify the application effect of this method in dynamic multi-site operation scenarios, a set of simulation verification examples and comparative examples are used to further illustrate the present invention. It should be noted that the following simulation verification examples and comparative examples are only used to illustrate the effectiveness of the present invention and are not intended to limit the scope of protection of the present invention.

[0173] In a simulation verification example, the following is adopted: Figure 1The flexible workshop transportation scenario shown is a specific application example of this method in a dynamic multi-site operation scenario. The system includes 120 passable stations, 1 input station, 1 output station, 9 processing devices, and 11 robotic arms. The AGV service stations corresponding to the input and output stations are as follows: and The nine processing machines are respectively labeled as... Their corresponding service sites are respectively The time it takes for an AGV to complete one movement between adjacent stations. The loading and unloading operations each take 4 seconds, with a time limit of 2 seconds. The process routes, candidate machine sets, and corresponding processing time settings for the three types of workpieces are shown in Table 1.

[0174] Table 1: Production process steps for three types of workpieces.

[0175]

[0176] To establish a fixed verification scenario, in this simulation verification example, the initial quantities of three workpiece types 1, 2, and 3 are set to 8, 8, and 8 pieces respectively, totaling 24 workpieces. Initially, all workpieces are located at the input station awaiting entry into the first process. Furthermore, the number of AGVs is set to 6, each with a carrying capacity of 1, and all are initially in an empty state. The initial positions of the AGVs are determined by… Figure 1 As can be seen, the system runs in a rolling fashion from the initial state until all workpieces have completed all processing steps and been transported to the output station.

[0177] In this simulation verification example, to ensure the reproducibility of the verification process, the main parameters related to the rolling allocation of transportation tasks are further set as follows: During the unloading target determination process, the weight of the same type of waiting load in the comprehensive unloading cost function is adjusted. Other types of wait load weights Take them respectively and .

[0178] In the process of matching tasks with positive demand, the weighting coefficient of age pressure is considered. Future impact on attenuation coefficient Weights of each component in the task queue pressure value , , The values ​​are set to 0.1, 1.0, and 0.1 respectively; the advance tolerance threshold in look-ahead pre-positioning allocation. and hysteresis tolerance threshold They were set to 6.0 and 5.0 respectively; , , .

[0179] In this validation example, the comprehensive demand assessment is based primarily on the current waiting pressure of the task, highlighting the role of this method in dynamic transportation demand response and forward-looking pre-positioning.

[0180] Under the same scenario and parameter conditions described above, the following comparison example is set up.

[0181] Comparative Example 1: Except for not performing the forward pre-positioning allocation of unmatched AGVs, the system layout, workpiece configuration, AGV quantity, transportation rules, parameter settings, and rolling allocation process are all the same as the proposed method. Specifically, after the matching of the current demand task is completed, the unmatched AGVs will no longer be pre-positioned according to potential transportation tasks, but will remain in their current idle state, waiting to participate in the allocation again at subsequent decision moments.

[0182] Comparative Example 2: A real-time allocation method without sequence and only involving target locking during execution is adopted. Except for not maintaining the AGV's pending transport intention sequence, the system layout, workpiece configuration, AGV quantity, transport rules, parameter settings, and transport allocation process are all the same as this method. Specifically, at each decision moment, the current real-time target is issued to the AGV only based on the current round's transport allocation result, without forming a task sequence that is updated across decision moments; empty AGVs only receive the current pickup target or pre-positioned target, and loaded AGVs only receive the current unloading target; only when the AGV has entered the loading or unloading execution phase does its currently executing target remain unchanged, and after the target is completed, the subsequent tasks are recalculated at the next decision moment.

[0183] In this simulation verification example, the following metrics are used to compare the various schemes: AGV idle travel time, average task waiting time, average transportation response time, AGV utilization rate, and system completion time. Specifically, AGV idle travel time reflects the cumulative movement time of the AGV in an unloaded state; average task waiting time reflects the average waiting level from when a transportation demand enters the waiting state until it leaves the waiting queue; average transportation response time reflects the average time cost from when a transportation demand is formed until it first receives a real pickup matching allocation; AGV utilization rate reflects the percentage of time the AGV is in a state of loading, unloading, or holding a transportation task to be executed; and system completion time reflects the overall completion efficiency of all workpieces being processed and delivered to the output station.

[0184] The statistical results of each scheme after running under the same initial conditions are shown in Table 2.

[0185] Table 2: Statistical Results of Operation.

[0186]

[0187] As shown in Table 2, compared with Comparative Example 1, the proposed method slightly increases the three local logistics indicators: AGV empty travel time, average task waiting time, and average transportation response time. However, the AGV utilization rate increases from 0.8572 to 0.9096, and the system completion time is shortened from 769 seconds to 745 seconds. This indicates that although the system needs to bear certain costs of active empty travel and local waiting after introducing the unmatched AGV forward-looking pre-positioning mechanism, it can improve the overall utilization level of AGVs and improve the system-level flow efficiency and overall completion efficiency. In other words, the main benefit of forward-looking pre-positioning is reflected in the improvement of system throughput and completion performance, rather than ensuring that all local transportation indicators are simultaneously optimal.

[0188] Furthermore, compared to Comparative Example 2, the proposed method demonstrates superior performance in terms of AGV empty travel time, average task waiting time, average transport response time, and system completion time. Specifically, the system completion time is reduced from 818 seconds to 745 seconds, and the average transport response time is significantly reduced from 5.517 seconds to 1.659 seconds. This indicates that by constructing and continuously updating the AGV's pending transport intention sequence, this method maintains a continuous connection between unloading tasks and subsequent pickup / pre-positioning tasks. It avoids the problems of lost subsequent tasks, repeated waiting, and interrupted connections caused by only issuing the current target in the real-time allocation method, thereby improving the continuity, responsiveness, and overall execution efficiency of transport organization.

[0189] Therefore, based on the simulation verification results above, it can be shown that this method can simultaneously handle the selection of unloading targets for cargo-carrying AGVs, matching of current active demand tasks, and forward pre-positioning of unmatched AGVs, thereby achieving coordinated allocation of current and potential demands. At the same time, by continuously maintaining the AGV transportation intention sequence, it can achieve continuous connection and stable execution of transportation tasks in a dynamic environment, thereby improving AGV utilization and shortening system completion time.

[0190] The above are merely specific embodiments of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial modifications made to the present invention using this concept shall be considered as infringing upon the protection scope of the present invention.

Claims

1. A method for rolling allocation of transportation tasks for mobile transport robots based on forward-looking pre-positioning, applied to a dynamic multi-site operation system including several service stations and several mobile transport robots, characterized in that, The following steps are included in the rolling execution: S1, obtain the status information of the dynamic multi-site operation system at the current decision moment, and obtain the current set of candidate mobile handling robots participating in transportation allocation, the current set of transportation tasks in demand, and the set of potential transportation tasks; the candidate set of mobile handling robots includes the set of loaded mobile handling robots and the set of unloaded mobile handling robots. S2, using the unloading comprehensive cost function to match the set of cargo machines with the information of optional unloading sites, to obtain the unloading allocation result at the current decision moment; S3, perform task capacity constraint matching on the current demand transportation task set and the candidate mobile handling robot set to obtain the current demand task matching result at the current decision time; and take the unmatched mobile handling robots in the candidate mobile handling robot set as unmatched mobile handling robots. S4, match the unmatched mobile transport robots with the set of potential transport tasks, and obtain the prospective pre-positioning allocation result of the unmatched mobile transport robots at the current decision time; S5. Based on the unloading allocation results, the current demand task matching results, and the forward-looking pre-positioning allocation results, the transportation allocation results for the current decision moment are formed. S6. Based on the unfinished tasks left over from the previous decision moment and the transportation allocation results at the current decision moment, update the sequence of transportation intentions to be executed for each mobile handling robot. Mobile transport robots execute tasks according to a sequence of transport intentions to be performed; In step S4, the unmatched mobile transport robots are matched with the set of potential transport tasks to obtain the forward-looking pre-positioning allocation result of the unmatched mobile transport robots at the current decision time, as follows: Perform a validity check on the look-ahead pre-positioning intentions left in the sequence of transport intentions to be executed by the mobile transport robot at the previous decision moment; Calculate the estimated arrival time of the unmatched mobile transport robot to the site corresponding to the potential task; then calculate the time difference between the mobile transport robot and the potential task by dividing the estimated arrival time by the estimated readiness time. For unmatched mobile transport robots, construct a feasible set of prospective pre-positioning tasks; represented as: ； in, This represents a set of feasible forward-looking pre-positioning tasks; Represents a set of potential transportation tasks; Indicates the actual number of people waiting; Indicates the estimated readiness time; Indicates the early tolerance threshold; Indicates mobile transport robot With potential tasks The time difference between them; Indicates the hysteresis tolerance threshold; Under a single capacity constraint with a task capacity constraint value of 1, bilateral matching is performed between the unmatched mobile transport robot and the set of feasible forward-looking pre-positioning tasks to obtain the forward-looking pre-positioning allocation results.

2. The method for rolling allocation of transportation tasks for mobile handling robots based on forward-looking pre-positioning as described in claim 1, characterized in that, The unloading allocation result at the current decision moment is obtained by matching the set of cargo machines with the information of available unloading sites using the unloading comprehensive cost function, as follows: Construct a set of candidate unloading sites for each mobile handling robot in the cargo handling robot set; For each candidate unloading site in the candidate unloading site set, construct the comprehensive unloading cost function, expressed as: ； in, This represents the overall cost function for unloading. Indicates candidate uninstallation sites The corresponding processing equipment input buffer area and the moving and transporting machine Waiting load for workpieces of the same type and stage currently under load; This indicates the pending load for other types of workpieces or workpieces at other stages; Indicates candidate uninstallation sites The remaining processing time for the workpiece currently being processed by the corresponding equipment; Indicates mobile transport robot Move from current location to candidate uninstallation site The estimated cost of driving; and Preset weighting coefficients; Each mobile handling robot in the cargo handling robot set selects the candidate unloading station with the smallest value of the unloading comprehensive cost function to obtain the unloading allocation result at the current decision time.

3. The method for rolling allocation of transportation tasks for mobile handling robots based on forward-looking pre-positioning as described in claim 1, characterized in that, The task capacity constraint is specifically implemented as follows: Downstream availability is constructed and expressed as: ； in, Indicates downstream availability; Indicates task Minimum expected waiting cost to proceed to optional downstream equipment after the completion of the current transport; Construct tasks based on downstream availability The overall demand value is expressed as: ； in, This represents the overall demand value; , and Indicates the preset weighting coefficient; Indicates task The queue pressure value; Indicates the final process quantity; Based on task Determining the overall demand value The task capacity constraint is expressed as: ； in, Indicates task capacity constraints; This represents the normalized overall demand value; This indicates the total number of current candidate mobile handling robots; Indicates the current number of tasks that are actually waiting; This indicates taking the minimum value; This indicates taking the maximum value.

4. The method for rolling allocation of transportation tasks for mobile handling robots based on forward-looking pre-positioning as described in claim 3, characterized in that, The task The queue pressure value is expressed as: ； in, Indicates task The normalized actual number of waiting items at the current decision-making moment; This represents the age stress value of the task after normalization; This represents the future impact of the task after normalization; The future impact of the task is represented as follows: ； in, This indicates that the current system can subsequently flow to tasks. The set of upstream related workpieces for the corresponding site; Indicates the process stage index; Indicates workpiece Stage of the process; Indicates the quantity of the corresponding workpiece; Indicates workpiece Arrive at the mission The estimated remaining time at the corresponding location. This is the preset attenuation coefficient.

5. The method for rolling allocation of transportation tasks for a mobile handling robot based on forward-looking pre-positioning as described in claim 3, characterized in that, The picking task matching under the task capacity constraint is based on the task-side preference relationship and the mobile handling robot-side preference relationship. The task-side preference relationship is as follows: Task Candidate mobile handling robots are sorted from smallest to largest according to their estimated arrival time to form a task-side preference list; when multiple mobile handling robots have the same estimated arrival time, the currently idle mobile handling robot is selected first. The preferred relationship of the mobile transport robot is as follows: for candidate tasks near the mobile transport robot, priority is given to candidate tasks with larger overall demand values; for candidate tasks that are farther away, priority is given to candidate tasks with lower travel costs after the mobile transport robot is released to reach the task station.

6. The method for rolling allocation of transportation tasks for a mobile handling robot based on forward-looking pre-positioning as described in claim 5, characterized in that, The estimated arrival time is expressed as: ； in, Indicates mobile transport robot Arrive at the mission The estimated arrival time; Indicates mobile transport robot The estimated release time for the current pickup task can begin. This indicates the release station corresponding to the release time; Indicates the distance between stations; Indicates the cost per unit of travel time; Indicates task The target site location.

7. The method for rolling allocation of transportation tasks for a mobile handling robot based on forward-looking pre-positioning as described in claim 1, characterized in that, The validity check of the forward-looking pre-positioning intentions left over from the previous decision-making moment specifically includes: If a potential task corresponding to a certain forward positioning intention has not yet been transformed into a current active demand task, and its estimated readiness time is still valid, and the time difference between the corresponding mobile transport robot and the potential task is still within the preset feasible window, then the forward positioning intention is retained; otherwise, the forward positioning intention is cleared. If the same potential task is held by multiple mobile transport robots at the same time, only the one with the earliest estimated arrival time and priority to the empty mobile transport robot is retained, and the other duplicate forward positioning intentions are cleared.

8. The method for rolling allocation of transportation tasks for a mobile handling robot based on forward-looking pre-positioning as described in claim 1, characterized in that, Under the single capacity constraint with a task capacity constraint value of 1, a two-sided matching is performed between the unmatched mobile transport robot and the set of feasible forward-looking pre-positioned tasks, and the matching is based on the potential task-side preference relationship and the mobile transport robot-side preference relationship. The potential task-side preference relationship is as follows: each candidate mobile handling robot is sorted from smallest to largest according to its expected arrival time. When the expected arrival times are the same, prioritize the use of an empty mobile handling robot; The preferred relationship of the mobile handling robot is as follows: sorting the potential tasks by the absolute value of the time difference between the mobile handling robot and the potential task from smallest to largest; when the absolute values ​​of the time differences of multiple potential tasks are the same, the task with the smaller expected arrival time is selected first; in the case of further equivalence, the potential task with the higher demand evaluation value is selected first.

9. The method for rolling allocation of transportation tasks for a mobile handling robot based on forward-looking pre-positioning as described in claim 1, characterized in that, Based on the unfinished tasks left over from the previous decision time and the transportation allocation results at the current decision time, the sequence of transportation intentions to be executed for each mobile handling robot is updated, satisfying the following constraints: the length of the sequence of transportation intentions to be executed for an empty mobile handling robot does not exceed 1; the length of the sequence of transportation intentions to be executed for a loaded mobile handling robot is 1 or 2, wherein the first intention is the unloading intention, and the second intention is the picking intention or the look-ahead pre-positioning intention; the look-ahead pre-positioning allocation is only executed for mobile handling robots that are not matched at the current decision time.

Images