System and method for dynamic network based automated path planning for freight stations

Abstract

A control device, connected to a dynamic path planning system for automated freight stations, includes multiple modules. A dynamic network construction module constructs a time-varying network structure based on the static locations of equipment and the real-time status of movable equipment; a virtual edge visualization module generates temporary connected paths and updates dynamic virtual edges based on the current locations of movable equipment; an equipment status management module maintains the current location, available time, task status, and operating parameters of movable equipment; a cost calculation module calculates the combined cost of each path segment, including movement and preparation costs, based on the network structure, dynamic virtual edges, and equipment status; a path search module performs a search in the time-dependent network based on the combined cost to obtain the optimal path; and a data integration and communication module receives real-time equipment data, integrates system status, provides it to the aforementioned modules, and transmits the generated optimal path to an execution processor to drive the movable equipment to perform transportation tasks.

Description

Technical Field

[0001] This disclosure relates to the field of automated logistics and route planning technology. More specifically, the present invention relates to an automated route planning system and method for freight stations based on dynamic network modeling, which can generate dynamic connectivity and calculate time-dependent shortest paths based on the real-time status of movable equipment. Background Technology

[0002] In the field of shortest path algorithms, previous applications have focused on static networks, such as road systems or fixed warehouse layouts. Traditional algorithms assume constant edge weights and predictable transport times, making them highly effective for applications where network topology and associated transport costs remain constant over time. While these methods can meet basic path requirements in automated freight stations and other logistics environments, their effectiveness diminishes when the network is subject to dynamic changes, such as equipment movement or real-time shifts in operating conditions.

[0003] In modern freight stations, the integration of mobile devices such as freight elevators and conveyor vehicles creates dynamic connections, rendering traditional path calculation methods inapplicable. Therefore, the challenge of shortest path algorithms lies in accurately modeling the freight station as a time-varying network.

[0004] For example, traditional solutions cannot cope with the dynamic changes brought about by mobile devices, which leads to the inability of route planning decisions to reflect the current operating status of freight stations, resulting in inefficiency and extended cargo handling time. In addition, traditional solutions also have the following shortcomings: (1) Lack of dynamic connectivity modeling: Existing route planning systems usually assume that the network topology is fixed and cannot express the temporary connectivity relationships caused by the movement of mobile devices. Since these dynamic changes are not included in the model, the routes generated by the system cannot reflect the real accessibility, thereby reducing the overall transportation efficiency. (2) Failure to integrate the real-time availability of mobile devices: Traditional algorithms cannot obtain and utilize real-time information such as the current location and occupancy status of devices, making route planning based on outdated or averaged data. Such route decisions are difficult to adapt to the real-time operating status of freight stations, resulting in increased waiting time and decreased flow efficiency. (3) Lack of a composite cost model that considers both movement and preparation delays: Traditional methods often only calculate costs based on fixed distances or static travel times, ignoring the time required for device repositioning and preparation. Since these delays are not taken into account, route planning often underestimates the actual transportation time, making the theoretically optimal route perform poorly in practice.

[0005] Therefore, a novel path planning solution is still needed, which is expected to be able to dynamically simulate network connectivity and integrate real-time device status to overcome the shortcomings of existing technologies. Summary of the Invention

[0006] One object of this invention is to provide a system and method to solve the aforementioned technical problems. The proposed technical solution includes a comprehensive framework designed to address the long-standing complex congestion problem in cargo material handling systems (e.g., air cargo material handling systems). Furthermore, it introduces several innovative technical means to improve congestion detection, analysis, and management, thereby enhancing the overall operational efficiency of the cargo logistics industry (e.g., air cargo logistics).

[0007] This invention proposes a dynamic path planning system, called "OptCargoRoute," specifically designed for automated freight stations. The proposed system models the freight station as a time-varying network, where static equipment locations serve as nodes, and connections between nodes are dynamically established through virtual edges. These virtual edges consist of movable equipment such as freight elevators and conveyor vehicles, and these devices change the network topology in real time. Unlike traditional static models, this proposed system considers the constantly changing operating state of the freight station, thereby enabling more accurate and efficient path planning.

[0008] The proposed technical solution includes a contributing technical feature: a state-dependent cost calculation function. This function consists of two key components: movement cost and setup cost. Movement cost quantifies transport time based on physical distance and equipment speed, while setup cost reflects the time required for movable equipment to reach the departure node. By integrating these two costs, the system can accurately estimate transport time, thus encompassing the inherent distance-related delays and operational delays in automated cargo handling. Since existing technologies typically neglect real-time equipment availability and setup delays, this comprehensive cost function overcomes these shortcomings.

[0009] The proposed technical solution also introduces an algorithm capable of efficiently calculating the optimal path in dynamic environments. For efficient optimal path calculation, time-dependent variants can be used, such as the improved Dijkstra's algorithm (TD-Dij) and the heuristically based time-dependent A* algorithm (TD-A-Star). These algorithms dynamically update network connectivity and edge costs using real-time operational data, ensuring that path decisions reflect the current state of the terminals. In practical operation, these improvements enable rapid recalculation and real-time adjustments, significantly improving the efficiency and response speed of path planning.

[0010] According to a first aspect of the present invention, a control device is provided, which is connected to a system for dynamic path planning in an automated freight station, characterized in that it comprises: a dynamic network construction module, a virtual edge visualization module, an equipment status management module, a cost calculation module, a path search module, and a data integration and communication module. The dynamic network construction module is used to construct a network structure that changes over time based on the location of static equipment and the real-time status of movable equipment. The virtual edge visualization module is used to generate temporary connected paths based on the current location of the movable equipment and update the dynamic virtual edges in the network structure. The equipment status management module is used to maintain the current location, available time, task status, and real-time operating parameters of the movable equipment to obtain the equipment status of the movable equipment. The cost calculation module is used to calculate the combined cost for each path segment based on the network structure, the dynamic virtual edges, and the equipment status, which includes movement costs and preparation costs. The path search module is used to perform a path search in a time-dependent network based on the combined cost to obtain the optimal path. The data integration and communication module is used to receive real-time device data and integrate system status, and provide the system status to the dynamic network construction module, the virtual edge realization module, the device status management module and the path search module, and transmit the optimal path to the execution processor, so that the execution processor drives the movable device to perform the transportation task.

[0011] According to a second aspect of the present invention, a system is provided for dynamic path planning in an automated freight station, characterized in that it comprises: a freight station, a sensor and data acquisition module, a control device, an execution processor, and a communication and data transmission network. The freight station includes multiple static devices and multiple movable devices. The static devices provide cargo transport paths at fixed locations. The movable devices perform horizontal or vertical cargo transport actions within the freight station. The sensor and data acquisition module acquires the current location, available time, operating status, and real-time operating parameters of the movable devices. The control device includes: a dynamic network construction module, a virtual edge visualization module, an equipment status management module, a cost calculation module, a path search module, and a data integration and communication module. The dynamic network construction module constructs a network structure that changes over time based on the location of the static devices and the real-time status of the movable devices. The virtual edge visualization module generates temporary connected paths based on the current location of the movable devices and updates the dynamic virtual edges in the network structure. The equipment status management module maintains the current location, available time, task status, and real-time operating parameters of the movable devices to obtain the equipment status of the movable devices. The cost calculation module calculates the combined cost of each path segment based on the network structure, the dynamic virtual edges, and the device status. This combined cost includes movement costs and preparation costs. The path search module performs a path search in the time-dependent network based on the combined cost to obtain the optimal path. The data integration and communication module receives real-time device data and integrates the system status, providing the integrated system status to the dynamic network construction module, the virtual edge realization module, the device status management module, and the path search module. The data integration and communication module transmits the optimal path to the execution processor, enabling the execution processor to drive the movable device to perform the corresponding transportation task based on the optimal path and to send the task execution status back to the control device. The communication and data transmission network transmits data between the sensors and data acquisition module, the control device, and the execution processor.

[0012] According to a third aspect of the present invention, a method is provided for dynamic path planning in an automated freight station, characterized by comprising the following steps: acquiring real-time operating data of movable equipment within the freight station, wherein the real-time operating data includes the current location, available time, task status, and operating parameters of the movable equipment; constructing a time-varying network structure based on the location of static equipment and the real-time operating data of the movable equipment, wherein the static equipment serves as a fixed node, and dynamic virtual edges are generated according to the current location of the movable equipment; maintaining the equipment status of the movable equipment, including recording the most recently encountered location, most recently available time, idle status, and task execution status of the movable equipment; calculating a combined cost for each path segment in the network structure, wherein the combined cost includes movement cost and preparation cost, wherein the movement cost describes the time for goods to physically move along the path segment, and the preparation cost describes the time required for the movable equipment to prepare to execute the path segment; performing a path search based on the combined cost using a time-related method to obtain an optimal path from the origin to the destination; and generating a transport instruction according to the optimal path, and causing the movable equipment in the freight station to execute the transport instruction.

[0013] Through the above configuration, dynamic network modeling, real-time device status integration, and a robust comprehensive cost function can be integrated, thus providing a comprehensive solution to the challenges currently faced by modern automated freight transport. The proposed technical solution not only improves cargo flow efficiency and minimizes transportation delays but also enhances the overall resilience of freight consolidation operations and can be applied to scenarios involving the Internet of Things, transportation, and industrial artificial intelligence technologies in the logistics field. Attached Figure Description

[0014] In the following sections, embodiments of the present invention will be described in more detail with reference to the accompanying drawings, wherein:

[0015] Figure 1A A schematic diagram of equipment types for an automated freight station according to an embodiment of the present invention is shown.

[0016] Figure 1B A schematic diagram of the layout of an example freight station according to an embodiment of the present invention is shown.

[0017] Figure 2A A schematic diagram illustrating the formation of virtual edges via a transport vehicle according to an embodiment of the present invention is shown.

[0018] Figure 2B A schematic diagram illustrating the formation of virtual edges via a cargo elevator according to an embodiment of the present invention is shown.

[0019] Figure 3 The execution logic of the time-varying Dijkstra (TD-Dij) algorithm is demonstrated.

[0020] Figure 4 Table 1 is provided, which shows the main performance comparisons of TD-Dij, TD-A-Star, RT, and Max-Prep-Dij on real-world query datasets.

[0021] Figure 5A A graph showing the average query response time for TD-Dij and TD-A-Star on the query set.

[0022] Figure 5B A distribution chart of OD types for each query set is displayed.

[0023] Figure 6 A schematic diagram illustrating the architecture of a system for dynamic path planning in an automated freight station according to an embodiment of the present invention is shown. Detailed Implementation

[0024] In the following description, as preferred examples, a system and method for automated route planning of freight stations based on dynamic networks are provided. Those skilled in the art will understand that modifications, including additions and / or substitutions, can be made without departing from the scope and spirit of the invention. Specific details may be omitted to avoid obscuring the invention; however, this disclosure is intended to enable those skilled in the art to implement the teachings herein without requiring extensive experimentation.

[0025] The automated route planning system for freight stations proposed in this invention optimizes freight flow by modeling the freight station as a time-varying network. In this model, the freight station can be represented as a directed graph, where nodes correspond to static equipment locations and edges represent feasible freight flow paths. The system enables the creation of virtual edges for the transfer of goods by movable equipment and assigns cost calculations related to equipment state to each virtual edge. The cost calculation can be the sum of movement costs (based on physical distance and equipment speed) and setup costs, where the setup cost considers the time required for the equipment to reach a usable state given its current state. To address the resulting time-varying shortest path problem, the technical solution of this invention introduces an improved algorithm and a time-varying variant of the A* algorithm, which utilizes an admissible heuristic to significantly reduce computation time while maintaining optimality. Ultimately, the technical solution of this invention can effectively model both static and dynamic aspects of freight stations, achieving more efficient and flexible route planning.

[0026] In order to better understand the technical effects achieved by the technical solution of the present invention, the difficulties or challenges encountered at present are first presented below.

[0027] 1-Related work

[0028] The technical field involved in this invention is related to shortest path algorithms, dynamic network modeling, and optimization of warehousing and logistics systems. The following is a brief description of the development of existing technologies to illustrate the limitations of current solutions.

[0029] 1.1-Path Planning Research in Warehousing and Material Handling

[0030] In traditional warehousing and logistics systems, shortest path algorithms are widely used for basic scheduling and path calculation. For example, existing technologies commonly employ shortest path algorithms to optimize picking routes, robot movement paths, or the travel paths of automated guided vehicles (AGVs) in warehouses. These methods typically focus on minimizing distance or time in fixed layout scenarios.

[0031] Some studies have proposed methods based on improved Dijkstra's algorithm or A* to handle variations in robot speed, uncertainties, or dynamic task assignments. While these studies involve dynamic factors, the network structure typically remains static, with its layout and connections unchanged during operation. Other works focus on local congestion control or regional scheduling in modular transport systems, improving overall throughput through local time windows or distributed coordination. However, these methods rely on a reconfigurable yet relatively fixed system structure, rather than a network topology that actually changes over time.

[0032] In general, existing research on warehouse routing focuses primarily on static or slightly dynamic environments, where network connections are predefined and do not change over time due to equipment movement. Therefore, these methods cannot handle the dynamic connectivity caused by movable equipment (such as freight elevators and conveyor vehicles) in automated freight stations, nor can they effectively describe the feasibility of real-world routes in such dynamic environments.

[0033] 1.2 - Time-dependent shortest path algorithms

[0034] The Time-Dependent Shortest Path (TDSP) algorithm aims to handle scenarios where edge costs are functions of time, extending the traditional static edge cost assumption. Some studies have analyzed directed graphs with time-dependent edge length functions in transportation networks satisfying the First-In-First-Out (FIFO) property, and based on this, have re-examined and extended core algorithm components, such as Dijkstra's algorithm, the heuristically guided ALT method, the partition-based Arc-Flags method, and hierarchical shrinkage structures, to suit time-dependent environments. The performance trade-offs of these methods have been evaluated in real-world networks.

[0035] Some studies have also proposed a bidirectional A* algorithm for time-dependent shortest paths, which operates on directed graphs with edge weights that are time functions and assumes the FIFO property. This method uses a lower bound function in the backward search to more effectively prune the forward search space, thereby accelerating the query while maintaining reasonable quality, and can be used to speed up the traditional Dijkstra method in road networks.

[0036] Furthermore, research has proposed a two-stage algorithm framework suitable for large-scale time-dependent shortest path graphs, representing the graph as a time-dependent graph structure containing positive edge delay functions. This method can handle both FIFO and non-FIFO networks simultaneously and allows users to define the starting time interval. Its core idea is to separate the path selection and time refinement processes, thereby efficiently calculating the shortest travel time and outperforming previous methods on large-scale graphs.

[0037] These studies demonstrate various approaches to modeling time-dependent graphs and developing efficient TDSP algorithms, each based on specific assumptions about time dependence and network characteristics. However, they all rely on the premise that the existence of edges is fixed and that time dependence is only reflected in the edge delay function.

[0038] In view of the challenges mentioned above, in order to make the technical solution of the present invention easier to understand, the theoretical basis of the technical solution will be explained below.

[0039] 2-Theoretical Basis and Definitions

[0040] 2.1 Equipment Classification and Characteristics

[0041] Figure 1A A schematic diagram of equipment types in an automated freight station according to an embodiment of the present invention is shown. To construct a freight transport path model, the equipment can be divided into two categories based on its mobility: static equipment and movable equipment. Static equipment includes powered rollers (PR), right-angle platforms (RA), and segmented driven rollers (FR), whose positions within the freight station are fixed. Static equipment can provide a continuous path for freight flow or act as a buffer. Movable equipment includes freight elevators (CH) and transport vehicles (TV). Freight elevators (CH) provide vertical transport between floors, while transport vehicles (TV) can run on tracks to achieve horizontal freight sorting. Figure 1B A schematic diagram of the layout of an example freight station according to an embodiment of the present invention is shown, focusing on describing the interaction between the aforementioned equipment types. Figure 1B The diagram shows the layout of static equipment and the operating range of movable equipment. White arrows indicate the direction of cargo flow, while dashed lines A and B represent the operating range of cargo lifts (CH) and conveyor belts (TV). This visually demonstrates how movable equipment dynamically expands the connectivity and operational capacity of the freight station.

[0042] In the illustrated layout, static equipment provides fixed connections, while movable equipment introduces dynamic connections and sources of state change, which directly affect the feasibility and cost of the route. How to determine these dynamic connections based on the real-time location and availability of movable equipment, and accurately reflect their changes during route planning, is a core consideration in the technical solution of this invention when constructing a cargo transportation route model.

[0043] 2.2 - Cargo Network and System Status

[0044] For clarity, the freight station can be modeled as a directed graph, taking into account the dynamic characteristics of movable equipment.

[0045] Definition 1 (Goods Network). A goods network is a directed graph. ,in It is the collection of all locations within the freight station. It is the set of all possible cargo flow paths between these locations.

[0046] The positions in the diagram correspond to the positions of static equipment, such as powered rollers (PR), right-angle platforms (RA), and segmented driven rollers (FR). The edges in the diagram represent feasible transport paths for goods, which can be accomplished by static or movable equipment, such as cargo lifts (CH) and transport vehicles (TV).

[0047] Definition 2 (System State). The system state of the network. It can be represented as a tuple It records the last known position and time of all movable devices. It can store each The last known floor and time, where express The last floor This indicates the time at which the last known location was reached. Can store each The last known location and time, where express The final location. This indicates the time at which the last known location was reached.

[0048] Each edge Each has an associated weight function. , indicating that in a given system state Below, position and The travel time between points. The weights are state-related and reflect the impact of the movable equipment's position on the travel time.

[0049] 2.3 - Problem Definition

[0050] This section can formally define the routing problem in freight networks.

[0051] Definition 3 (Path). From the starting position To the target location path Represented as a position sequence ,in , And in Under all conditions, .

[0052] Definition 4 (Path Cost). Given a path... An initial system state and a start time Then the cost of the path for:

[0053] in It is the location reached. The time is recursively calculated as And for calculate ; It is crossing the border The previous system state; and In a given system state and nodes Arrival time below The weight (travel time). Here, we can assume that the First-In-First-Out (FIFO) property means that goods that arrive at a location earlier will not leave later than goods that arrive later.

[0054] Problem Statement: Given a goods network A starting position A destination location An initial system state and a start time The goal is to find a way from arrive path This makes the path cost It can be minimized. Specifically, this is a shortest path problem in a time-varying network with state-dependent edge costs.

[0055] 3-Execution Method

[0056] 3.1-Concretizing Virtual Edges

[0057] To facilitate the application of the shortest path algorithm, the initial conceptual model of the freight station can be converted into a specific graph suitable for time-varying analysis.

[0058] Figure 2A A schematic diagram illustrating the formation of virtual edges using a transport vehicle (TV) according to an embodiment of the present invention is shown. The diagram illustrates the process of forming virtual edges along a cargo transport path using a transport vehicle (TV). Dashed arrows extend from nodes... Extend to nodes , and This indicates the potential directional paths that the transport vehicle (TV) can achieve. Furthermore, the arrows in the diagram are intended to emphasize the nodes. Goods cannot be received from the transport vehicle (TV), and nodes... Goods can only be received from the transport vehicle (TV), which clearly defines the direction of transport and the specific role of the transport vehicle (TV) in enhancing network connectivity.

[0059] Figure 2B A schematic diagram illustrating the formation of virtual edges using a freight elevator (CH) according to an embodiment of the present invention is provided. The diagram illustrates the process of creating a virtual edge for vertical cargo movement using a freight elevator (CH). The dashed arrows between floors represent the actual paths implemented by the freight elevator (CH), demonstrating how the internal vertical connectivity of the freight station is established. Furthermore, the arrows in the diagram indicate the connection relationships of the freight elevator (CH) across different floors, thereby enhancing the overall operational capacity of the freight station.

[0060] Step 1: Construct an initial graph. We can start with an initial graph, denoted as [example graph]. ,in A set representing static device locations (e.g., PRs, RAs, FRs). Initial edge set. Only edges representing direct and static connections between these locations are included. This initial graph does not include connections formed by movable devices (CHs and TVs), which are initially treated as virtual edges.

[0061] Step 2: Realize the virtual edges. The realization process converts these virtual connections into explicit edges, thus producing a realized graph. Vertex set Remaining unchanged represents the static device location. However, the edge set... Expanded to include static-to-static edges (i.e., those existing in...) (Items in the document) and new edges representing connections facilitated by movable devices. To determine which virtual edges need to be materialized, the location of each pair of static devices needs to be considered. .function Able to identify potential promoters and A set of movable equipment instances (CHs or TVs) along direct cargo flow paths. (Function) It is defined based on the physical layout of the freight station and the operational capacity of its equipment. If Return to the empty set ,but and There cannot be a direct connection between them (whether static or movable). For each pair If among them For a non-empty set, a directed edge is then... Add to the materialized edge set The middle. This side represents and Even if it relies on movable equipment, its potential for cargo transfer within a given scenario remains. For example... Figure 2A and Figure 2B As shown, it illustrates an example of a visualization graph that includes newly added edges representing the connections facilitated by the transport vehicle (TV) and the cargo elevator (CH).

[0062] In order to calculate the state-dependent edge cost, each edge can be... This is associated with the set of movable device instances that facilitate the transformation represented by the edge. This set can be represented as... .if If it is a static-to-static edge, then empty set .if The connection facilitated by the cargo elevator (CH) is then... yes A subset of (the set of all freight elevators (CH)). If The connection facilitated by the transport vehicle (TV) is then... yes A subset of (the set of all transport vehicles (TVs)). This association between edges and movable equipment is crucial for determining the prepare cost component in the time-varying edge cost function, as the prepare cost will depend on... The location and availability of specific movable device instances within the diagram. Therefore, the visualization diagram... Can connect each edge The mapping provides a complete representation of freight station connectivity, including the dynamic effects of movable equipment.

[0063] 3.2 - Constructing Time-Varying Networks

[0064] After materializing the virtual edges, we can obtain the materialized graph. The next step is to construct it as a time-varying network. This means traversing one of its edges... The resulting costs (travel time) are not fixed, but depend on when the goods begin to cross the edge and the state of the system at that time. This time dependence primarily stems from movable equipment (transport vehicles (TVs) and cargo lifts (CHs)). The time required for a transport vehicle (TV) or cargo lift (CH) to facilitate a connection may vary depending on its current location.

[0065] State-dependent edge cost function. Node. and The travel time between depends on the distance from Departure time (referred to as ) and system status The edge cost function of state dependency can be defined as follows: It represents the state of a given system. Below, at time From node Depart for the node The travel time. This cost consists of two parts: Move Cost and Prepare Cost, which are explained below.

[0066] (1) Movement Costs: Movement costs can be denoted as... This indicates that when the assumed necessary equipment is ready, and The travel time between these points can vary depending on the distance and the speed of the equipment.

[0067] in:

[0068] yes and The physical distance between them.

[0069] Return to the edge The device type (e.g., PR, RA, FR, CH, or TV).

[0070] These are constant speeds corresponding to the device type. It is assumed here that these speeds have been verified and are known beforehand.

[0071] (2) Preparation costs: Preparation costs can be recorded as follows: It indicates that the movable equipment is at the node. The time required to become available. For static-to-static edges, this cost is zero. For edges facilitated by CH / TV, it depends on the current location of the required device (in state). (is in) and from Departure time .

[0072] For the edge facilitated by the cargo elevator (CH), let For the required cargo lift (CH), we can obtain:

[0073] For the edge facilitated by the transport vehicle (TV), let... For the required transport vehicle (TV), we can obtain:

[0074] in yes Along the specific location of the track, yes The last known location (from which) of Quantity, specifically with Related tuples). It is the physical distance along the track, and That's the speed of a TV.

[0075] The item implements an "overlay mechanism". If the device is in time... If the arrival is earlier, the preparation cost is practically zero.

[0076] (3) Total marginal cost: Total marginal cost is the sum of moving costs and setup costs, and can be expressed as:

[0077] This paper applies Dijkstra's algorithm to time-varying networks. Since edge costs change over time, the standard Dijkstra algorithm cannot be directly applied. Therefore, Dijkstra's algorithm can be improved to handle these dynamic costs, resulting in the time-varying Dijkstra (TD-Dij) algorithm, whose execution logic is as follows: Figure 3 Algorithm 1 is shown.

[0078] In Algorithm 1, for each node The algorithm can maintain the known earliest arrival time. Priority queue Storeable by The nodes are sorted by value, allowing for efficient selection of the node with the shortest known arrival time. The core of the algorithm is the relaxation step. For the current node... Each neighbor It can be used in Current arrival time ( ) and initial system state Calculate the time-varying edge cost This calculation method can directly use the cost function defined in the previous section. If crossing an edge... Resulting in arrival Earlier (i.e.) ), then can Update to this new smaller value, The front drive is set as (It is used for path reconstruction) and updated using "decreaseKey". Priority in the priority queue. The algorithm can visualize the graph. Source node Target Node Initial system state and start time As input.

[0079] To further improve computational performance, especially for queries with known target locations, the A* search algorithm can be further adjusted for time-varying environments. This time-varying A-Star (TD-A-Star) method modifies the node priority strategy in the search. Although TD-Dijkstra (i.e., the aforementioned Algorithm 1) only considers the earliest known arrival time... Priority queue The nodes in the algorithm are sorted, but TD-A-Star can use composite priority values. .

[0080] function Can be used as a slave node To the target node Heuristic estimation of remaining driving time. Regarding the aforementioned TD-A-Star implementation, this involves using the function... Defined as considering only the static movement cost components of all edges. At that time, from arrive The shortest path cost. Due to the actual time-varying edge cost. The value is always greater than or equal to the corresponding movement cost (due to the non-negative setup cost), hence the heuristic function... It is admissible and acceptable, meaning it never overestimates the true minimum remaining time-varying cost. Therefore, TD-A-Star maintains the guarantee of finding the optimal (minimum cost) path, the same as TD-Dijkstra's path, but aims to improve search efficiency by more effectively guiding exploration towards the goal. For a given goal... Standard shortest path algorithms (such as Dijkstra's algorithm) can be used to find the shortest path from... Initially, the graph is weighted solely by movement costs, and all relevant nodes are pre-calculated. heuristic values .

[0081] 4- Conduct experiments

[0082] The experimental section provides an empirical evaluation of the proposed time-dependent shortest path algorithm in freight station route planning scenarios. Here, we focus on evaluating the time-dependent Dijkstra algorithm (TD-Dij) and its heuristic variant, the time-dependent A* algorithm (TD-A-Star), observing their performance in terms of path quality and computational efficiency. To gain a more comprehensive understanding of the effectiveness of these two methods, we also compare them with two other representative baseline methods: one is the Dijkstra method using the worst-case preparation time (Max-Prep-Dij) as the static weight, and the other is the routing table (RT) lookup method commonly used in this field. The experiments use real-world operational data and manually constructed query data to cover common operational scenarios under different conditions in automated freight stations.

[0083] 4.1 Experimental Setup

[0084] The experiments can be performed using a server.

[0085] 4.2-Dataset

[0086] To ensure the validity of the assessment, both real-world and synthetic data can be used simultaneously. The basic network model can be based on a large-scale, operational air cargo terminal. The air cargo terminal facility has a multi-story structure spanning six floors and is logically divided into two zones connected by only a few points. This structural characteristic will significantly affect cargo flow patterns and make route planning more complex.

[0087] Real-world data. To capture realistic operational scenarios, a dataset of 818 origin-destination (OD) queries was used, taken from freight station operational logs. The data was anonymized, sensitive information was removed, but the core content needed for route planning was retained. For ease of analysis, the real-world queries were categorized based on the relative relationships between origin and destination within the freight station structure. These categories represent different cargo flow patterns, including: Same Floor, Same Area (SFSW), indicating cargo movement within the same floor and area; Same Floor, Different Areas (SFDW), indicating cargo movement between different areas on the same floor; Different Floors, Same Area (DFSW), involving vertical movement of cargo within the same area via elevators; and Different Floors, Different Areas (DFDW), the most complex case requiring simultaneous movement across floors and areas. Performance analysis based on these categories helps to understand the algorithm's performance in addressing different route planning challenges in greater detail.

[0088] Synthetic Data. To further analyze algorithm performance within a controlled path complexity range, five synthetic query datasets, denoted Q1 to Q5, were generated, each containing 1000 randomly generated OD pairs. These synthetic datasets allow for systematic evaluation but are incompatible with RT baselines because pre-compiled routing tables do not necessarily contain entries for all possible OD pairs within the network. The generation process categorizes queries based on estimated path distances. First, the network diameter is approximated by calculating the shortest path for all node pairs using a static graph. The edge weights include the move cost and the average prepare cost for potential movable device edges. Subsequently, for each synthetic dataset... ( Randomly selected OD pairs (from 1 to 5) are filled, and their estimated shortest path distances fall within the range. Within. For example, Q1 contains a distance approximately within The pairings between them represent relatively short paths, while Q5 includes distances within... The pairings between the two represent the longest paths in the network. This construction method is sufficient to study how the algorithm's performance scales with increasing path length and complexity.

[0089] 4.3 Baseline Algorithm

[0090] The performance of TD-Dij and TD-A-Star can be compared with two baseline methods that represent different approaches to computing paths in such environments.

[0091] Max-Prep-Dij: Maximum preparation time weights. This baseline utilizes the standard Dijkstra algorithm on static graphs, where edge weights can only be pre-computed. For any edge that might involve movable equipment... Its weight The calculation is the sum of move cost and maximum possible prepare cost. This maximum prepare time assumes that the required movable equipment (cargo lift (CH) or transport vehicle (TV)) must travel from the furthest point within its operating range to the starting node. For example, for a freight elevator (CH) operating between floors 1 and 6, the maximum preparation time for a request to serve floor 2 will take into account the time required to travel from floor 1 to floor 2 and from floor 6 to floor 2, and take the larger of the two. Formally, this can be expressed as: This method represents a computationally efficient static approach, but due to its conservative weights, it is expected to produce only suboptimal path costs.

[0092] Route Table (RT). The baseline method replicates a common practice in this field, using a pre-computed route table for route decisions. Unlike dynamic route calculation, this method determines the next piece of equipment or the next location at each step by looking up the table. The route table is indexed by a triple containing: the origin region, the destination region, and the equipment currently handling the cargo. Each entry stores a list of possible next actions, ordered according to criteria defined by the freight station operator (e.g., rating or index). Typically, this method selects the first option in the list. The route table embodies a fixed, policy-based route selection approach, optimized for extremely fast lookup speeds. However, this method is entirely static and cannot be adjusted based on real-time system status, equipment availability, or congestion. Furthermore, the route table may not contain all possible origin-destination combinations in the network.

[0093] 4.4 - Evaluation Indicators

[0094] The performance of the evaluated algorithms (TD-Dij, TD-A-Star, Max-Prep-Dij, and RT) can be quantitatively assessed using three main metrics. First, average path cost (Travel Time) measures the quality of the computed path; this cost is the sum of the time dependency weights of each edge on the path, representing the total travel time (in seconds); a lower value indicates a more efficient path. Second, average query response time (computation time) quantifies the computational cost of each algorithm; it is measured as the average time required to compute a single path, in milliseconds; a lower value is more desirable in real-time deployments.

[0095] 5. Experimental Results and Analysis

[0096] This section interprets the experimental results. It analyzes results from real and synthetic datasets, highlighting the advantages and limitations of the TD-Dijkstra algorithm, and discusses its practical application value.

[0097] 5.1 - Real-world Data Analysis

[0098] Figure 4 Table 1 presents a comparison of the main performance of TD-Dij, TD-A-Star, RT, and Max-Prep-Dij on real-world query datasets. The results are categorized by query type (SFSW, SFDW, DFSW, DFDW) and an overall summary is provided. The results clearly demonstrate the effectiveness of the proposed time-related method in finding high-quality paths.

[0099] In terms of average path cost, TD-Dij and TD-A-Star consistently achieve the lowest values ​​across all query types, indicating that they have found the optimal path in a time-dependent sense. The average costs for both are identical (e.g., an overall average of 17.87 seconds) and significantly lower than the two baseline methods. Compared to commonly used RT methods in the field, the proposed dynamic method reduces the overall average path cost by approximately 17.7%. Compared to the static Max-Prep-Dij baseline, the average cost reduction is even greater, with an overall reduction of approximately 18.6%. These improvements demonstrate that incorporating real-time device status into path calculation significantly improves performance. Furthermore, the small standard deviations of TD-Dij and TD-A-Star indicate stable and consistent performance across queries of the same category.

[0100] Regarding average query response time, both baseline methods exhibit expected results: the RT method relies only on simple table lookups, thus completing queries almost instantaneously (approximately 0.04 milliseconds overall), while Max-Prep-Dij executes only a static Dijkstra's algorithm once, resulting in lower overhead (approximately 1.42 milliseconds overall). TD-Dij, due to its need for dynamic edge cost calculation, has significantly higher query times (approximately 20.99 milliseconds overall), reaching 42.63 milliseconds, especially in more complex DFDW queries. However, TD-A-Star, through its acceptable heuristic function, significantly reduces computation time, achieving a substantial speedup compared to TD-Dij, with an overall average query response time of only 0.53 milliseconds. Therefore, TD-A-Star maintains the optimal path quality of TD-Dij while significantly improving computational efficiency, making it more practical.

[0101] Overall, real-world data analysis validates that TD-A-Star achieves a good balance between path quality and query efficiency, obtaining the same optimal path cost as TD-Dij, but with significantly lower computational overhead, making it superior in practice compared to naive dynamic methods and static baseline schemes.

[0102] 5.2 - Synthetic Data Analysis

[0103] Figure 5A The graph shows the average query response time (milliseconds, logarithmic scale) of TD-Dij and TD-A-Star on query sets Q1 (lowest cost) to Q5 (highest cost), with TD-A-Star showing better scalability. Figure 5B The chart shows the distribution of OD types for each query set (in the order of SFSW, DFSW, SFDW, DFDW for each query set Q1 to Q5). The chart reveals an increased proportion of complex DFDW types in the higher-cost query sets. This structural change explains... Figure 5A The performance differences observed in [the study].

[0104] Experimental results show a significant correlation between path complexity, query type distribution, and algorithm performance. As the algorithm moves from query set Q1 to query set Q5—that is, from a query set with shorter paths to a query set with longer paths and higher complexity—the performance improves. Figure 5B The study reveals a clear trend in query composition: the proportion of simple queries within the same region or floor (such as SFSW and DFSW) is gradually decreasing, while complex queries requiring cross-floor and cross-region access (mainly DFDW) are increasing significantly. For example, DFDW queries accounted for only about 5% in query set Q1, but this figure rose to over 51% in query set Q5.

[0105] This increase in query complexity directly impacts computation time, such as... Figure 5A As shown, although the query time of both algorithms increases with the complexity of query set Q1 to Q5, the increase is more significant for TD-Dij. For example, the average query response time of TD-Dij increases sharply from 1.71 milliseconds for query set Q1 to 162.99 milliseconds for Q5. In contrast, TD-A-Star's query time increases more moderately due to the effective guidance of the heuristic function, rising only from 0.50 milliseconds for query set Q1 to 9.45 milliseconds for query set Q5. Therefore, as path complexity increases, TD-A-Star exhibits a more significant speedup compared to TD-Dij.

[0106] This analysis demonstrates that heuristic search plays a crucial role in narrowing the search space, particularly in more complex query types with longer paths, such as those represented by query set Q5. Since complex queries are representative of actual freight station operations, this further reinforces the practical value and advantages of TD-A-Star in real-world deployments.

[0107] As can be seen from the above theories and experiments, the technical solution proposed in this invention can effectively make substantial contributions in the real world. The following will further explain the dynamic path planning system for automated freight stations implemented using the technical solution.

[0108] Figure 6 This diagram illustrates the architecture of a system 100 for dynamic route planning in an automated freight station according to an embodiment of the present invention. System 100 is a dynamic route planning system deployed in an automated freight station, capable of constructing a dynamic network based on real-time operating conditions and calculating time-dependent shortest paths by a control device. System 100 includes various physical devices, data acquisition devices, communication networks, and control devices for performing route planning and scheduling within the freight station, enabling the freight station to adapt to changing operating conditions over time and achieve higher cargo handling efficiency.

[0109] System 100 may include a freight station 110, sensors and data acquisition modules 120, a communication and data transmission network 130, a control device 140, and an execution processor 150. The various components of system 100 can communicate and exchange data and information through wired or wireless connections.

[0110] Freight station 110 is equipped with various devices for cargo transport and handling, including static devices 112 that are fixed in place and movable devices 114 that can move within the freight station. The static devices remain in place and serve as fixed nodes in a dynamic network; these may include, as mentioned above, powered rollers (PR), right-angle platforms (RA), and segmented driven rollers (FR), providing basic cargo transport paths or temporary buffer spaces. In contrast, the movable devices 114 dynamically change position during operation; these may include, as mentioned above, cargo elevators (CH) and transport vehicles (TV). Cargo elevators are used for vertical transport between different floors, while transport vehicles perform horizontal transport within floors or across areas. The real-time location and availability of the movable devices 114 directly affect the connectivity of freight station 110 and the feasible transport paths for goods between different nodes.

[0111] The sensor and data acquisition module 120 is configured to collect the position, availability, operating speed, and other operating parameters of the movable equipment 114, and transmit this information to the control device 140 via wired or wireless communication. The sensor and data acquisition module 120 may include Internet of Things (IoT) devices, location tracking devices, industrial control devices, or other devices capable of providing real-time operating data. By continuously updating the equipment status, the system 100 can accurately reflect the actual operating conditions of the freight station 110 at every point in time.

[0112] The communication and data transmission network 130 is used for data exchange between the sensor and data acquisition module 120, the control device 140, and other components within the freight station 110. The communication and data transmission network 130 may include a local area network, a wireless network, an industrial automation communication protocol, or a combination thereof, to enable the timely transmission of equipment status within the freight station 110 to the control device 140 and to ensure that scheduling commands generated by the control device 140 can be sent back to the execution processor 150.

[0113] The control device 140 can be deployed at the freight station 110 or in a remote server. It receives real-time data from sensors and the data acquisition module 120, constructs a dynamic network model, evaluates the transportation costs of different routes, and executes a time-dependent shortest path search algorithm to generate the optimal route. In some embodiments, the control device 140 may include at least one processor, memory, and modular software running thereon for performing functions such as dynamic network modeling, cost calculation, route planning, and data integration. The internal structure and operating mode of the control device will be further described later.

[0114] The execution processor 150 can issue instructions to the movable device 114 according to the path or scheduling scheme generated by the control device 140, coordinate the movement of the cargo elevator and the conveyor, and feed back the task execution results to the control device 140, thereby forming a data closed loop. In some embodiments, the execution processor 150 may include a scheduling server, an equipment control unit, a PLC system action execution module, or a combination thereof, to convert the scheduling instructions generated by the control device 140 into specific execution actions for the movable device 114.

[0115] During the operation of system 100, the sensor and data acquisition module 120 continuously collects real-time information from the movable device 114 and transmits it to the control device 140 via a communication network. The control device 140 constructs a dynamic network structure based on these real-time states, including the positions of the static devices 112 as nodes and the dynamic virtual edges formed by the movable devices 114. The control device 140 calculates the cost of each edge in the network based on a state-dependent cost function and a time-dependent model, and uses a shortest path algorithm to generate the optimal path for goods from one location to another. The generated results are distributed to the corresponding devices via the execution processor 150, and then the movable devices 114 perform the actual goods transfer. After execution, the state of system 100 is updated and fed back to the control device 140 via the sensor and data acquisition module 120, forming a continuous loop of path planning and optimization. In this way, the temporary connectivity (virtual edges) determined by the real-time position of the movable devices 114 allows the reachability between nodes to change over time.

[0116] Through the above structure, system 100 can accurately model the complex and constantly changing operating conditions within freight station 110 and generate adaptive optimal path planning schemes in real time. By integrating the functions of equipment, sensor data, communication networks, and control devices in freight station 110, system 100 can maintain high-efficiency cargo flow performance in dynamic environments.

[0117] The control device 140 may include multiple functional modules to support dynamic route planning in automated freight stations. Specifically, the control device 140 includes a dynamic network construction module 210, a virtual edge visualization module 220, an equipment status management module 230, a cost calculation module 240, a route search module 250, and a data integration and communication module 260. The modules can communicate and exchange data and information through wired or wireless connections.

[0118] The dynamic network construction module 210 is used to establish a network structure that changes over time based on the location of the static device 112 and the real-time status of the movable device 114. The dynamic network construction module 210 can treat the static device 112 as a fixed node and construct a basic network based on the fixed connectivity relationships between the static devices 112. Simultaneously, the dynamic network construction module 210 can combine the actual location of the movable device 114 at different points in time to make the network reflect the actual connectivity currently possessed by the freight station 110, providing a basis for subsequent path planning. In some embodiments, the dynamic network construction module 210 can determine whether certain paths are reachable at the current time based on the nearest location and the most recent available time of the movable device 114, ensuring that the network structure remains consistent with the system state. The network structure generated by the dynamic network construction module 210 can be used by the virtual edge realization module 220 and the path search module 250.

[0119] The virtual edge visualization module 220 generates temporary connectivity paths based on the current position of the movable device 114. The virtual edge visualization module 220 can determine whether the movable device 114 can form a transport capability between two locations at a specific point in time. For example, when a conveyor is located on a certain track section, the movement between all nodes within its reach constitutes a temporarily usable path. Or, when a freight elevator is stopped at a certain floor, it can provide vertical transport capability between the stopped floor and other service floors. The virtual edge visualization module 220 can continuously track changes in these capabilities, enabling the dynamic network to reflect the actual transport connections available to the freight station 110 at each moment. In some embodiments, when generating a temporary connectivity path, the virtual edge visualization module 220 can also determine whether the path can be actually used at the current point in time based on the availability of the movable device 114. The temporary connectivity paths generated by the virtual edge visualization module 220 can be provided to the cost calculation module 240 and the path search module 250 for path segment feasibility assessment.

[0120] The equipment status management module 230 is used to maintain the latest status of the movable equipment 114. The equipment status management module 230 can record the most recent location, most recent available time, current idle status, whether it is performing a task, or whether it is in a fault state for each movable equipment 114. The equipment status management module 230 can provide other modules with judgment criteria to determine whether a specific movable equipment 114 can participate in path planning or form a virtual edge at a specific point in time, thereby generating a realistic and feasible transportation path. In some embodiments, the status information maintained by the equipment status management module 230 can be used by the cost calculation module 240 to assess the preparation time required for the movable equipment 114 to transition from its current state to perform a new transportation action.

[0121] The cost calculation module 240 generates a cost value reflecting the actual transportation time for each feasible path segment in the dynamic network. This cost value may include a movement cost describing the time it takes for goods to move along the path segment, and a preparation cost describing the time required for the movable device 114 to reach an executable state before performing the transportation segment. The definitions of movement cost and preparation cost used by the cost calculation module 240 correspond to the state-dependent edge cost model described in the aforementioned theoretical foundation section, where the total cost can be considered as a combination of movement cost and preparation cost. In some embodiments, the movement cost can be calculated by dividing the physical distance between nodes by the speed of the target movable device 114, while the preparation cost considers any delays caused by repositioning or waiting of the target movable device 114 before commencing cargo transfer.

[0122] The cost calculation module 240 can estimate the movement time required for goods to physically move along a path segment based on the physical distance between nodes, the type of equipment the path depends on, and the operating speed of the corresponding equipment. Simultaneously, the cost calculation module 240 can also estimate the preparation time required for the movable equipment 114 to perform this transportation segment based on its current status. This preparation time can be assessed based on factors such as the current location of the target movable equipment 114, its most recent availability time, whether the target movable equipment 114 is currently performing other tasks, the time required for the target movable equipment 114 to move to the path start point, and whether the target equipment needs to wait for idle time.

[0123] When calculating the cost of a path segment, the cost calculation module 240 integrates the travel time and preparation time, so that the cost reflects the total time required to complete the transportation of the target path segment under the current system state. The integrated cost value can be used as the edge weight in the path search process, enabling the path search module 250 to judge the merits of different path segments based on the combined magnitude of travel cost and preparation cost, and accordingly decide which path segments should be explored first or skipped in the dynamic network.

[0124] Since the availability, location, and operational status of movable equipment 114 change continuously over time, the cost calculation module 240 can consider the actual arrival time of goods on the route segment when generating route segment costs, and determine whether the target equipment can provide transportation capacity at that time, thus making the route segment cost time-dependent. When the target equipment can immediately perform transportation, the preparation time can approach zero, making the combined cost mainly determined by the movement time (preparation cost approaches zero); however, when the target equipment has not yet arrived at the route start point or is performing other tasks, the preparation time will increase significantly, leading to an increase in preparation costs, thereby increasing the overall cost of the route segment. The cost calculation module 240 can continuously update the cost estimate based on the latest status information provided by the equipment status management module 230, so that the cost value of each route segment can reflect the real transportation conditions in the dynamic network environment. The cost value output by the cost calculation module 240 can be used as the basis for the route search module 250 to determine the optimal path.

[0125] The path search module 250 is used to find the transportation route with the lowest transportation cost in the time-related network established by the above modules. The path search module 250 can calculate the earliest reachable time of each node at different time points step by step based on the combined transportation cost provided by the cost calculation module 240, and determine the optimal path from the origin to the destination accordingly.

[0126] In some embodiments, the path search module 250 may employ the time-dependent Dijkstra method (TD-Dij), which, during the search process, considers both the arrival time of goods at each node and the cost of the path segment at that time, ensuring that each path extension reflects the current system state and the availability of movable equipment 114. When a change in equipment state causes a virtual edge to become unavailable or its cost to increase, TD-Dij can also adjust the search direction in a timely manner based on the new combined cost. In some embodiments, the time-dependent TD-Dij method can perform path search by sequentially selecting the earliest reachable node, updating the arrival time of the path segments adjacent to that node, and marking the node as processed after the update.

[0127] In some embodiments, the path search module 250 may employ the time-dependent A* method (TD-A-Star). TD-A-Star, in addition to executing the search logic based on the earliest reachable time described above, further utilizes heuristics to predict the remaining transport time from the current node to the target node, enabling the search process to prioritize exploring path segments more likely to constitute the optimal solution. In time-dependent networks, heuristic estimation can be based on the shortest travel time under static conditions or other lower bound estimates that satisfy the acceptability principle, thereby reducing the search volume and increasing the path generation speed while maintaining optimality. In some embodiments, the time-dependent TD-A-Star method can perform path search by selecting the next expanded node based on heuristic estimation, calculating the arrival times of the adjacent path segments of that node, and adding the node to the open or closed set. Here, "based on heuristic estimation" means: using the estimated shortest travel time from the node to the target node under static conditions as the ranking criterion for prioritizing the expansion of the next candidate node, and selecting the expansion order of the path search accordingly.

[0128] In some embodiments, the path search module 250 may include a first method and a second method to meet the path planning requirements under different operational scenarios. The first method may be driven by the time-dependent Dijkstra method (TD-Dij), and the second method may be driven by the time-dependent A* method (TD-A-Star). The path search module 250 may select to use the first method or the second method based on the real-time computing resources of the freight station 110, the status of the movable equipment 114, or the task priority, and switch between the two methods when necessary to balance the accuracy and real-time performance of path planning. By combining the time-dependent extended logic of TD-Dij and TD-A-Star, the path search module 250 can generate a transportation path that is both adaptive and efficient in real time when the network structure is dynamically updated due to the movement and changes of the movable equipment 114.

[0129] In some embodiments, when calculating the cost of any path segment, the path search module 250 can use the arrival time corresponding to the starting point of the current path to the beginning of the path segment as the input for edge cost calculation, so that the path search can reflect the time dependence characteristics of edge cost.

[0130] The data integration and communication module 260 receives real-time device data from the sensor and data acquisition module 120 and shares this data among the modules within the control device 140. The data integration and communication module 260 can transmit the updated network structure, path cost results, and path planning results generated by the path search module 250 to the execution processor 150, enabling the movable device 114 to perform actual transportation according to the latest path. After the task is completed, the data integration and communication module 260 can receive the task status feedback and status changes caused by device movement from the execution processor 150, and integrate this information into a new system state. This allows the dynamic network construction module 210, virtual edge realization module 220, device status management module 230, and path search module 250 to update the network topology, virtual edge availability, device availability, and path feasibility based on the latest system state, thereby enabling the control device 140 to enter the next round of path planning, forming a continuous closed-loop optimization mechanism. In other words, the data integration and communication module 260 can use the integrated system state as input to the aforementioned modules to trigger a new round of network updates and path calculations. In some embodiments, the data integration and communication module 260 can be connected to an existing terminal management system, enabling the dynamic path planning function of the present invention to work in conjunction with the scheduling system, monitoring system or equipment management system of the freight station 110, adapting to the equipment layout and operation process of different freight stations 110.

[0131] Through the coordinated operation of the above modules, the control device 140 can continuously maintain an accurate network model in the highly dynamic environment of the automated freight station and calculate the optimal path in real time, so that goods can be transported in a highly efficient manner in the complex environment of the freight station 110.

[0132] System 100 may also include a cost parameter adaptive tuning module 160 and a predictive analysis module 170.

[0133] The cost parameter adaptive tuning module 160 dynamically adjusts the weight parameters in the combined cost function based on the actual transportation performance of goods in the freight station 110. The cost parameter adaptive tuning module 160 continuously monitors the actual transportation time of each transportation route, the waiting time of goods in different sections, and the response time of movable equipment 114 under high load conditions, comparing this feedback information with the estimated cost output by the cost calculation module 240. When monitoring results show a deviation between the actual transportation time and the estimated value for certain route segments, the cost parameter adaptive tuning module 160 can adjust the weights of movement costs and preparation costs, ensuring that the combined cost function reflects transportation costs closer to actual operation under different load conditions, thereby improving the overall accuracy of route planning and system efficiency. In some embodiments, the cost parameter adaptive tuning module 160 can periodically or trigger updates when changes in operating status are detected, enabling the system to continuously self-optimize according to the operating conditions of the freight station 110.

[0134] The predictive analysis module 170 is used to predict the availability and potential delays of movable equipment 114 in the future time period based on historical operating data, equipment status information, and real-time trends. The predictive analysis module 170 can use statistical models, rule models, or machine learning models to analyze information such as equipment movement patterns, task execution times, and the frequency of congestion points to estimate potential equipment occupancy conflicts, response delays, or path blockages. The predictive analysis module 170 can provide the prediction results to the path search module 250, enabling the path search module 250 to proactively avoid potentially congested path segments or unavailable equipment resources before delays occur, thereby reducing the number of path replannings and improving overall operating efficiency. In some embodiments, the predictive analysis module 170 can be linked with the equipment status management module 230, so that the predicted availability of equipment in the future can be used as one of the bases for determining path feasibility.

[0135] During the operation of system 100, the functional modules within control device 140 can work collaboratively according to a predetermined data flow to continuously generate optimal paths that conform to the current operating state of freight station 110. When system 100 starts, dynamic network construction module 210 can establish a basic network structure based on the fixed layout of static equipment 112 and the initial state of movable equipment 114. As sensors and data acquisition module 120 continuously transmit changes in device position and status, virtual edge visualization module 220 can generate or update temporary connectivity paths formed by movable equipment 114, enabling the network structure to reflect the real-time reachability of freight station 110.

[0136] When system 100 receives a new transport task or detects a change in network structure, device status management module 230 can provide the latest status information of movable device 114, including whether it is currently idle, the time of its most recent task execution, and its reachability. Cost calculation module 240 can combine this status information with the dynamic network structure to recalculate the transport cost of each path segment in the network to obtain the comprehensive cost of the required travel time and preparation time under the current conditions. For example, cost calculation module 240 can calculate the corresponding cost for the dynamic path segment formed by movable device 114 based on the temporary connectivity path generated by virtual edge realization module 220.

[0137] The path search module 250 can perform time-related path planning based on the path segment costs output by the cost calculation module 240 to find the optimal transportation path from the origin to the target node. The path search module 250 can choose to use either the first time-related method (Dijkstra's method) or the second time-related method (A* method), or switch between the two methods, depending on system requirements. In some embodiments, when the predictive analysis module 170 is present, the path search module 250 can also incorporate predictions of future equipment availability into its judgment, enabling the system to avoid potential bottlenecks or delays in advance.

[0138] The data integration and communication module 260 can transmit the route planning results generated by the route search module 250 to the execution processor 150, enabling the movable device 114 to perform transportation tasks based on the latest route. During task execution, the execution processor 150 continuously reports on device movement, task progress, and abnormal conditions. The data integration and communication module 260 can integrate this information into a new system state and send it back to the dynamic network construction module 210, the virtual edge realization module 220, and the device state management module 230 for the next round of route modeling and cost calculation. In some embodiments, when the cost parameter adaptive tuning module 160 detects a deviation between transportation performance and estimated cost, it can dynamically adjust the weight of the combined cost function, enabling the system 100 to maintain accurate route cost assessment even under high load or special operating environments. In this way, the system 100 can form a closed-loop mechanism of continuous updating and self-optimization.

[0139] In summary, the dynamic path planning system proposed in this invention achieves significantly better technical results than traditional static path planning methods in the highly time-varying and complex environment of automated freight stations. Firstly, the dynamic network modeling mechanism employed in this invention represents the freight station as a network structure that changes continuously over time. Static equipment forms fixed nodes, while movable equipment expands network connectivity in real time as dynamic virtual edges. Through this modeling approach, the system can accurately reflect the real-time location and availability of movable equipment, ensuring that path planning truly corresponds to the actual operating state of the freight station and significantly improving overall transportation efficiency.

[0140] Secondly, the combined cost function introduced in this invention integrates movement costs and preparation costs, enabling it to simultaneously capture the basic movement time required for goods to move between devices and the delay time required for movable equipment to prepare for transportation from its current state. This cost structure can quantify factors such as equipment scheduling delays, waiting times, and task occupancy that traditional methods cannot handle, making the transportation cost of each route segment closer to real transportation conditions, thereby improving the accuracy and rationality of route selection.

[0141] Furthermore, by employing time-dependent path search methods, including the time-dependent Dijkstra's improved method and the time-dependent A* method, the system can promptly calculate the optimal path best suited to the current conditions based on real-time updated network structure and cost information. In particular, the time-dependent A* method reduces the search range through heuristic estimation, significantly improving computational efficiency while maintaining the optimal solution, making it more suitable for automated freight transport environments requiring real-time response.

[0142] Finally, by integrating real-time data from sensors, IoT devices, and the terminal management system, this invention can continuously update the network model, equipment status, and path cost parameters, ensuring that the path planning results always reflect the latest operational status. Combined with optional cost parameter adaptive tuning mechanisms and equipment availability prediction mechanisms, the system can maintain stable performance under high load or congestion conditions and proactively avoid potential delays, thereby achieving higher cargo flow efficiency and more resilient overall operational performance.

[0143] Based on the above technical features, the present invention can achieve accurate, efficient and adaptive path planning in complex, dynamic and high-throughput automated freight station environments, significantly improving the shortcomings of traditional technologies in terms of inaccurate paths, inability to reflect equipment status, and inability to handle time-related delays.

[0144] According to the embodiments disclosed herein, the functional units and modules of the system may be implemented using computing devices, computer processors, or electronic circuits, including but not limited to application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), microcontrollers, and other programmable logic devices, which may be configured or programmed in accordance with the teachings of this disclosure. Those skilled in the art of software or electronics can readily write computer instructions or software code that run on computing devices, computer processors, or programmable logic devices based on the teachings of this disclosure.

[0145] According to these embodiments, all or part of the method can be executed in one or more computing devices, including servers, personal computers, laptops, and mobile computing devices such as smartphones and tablets.

[0146] These embodiments may include computer storage media, transient and non-transient storage devices, wherein computer instructions or software code are stored, which can be used to program or configure computing devices, computer processors, or electronic circuits to perform any of the processes of the present invention. The storage media, transient and non-transient storage devices may include, but are not limited to, floppy disks, optical disks, Blu-ray discs, DVDs, CD-ROMs, magneto-optical disks, ROMs, RAMs, flash memory devices, or any medium or device suitable for storing instructions, code, and / or data.

[0147] According to various embodiments, each functional unit and module may also be implemented in a distributed computing environment and / or cloud computing environment, wherein all or part of the machine instructions are executed in a distributed manner by one or more processing devices interconnected by a communication network, such as an intranet, a wide area network (WAN), a local area network (LAN), the Internet, and other forms of data transmission media.

[0148] The above description of the present invention is for illustrative purposes only and is not intended to be exhaustive or to limit the invention to the specific forms disclosed. Those skilled in the art can make numerous modifications and variations to the present invention.

[0149] These embodiments were chosen and described in order to better explain the principles of the invention and its practical application, thereby enabling those skilled in the art to understand the various embodiments of the invention and the various modifications suitable for particular purposes.

Claims

1. A control device connected to a system for dynamic path planning in an automated freight station, characterized in that, include: The dynamic network construction module is used to build a network structure that changes over time based on the location of static devices and the real-time status of movable devices. The virtual edge realization module is used to generate temporary connectivity paths based on the current position of the movable device and update the dynamic virtual edges in the network structure. The equipment status management module is used to maintain the current location, available time, task status and real-time operating parameters of the movable equipment in order to obtain the equipment status of the movable equipment. The cost calculation module is used to calculate the combined cost of each path segment based on the network structure, the dynamic virtual edge, and the device status, which includes movement cost and preparation cost. The path search module is used to perform path search in a time-related network based on the combined cost to obtain the optimal path; as well as The data integration and communication module is used to receive real-time device data and integrate system status, and provide the system status to the dynamic network construction module, the virtual edge realization module, the device status management module and the path search module, and transmit the optimal path to the execution processor, so that the execution processor drives the movable device to perform the transportation task.

2. The control device according to claim 1, wherein the dynamic network construction module is further used to represent the static device as a fixed node, and to construct the basic configuration of the network structure based on the static connectivity relationship.

3. The control device according to claim 2, wherein the movable equipment includes a conveyor vehicle and a cargo elevator, and the virtual edge realization module is further configured to generate a horizontal connectivity path based on the horizontal reachability of the conveyor vehicle and update the network structure thereon, and generate a vertical connectivity path based on the service floor of the cargo elevator and update the network structure thereon.

4. The control device according to claim 1, wherein the device status management module is further used to record the most recent location of the movable device, the most recent available time, whether it is in an idle state, and whether it is performing a task.

5. The control device according to claim 1, wherein the cost calculation module is further configured to calculate the movement cost by combining the physical distance between nodes with the operating speed of the movable device.

6. The control device according to claim 5, wherein the cost calculation module is further configured to obtain the preparation cost based on the current position of the movable device, the most recent available time, whether it is performing other tasks, and the time required to move to the starting point of the path.

7. The control device according to claim 6, wherein the movement cost is calculated by the ratio of the distance between nodes to the operating speed of the movable device, and the preparation cost is calculated by the sum of the time required for the movable device to move to the starting point of the path segment and the available waiting time.

8. The control device according to claim 1, wherein the path search module employs and executes a time-related first method, which performs path search by sequentially selecting the earliest reachable node, updating the arrival time of its adjacent path segments, and marking the node as processed.

9. The control device according to claim 1, wherein the path search module employs and executes a time-related second method, which performs path search by selecting the next expansion node based on heuristic estimation, updating the arrival time of adjacent path segments, and managing open and closed sets, wherein the heuristic estimation is based on the shortest travel time from the current node to the target node under static conditions, which serves as the sorting basis for prioritizing expansion to the next candidate node, and the expansion order of the path search is selected accordingly.

10. A system for dynamic path planning in automated freight stations, characterized in that, include: Freight stations, including: Multiple static devices are used to provide a cargo transport path at a fixed location; Multiple movable devices are used to perform horizontal or vertical cargo transport operations within the freight station; Sensors and data acquisition modules are used to obtain the current position, available time, operating status and real-time operating parameters of the movable device; Control device, including: A dynamic network construction module is used to construct a network structure that changes over time based on the location of the static device and the real-time state of the movable device. The virtual edge realization module is used to generate temporary connectivity paths based on the current position of the movable device and update the dynamic virtual edges in the network structure. The equipment status management module is used to maintain the current location, available time, task status and real-time operating parameters of the movable equipment in order to obtain the equipment status of the movable equipment. The cost calculation module is used to calculate the combined cost of each path segment based on the network structure, the dynamic virtual edge, and the device status. The combined cost includes movement cost and preparation cost. A path search module is configured to perform a path search in a time-dependent network to obtain an optimal path based on the combined cost; and The data integration and communication module is used to receive real-time device data and integrate system status, and provide the integrated system status to the dynamic network construction module, the virtual edge visualization module, the device status management module and the path search module; An execution processor, wherein the data integration and communication module transmits the optimal path to the execution processor, causing the execution processor to drive the movable device to perform a corresponding transportation task according to the optimal path, and to send the task execution status back to the control device; and A communication and data transmission network is used to transmit data between the sensors and data acquisition module, the control device and the execution processor.

11. The system according to claim 10 further includes a cost parameter adaptive tuning module for dynamically adjusting the weights of the movement cost and the preparation cost in the combined cost based on the actual transportation performance of the goods in the freight station, wherein the cost parameter adaptive tuning module updates the weight parameters based on the difference between the actual transportation time of each route segment and the estimated cost output by the cost calculation module.

12. The system of claim 10 further includes a predictive analysis module, which uses a machine learning model to analyze the device movement pattern of the movable device, thereby predicting the availability and potential delay of the movable device in a future time period based on historical operating data, device status information and real-time trends, wherein the predictive analysis module provides the prediction results to the path search module so that the path search module can actively adjust the path search before the path becomes unavailable or a delay occurs.

13. In the system according to claim 10, the data integration and communication module is further configured to receive the task execution status and integrate it into an updated system status after the execution processor completes the transportation task, so that the dynamic network construction module, the virtual edge realization module, the device status management module and the path search module can perform the next round of path planning based on the updated system status, thereby forming a closed-loop path planning optimization mechanism.

14. A method for dynamic path planning in automated freight stations, characterized in that, include: Acquire real-time operating data of movable equipment within the freight station, wherein the real-time operating data includes the current location, available time, task status, and operating parameters of the movable equipment; Based on the location of the static device and the real-time operating data of the movable device, a network structure that changes over time is constructed, wherein the static device is used as a fixed node, and dynamic virtual edges are generated according to the current location of the movable device. Maintain the device status of the movable device, including recording the most recent location, most recent available time, idle status and task execution status of the movable device; Calculate the combined cost for each path segment in the network structure, wherein the combined cost includes movement cost and preparation cost, the movement cost is used to describe the time it takes for goods to physically move along the path segment, and the preparation cost is used to describe the time required for the movable equipment to prepare to execute the path segment; A time-related method is used to perform path search based on the combined cost to obtain the optimal path from the starting point to the destination; as well as A transport instruction is generated based on the optimal route, and the movable equipment in the freight station executes the transport instruction.

15. The method of claim 14, wherein the movable device comprises a conveyor vehicle and a cargo elevator, and the method further comprises: A horizontal virtual connectivity path is generated based on the horizontal reachability of the transport vehicle, and a vertical virtual connectivity path is generated based on the service floor of the cargo elevator, so as to update the dynamic virtual edges in the network structure.

16. The method of claim 14, further comprising: The movement cost is calculated by dividing the physical distance between nodes by the operating speed of the corresponding movable device. as well as The preparation cost is calculated based on the sum of the time required for the movable equipment to move to the starting point of the path segment and the available waiting time.

17. The method of claim 14, further comprising: A time-related first method is used to perform path search, the first method including sequentially selecting the earliest reachable node, updating the arrival time of the adjacent path segments of the node, and marking the node as processed.

18. The method of claim 14, further comprising: A time-related second method is used to perform path search. The second method includes selecting expansion nodes based on heuristic estimation, updating the arrival times of adjacent path segments, and managing open and closed sets. The heuristic estimation is based on the shortest travel time from the current node to the target node estimated under static conditions to determine the node expansion order.

Images