A logistics sorting, package collecting and transmitting and end access system and method based on cross-space dynamic mapping

Abstract

This invention provides a logistics sorting, package collection and transportation, and last-mile storage and retrieval system and method based on cross-spatial dynamic mapping, belonging to the field of logistics technology. It solves the technical problems of insufficient flexibility and fragile information links in existing logistics systems. This logistics sorting, package collection and transportation, and last-mile storage and retrieval system based on cross-spatial dynamic mapping includes a central dynamic addressing layer deployed in the cloud for real-time maintenance of the occupancy status matrix of all physical storage bins in the last-mile station; a sorting center mapping execution layer located below the sorting line, containing "streetlight"-style multi-dimensional sensing units and micro-thermal labeling units; and a last-mile station execution layer containing multiple storage bins with unique physical codes and an integrated labeling and printing machine. The storage bins are used to hold packages and correspond one-to-one with physical bin IDs. This invention has the advantages of dynamically decoupling sorting decisions from physical containers and ensuring the reliability of information throughout the entire chain.

Description

Technical Field

[0001] This invention belongs to the field of logistics technology and relates to a logistics sorting, package collection and transportation and last-mile storage system, particularly a logistics sorting, package collection and transportation and last-mile storage system and method based on cross-space dynamic mapping. Background Technology

[0002] In the existing logistics system, the sorting process and the parcel storage and retrieval process at last-mile delivery stations operate independently. Sorting machines typically assign parcels to fixed physical slots, each pre-assigned to a destination or delivery station. This static binding method has at least the following drawbacks: Insufficient flexibility: The number of compartments is limited and fixed, making it impossible to dynamically adjust the allocation strategy based on the real-time status of the parcels at the station. When the number of parcels at a station surges suddenly, the corresponding compartments may overflow, while other compartments remain idle.

[0003] Unable to aggregate packages: When the same recipient has multiple packages at the same station, the existing system has difficulty allocating them to nearby storage bins at the sorting source, causing users to have to search for them in multiple locations when picking up their packages, resulting in a poor user experience.

[0004] The information chain is fragile: After sorting, packages typically rely on a single printed waybill or label to establish their connection with the target container. If the label is damaged or detached during transport, the last-mile personnel can only relocate it manually, which is inefficient and prone to errors.

[0005] Uneven utilization of basket spaces: Static allocation does not take into account the load-bearing capacity, historical popularity, and ease of operation of the basket spaces, which may result in some basket spaces being overloaded, some being idle, or inconvenient basket spaces at high or low positions being used frequently.

[0006] Therefore, there is an urgent need for a system and method that can dynamically decouple sorting decisions from physical containers and ensure the reliability of information throughout the entire chain. Summary of the Invention

[0007] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a logistics sorting, package transport, and end-of-line storage system and method based on cross-space dynamic mapping. The technical problem to be solved by this invention is: how to dynamically decouple sorting decisions from physical containers and ensure the reliability of information throughout the entire chain.

[0008] The objective of this invention can be achieved through the following technical solutions: A logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping includes: A central dynamic addressing layer, deployed in the cloud, is used to maintain the occupancy status matrix of all physical storage bins in the last-mile delivery stations in real time. The occupancy status matrix records the current status of each storage bin and the unique identifier of the package it is bound to. The occupancy status includes at least an idle state, a pre-occupied state, and a fully occupied state. The central dynamic addressing layer is configured to: respond to a package sorting request, run a multi-constraint greedy addressing algorithm to dynamically allocate the target physical bin ID, establish a dynamic mapping relationship between the package unique identifier and the target physical bin ID, and persistently store this mapping relationship; while establishing the mapping relationship, use atomic operations to update the status of the selected bin in the occupancy status matrix from idle to pre-occupied. The sorting center mapping execution layer, located below the sorting line, includes a "streetlight"-style multi-dimensional sensing unit and a micro-thermal labeling unit. The "streetlight"-style multi-dimensional sensing unit is an inverted U-shaped structure spanning above the sorting conveyor line, integrating a vision sensor and an infrared grating. It is configured to collect waybill information the instant the package leaves the sorting chute. When the vision sensor fails to capture a clear image, the system uses the package's timestamp recorded by the infrared grating to retrieve the corresponding waybill number from the chute allocation record of the sorting system, ensuring reliable acquisition of waybill information. The micro-thermal labeling unit is synchronized with the sensing unit via an industrial communication network. Within the same communication cycle, based on the target physical basket ID issued by the central dynamic addressing layer, it synchronously prints and attaches auxiliary mapping labels, affixing the labels to the non-waybill area of ​​the package surface. The last-mile delivery station execution layer includes multiple storage baskets with unique physical codes and a labeling and printing machine. The storage baskets are used to hold packages and correspond one-to-one with physical basket IDs. The labeling and printing machine is configured to, when a package arrives at the station, if the auxiliary mapping label attached to its surface is missing or unreadable, query the central dynamic addressing layer by scanning the original package tracking number to obtain the dynamic mapping relationship corresponding to that tracking number, acquire the assigned target physical basket ID, and regenerate the auxiliary mapping label on-site. When reprinting the label, the labeling and printing machine verifies the validity of the dynamic mapping relationship with the central dynamic addressing layer, and only allows the reprinting operation if the mapping relationship is valid. After reprinting, the original label mapping relationship corresponding to the package is marked as used, and the system rejects secondary basket binding requests based on the used mapping relationship.

[0009] Preferably, the multi-constraint greedy addressing algorithm includes a hard constraint pruning layer and a multi-dimensional weighted scoring layer; The hard constraint pruning layer is used to sequentially perform the following filtering from all available baskets at the current target station to obtain candidate set C1: Weight constraints: Remove baskets whose package weight w exceeds the maximum load capacity W_max. Size constraints are applied to exclude baskets whose three-dimensional dimensions (l, d, h) exceed the basket's capacity (L, D, H), where l, d, and h are the length, width, and height of the package, respectively, and L, D, and H are the length, width, and height of the basket. Lock constraints and remove baskets marked as under maintenance or reserved. If the candidate set C1 is empty after trimming, the no-solution exception handling process is triggered, and a manual intervention instruction is sent to the sorting center. The multi-dimensional weighted scoring layer is used to calculate a weighted score (Score) for each candidate basket in the candidate set C1. The scoring function is: ; Wherein, α, β, γ, and δ are configurable weight coefficients, with default values ​​of 0.3, 0.2, 0.1, and 0.4, respectively; Score_dist is the aggregation score, reflecting the degree of package aggregation; it is defined as: when the same recipient already has packages in a pre-occupied or occupied state at the target station, obtain the set of coordinates of the baskets where these existing packages are located. Let the coordinates of the candidate basket position be (x, y, z), where x, y, and z are the column coordinates, height coordinates, and depth coordinates of the candidate basket position, respectively, and x_q, y_q, and z_q are the column coordinates, height coordinates, and depth coordinates of the existing basket position, respectively. Then:

[0010] ;

[0011] The smaller the value, the higher the physical aggregation with the assigned package; if the recipient currently has no other packages, then Score_dist = 0; Score_weight is the weight fit score, defined as: Score_weight = (w / W_max)×100. The smaller the value, the better the weight fit. Score_heat is the thermal balance score, defined as: Score_heat=f / T, where f is the historical usage frequency of the basket within the preset time window T, and T is the length of the preset time window. The smaller the value, the more idle the basket is recently. Score_ergo is the ergonomic score, defined as: Score_ergo=|y-H_center|, where y is the height coordinate of the candidate basket position, and H_center is the center value of the preset optimal operating height interval [H_min, H_opt], ​​H_center= (H_min + H_opt) / 2, the smaller the value, the closer the height is to the optimal operating center; The system selects the basket with the smallest score value as the target physical basket ID.

[0012] Preferably, the central dynamic addressing layer also maintains a time sliding window Δt. When n packages (n>1) belonging to the same recipient are detected to request basket allocation in sequence within Δt, an aggregation mode is triggered. In the aggregation mode, for the second and subsequent packages of the same recipient, the weight coefficient α of the aggregation degree score is temporarily amplified to α', α' = α×K, where K is an aggregation gain factor greater than 1, with a default value of 6.67. This allows subsequent packages to be preferentially allocated to baskets that are closer to the Manhattan distance of the baskets where the already allocated packages are located, thereby strengthening the spatial clustering of packages belonging to the same recipient.

[0013] Preferably, the size constraint further includes: if the package type is marked as large or irregular, in addition to checking the basic three-dimensional dimensions, it is also necessary to check whether the longest side of the package satisfies the redundancy space condition of 2×max(L,D,H), where max(L,D,H) represents the maximum value among the three dimensions of the basket position: length, width, and height; if the redundancy space condition is not met, the basket position is removed from the candidate set.

[0014] Preferably, the dynamic mapping relationship is time-sensitive, and its state transition logic is uniformly controlled by the central dynamic addressing layer. Specifically, the idle state is converted to the pre-occupied state by the labeling event, the pre-occupied state is converted to the occupied state by the basket entry confirmation event, and the occupied state is converted to the idle state by the item retrieval confirmation event. When a basket is in the pre-occupied state for more than a preset time T_timeout and has not been converted to the occupied state, the central dynamic addressing layer actively pushes a verification and reprint task to the labeling and printing all-in-one machine of the terminal station execution layer. Here, T_timeout is the pre-occupied timeout threshold.

[0015] Preferably, the physical storage basket has a built-in gravity sensor; the central dynamic addressing layer is configured to automatically trigger a pickup notification to the recipient when the gravity sensor detects that the total weight of the items in the basket reaches a preset weight threshold W_notify, where W_notify is the weight threshold that triggers the notification.

[0016] Preferably, the atomic operation is implemented through a distributed lock mechanism, with the lock granularity being a single basket position. Before performing a state change on the target basket position, the distributed lock of the basket position is acquired first, and the state is updated to pre-occupied only after verifying that the state is still idle under the protection of the lock. This ensures that the same idle basket position will not be allocated to two different packages at the same time in a high-concurrency sorting scenario.

[0017] A logistics sorting, package collection and transportation, and last-mile storage method based on cross-space dynamic mapping includes the following steps: S1. At the sorting center, the "streetlight" type multi-dimensional sensing unit collects the waybill information the instant the package leaves the sorting chute. If the visual sensor captures a clear image, the waybill number is obtained. If the visual sensor fails to capture a clear image, the waybill number is retrieved from the chute allocation record of the sorting system using the timestamp recorded by the infrared grating. Based on the obtained waybill number, the cloud-based central dynamic addressing layer determines the target end station and runs a multi-constraint greedy addressing algorithm to allocate the target physical basket ID: First, hard constraint clipping is performed from the available baskets at the target station, including weight constraints w≤W_max, size constraints l≤L and d ≤D and h ≤ H, and locking constraints to exclude baskets under maintenance. For packages marked as large or irregularly shaped, the size constraint must also check whether the longest side of the package satisfies the redundancy space condition of 2×max(L, D, H). For the candidate set that passes the hard constraint clipping, a weighted score is calculated according to the following formula: ; Select the basket with the smallest score as the target physical basket ID, update the basket status from idle to pre-occupied through distributed lock atomic operations, and establish a dynamic mapping relationship between the unique identifier of the package and the basket ID. S2. Within the same communication cycle, the micro-thermal labeling unit prints an auxiliary mapping label containing the target physical basket ID issued by the central dynamic addressing layer, and immediately attaches it to the non-sheet area of ​​the package surface to complete the physical-information mapping marking. S3. The parcel is transported to the corresponding terminal station via the collection container. The operator puts the parcel into the storage basket with the corresponding unique physical code according to the basket ID on the auxiliary mapping label, and completes the warehousing. S4. When a parcel is found to have a missing or unreadable auxiliary mapping label at the station, the original waybill number of the parcel is scanned by the end-point labeling and printing all-in-one machine. The system retrieves the dynamic mapping relationship corresponding to the waybill number from the central dynamic addressing layer and verifies the validity of the mapping. Only when the mapping relationship is valid is a new auxiliary mapping label allowed to be reprinted. After the reprint is completed, the old label mapping relationship corresponding to the parcel is marked as used, and the system rejects any secondary basket binding requests based on the used mapping relationship.

[0018] Preferably, in step S1, when the central dynamic addressing layer detects multiple packages for the same recipient within a preset time window Δt, an aggregation mode is triggered. When calculating the score, the aggregation degree score weight coefficient α is temporarily amplified, so that the subsequent package basket allocation results are preferentially clustered in the adjacent area of ​​the already allocated baskets. At the same time, when the visual sensor fails to capture a clear image, the waybill number is retrieved from the sorting system using the timestamp recorded by the infrared grating to ensure the completeness of the waybill information collection, thereby continuously completing the labeling and mapping establishment.

[0019] Preferably, a delivery confirmation identifier is set up at the terminal station. After the deliveryman completes the delivery of the package, he / she scans the delivery confirmation identifier to generate a confirmation signal and sends it to the central dynamic addressing layer. After receiving the confirmation signal, the central dynamic addressing layer triggers the sending of a pickup notification to the recipient.

[0020] Compared with existing technologies, this logistics sorting, package collection and transportation, and last-mile storage and retrieval system and method based on cross-space dynamic mapping has the following advantages: By using cross-space dynamic mapping, the rigid binding between sorting decisions and physical bin locations is removed, significantly improving system flexibility and bin resource utilization; multi-constraint greedy algorithms and aggregation modes achieve spatial aggregation of packages from the same recipient while ensuring feasibility, improving user experience; dual-mode perception and tag re-printing mechanisms build full-link information redundancy and self-healing capabilities from sorting to storage and retrieval, significantly reducing the cost of handling abnormal items; the introduction of dimensions such as weight and ergonomics makes bin allocation safer and more human-centered. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention.

[0022] Figure 2 This is a schematic diagram of the structure and workflow of the sorting center's mapping execution layer.

[0023] Figure 3 This is a schematic diagram of the execution layer of the terminal station.

[0024] Figure 4 This is a schematic diagram of the state transition logic for the basket position.

[0025] Figure 5 This is a flowchart illustrating a multi-constraint greedy addressing algorithm.

[0026] Figure 6 A diagram showing the comparison of the aggregation and distribution effects of packages for the same recipient. Detailed Implementation

[0027] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0028] Example 1

[0029] like Figures 1-6 As shown, this logistics sorting, package transport, and last-mile storage system based on cross-space dynamic mapping includes:

[0030] The central dynamic addressing layer, deployed in the cloud, is used to maintain the occupancy status matrix of all physical storage bins in the last-mile delivery stations in real time. The occupancy status matrix records the current status of each storage bin and the unique identifier of the package it is bound to. The occupancy status includes at least three states: idle, pre-occupied, and occupied. The idle state indicates that the bin is not currently associated with any package and can be reassigned. The pre-occupied state indicates that the bin has been reserved by a package in the sorting process through central scheduling, but the package has not yet been actually placed in the bin. The occupied state indicates that the package has been physically placed in the bin, and the bin and the package are bound together. This three-state design decouples information flow from physical flow, allowing for logical reservation of basket slots before packages arrive at the delivery station, significantly improving system throughput. The central dynamic addressing layer is configured to: respond to package sorting requests by running a multi-constraint greedy addressing algorithm to dynamically allocate target physical basket slot IDs, establishing a dynamic mapping between the package's unique identifier and the target physical basket slot ID, and persistently storing this mapping relationship, for example, by writing it to a distributed database with a timestamp and version number, to support subsequent end-to-end traceability and status verification; simultaneously, by atomically updating the selected basket slot's status in the occupancy status matrix from idle to reserved. These atomic operations can be implemented using Redis's Lua scripts or the database's CAS (Compare-And-Swap) primitive, ensuring that status changes are indivisible under concurrent requests, thereby preventing overselling of basket slots. The sorting center mapping execution layer, located on the lower section of the sorting line, includes a "streetlight"-style multi-dimensional sensing unit and a micro-thermal labeling unit. The "streetlight"-style multi-dimensional sensing unit is arranged in an inverted U-shape above the sorting conveyor line. Its inverted U-shaped support spans the conveyor belt, with visual sensors and infrared gratings integrated on the inner sides of the two side columns. A supplementary lighting and ranging module, also integrating a visual sensor and infrared grating, can be added to the top. This module is configured to collect waybill information the instant the package leaves the sorting chute. Specifically, when the package slides from the chute into the free-fall stage of the connecting conveyor belt or at the moment of falling, the visual sensor attempts to acquire a waybill image at high frequency (e.g., 60 frames per second) and uses OCR to identify the waybill number. If the visual sensor fails to capture a clear image, for example, if the waybill is obscured, damaged, or the package's posture is abnormal and cannot be identified, the system uses the timestamp recorded by the infrared grating to retrieve the corresponding waybill number from the chute allocation record of the sorting system, ensuring reliable acquisition of waybill information. The infrared grating consists of a row of through-beam infrared emitting and receiving tubes. When the package cuts off the light path, it generates a pulse signal. The control system records the event time with microsecond precision and associates it with the chute number, logical grid number, and time sequence assigned by the sorting machine, thereby accurately tracing the waybill identity and forming a dual-mode acquisition mechanism of "vision as the primary method and grating as a backup." The micro-thermal labeling unit and the sensing unit are synchronized through an industrial communication network (such as EtherCAT or Profinet). Within the same communication cycle, based on the target physical basket ID issued by the central dynamic addressing layer, the auxiliary mapping label is printed and attached synchronously. The label is pasted on the non-label area of ​​the package surface, such as the side, back, or blank space next to the label, avoiding covering the original label barcode and recipient information. This auxiliary mapping label can contain plain text numbers of the basket ID, a one-dimensional barcode or a two-dimensional barcode, and intuitive information such as the station code, facilitating subsequent manual or semi-automatic reading. The last-mile delivery station execution layer includes multiple storage baskets with unique physical codes and a labeling and printing machine. The storage baskets hold packages and correspond one-to-one with physical basket IDs. The physical codes can be laser-etched metal nameplates or high-strength RFID tags, ensuring long-term identification even in the complex lighting and dusty environment of the delivery station. The labeling and printing machine is configured to, when a package arrives at the station, if the auxiliary mapping label attached to its surface is missing or unreadable (e.g., due to label detachment caused by transport friction or fading of thermal paper), scan the original waybill number of the package to query the central dynamic addressing layer for the dynamic mapping relationship corresponding to that waybill number, obtain the assigned target physical basket ID, and regenerate the auxiliary mapping label on-site. When reprinting labels, the labeling and printing machine verifies the validity of the dynamic mapping relationship with the central dynamic addressing layer, and only allows reprinting if the mapping relationship is valid. The valid states include pre-occupied state and actual occupied state. If the mapping has been rolled back due to timeout or manual cancellation, reprinting will be refused and an alarm will be issued. After reprinting is completed, the original tag mapping relationship corresponding to the package will be marked as used. The system will refuse secondary basket binding requests based on the used mapping relationship, thereby preventing multiple deliveries of one order and basket misalignment caused by tag copying or replay attacks.

[0031] The multi-constraint greedy addressing algorithm includes a hard constraint pruning layer and a multi-dimensional weighted scoring layer. The algorithm is deployed in the microservice of the central dynamic addressing layer. Each time an allocation call is made, it quickly reads the current station basket status snapshot from memory or cache to complete the layer-by-layer filtering and scoring sorting. The hard constraint pruning layer is used to sequentially perform the following filtering from all available baskets at the current target station to obtain candidate set C1: Weight constraints are applied to exclude baskets whose package weight w exceeds the maximum load capacity W_max. Weight data can be obtained from dynamic scales on the sorting line or upstream systems. If no weight data is collected, the default maximum value is conservatively applied, and filtering is based solely on size constraints. Size constraints are applied to reject baskets whose three-dimensional dimensions (l, d, h) exceed the basket's capacity (L, D, H), where l, d, and h are the length, width, and height of the package, and L, D, and H are the length, width, and height of the basket. The system allows packages to be slightly smaller than the basket's volume, but any package exceeding this limit will be rejected to ensure that it can be completely placed inside the basket and prevent it from protruding and tipping over. Lock constraints and remove baskets marked as "under maintenance" or "reserved". Baskets in the "under maintenance" state can be locked by the station administrator via a handheld terminal, while those in the "reserved" state are reserved by the system for VIP customers, temporary storage of large items, or emergency buffering. If the candidate set C1 is empty after trimming, the no-solution exception handling process is triggered, and a manual intervention instruction is sent to the sorting center. For example, an alarm is popped up on the sorting operation screen, an audio-visual prompt is triggered, and the sorting of the package is temporarily suspended. The manual confirmation is made as to whether the station needs to be changed or the basket position attribute needs to be temporarily adjusted. The multi-dimensional weighted scoring layer is used to calculate a weighted score (Score) for each candidate basket in the candidate set C1. The scoring function is: ; Wherein, α, β, γ, and δ are configurable weight coefficients, with default values ​​of 0.3, 0.2, 0.1, and 0.4, respectively; Score_dist is the aggregation score, reflecting the degree of package aggregation; it is defined as: when the same recipient already has packages in a pre-occupied or occupied state at the target station, obtain the set of coordinates of the baskets where these existing packages are located. Let the coordinates of the candidate basket position be (x, y, z), where x, y, and z are the column coordinates, height coordinates, and depth coordinates of the candidate basket position, respectively, and x_q, y_q, and z_q are the column coordinates, height coordinates, and depth coordinates of the existing basket position, respectively. Then:

[0032] ;

[0033] A smaller value indicates a higher degree of physical aggregation with already assigned packages, meaning a smaller Manhattan distance, which is more conducive to recipients picking up packages in one go. If the recipient currently has no other packages, then Score_dist = 0, in which case the score for this dimension is zero, and it has no biased effect. Score_weight is the weight matching score, defined as: Score_weight = (w / W_max) × 100. The smaller the value, the better the weight matching, which can prioritize the allocation of heavy objects to baskets with stronger load-bearing capacity, while avoiding light objects occupying high-load-bearing baskets and causing resource waste. Score_heat is the heat balance score, defined as: Score_heat = f / T, where f is the historical usage frequency of the basket within the preset time window T, and T is the preset time window length, such as 30 minutes or 1 hour. The smaller the value, the more vacant the basket is recently, and the allocation algorithm tends to use "cold" baskets, so that the utilization rate of the baskets in the entire station tends to be balanced in the time dimension, and the fatigue wear of some popular baskets is reduced. Score_ergo is an ergonomic score, defined as: Score_ergo=|y-H_center|, where y is the height coordinate of the candidate basket position, and H_center is the center value of the preset optimal operating height range [H_min, H_opt], ​​H_center= (H_min + H_opt) / 2. Typically, H_min is set to 0.6 meters and H_opt to 1.2 meters, corresponding to a comfortable access range for adults without bending over or tiptoeing. The smaller the value, the closer the height is to the optimal operating center, which helps reduce the labor intensity of operators or recipients and improves the human-computer interaction experience at the last-mile delivery station. The system selects the basket with the smallest score as the target physical basket ID. This greedy strategy completes the decision in polynomial time, is easy to implement and optimize in engineering, and can adapt to different station layouts, business hours, and operational strategies through flexible configuration of weight coefficients.

[0034] The central dynamic addressing layer also maintains a time sliding window Δt. When n packages (n>1) belonging to the same recipient are detected sequentially requesting basket allocation within Δt, an aggregation mode is triggered. In aggregation mode, for the second and subsequent packages of the same recipient, the weight coefficient α of the aggregation score is temporarily amplified to α', where α' = α×K, and K is an aggregation gain factor greater than 1, with a default value of 6.67. This ensures that subsequent packages are preferentially allocated to baskets closer to the Manhattan distance of already allocated baskets, thereby strengthening the spatial clustering of packages belonging to the same recipient. The amplified α' will significantly increase the weight of the aggregation score in the total score. Even at the cost of some weight matching or thermal balance, it ensures that packages belonging to the same recipient are concentrated in a few adjacent or nearby baskets, thus significantly reducing the walking distance and basket search time for the recipient.

[0035] The size constraints further include: if the package type is marked as large or irregular (e.g., identified by the main / irregular item detection sensor of the sorting system, or marked with a "large" label in the waybill information), in addition to checking the basic three-dimensional dimensions, it is also necessary to check whether the longest side of the package meets the redundancy space condition of 2×max(L,D,H), where max(L,D,H) represents the maximum value among the length, width, and height of the basket position; if the redundancy space condition is not met, the basket position is removed from the candidate set. This redundancy space condition ensures that large or irregular packages have sufficient adjustment and buffer space in the basket, avoiding the package from jamming the basket or affecting the access of adjacent basket positions due to excessive length in a certain dimension, while also taking into account the diagonal tilting requirements that may exist when the package is actually placed.

[0036] The dynamic mapping relationship is time-sensitive, and its state transition logic is uniformly managed by the central dynamic addressing layer. Specifically, the idle state transitions to the pre-occupied state via a labeling event, the pre-occupied state transitions to the occupied state via an inbound confirmation event, and the occupied state transitions back to the idle state via a pickup confirmation event (triggered by changes in the photoelectric sensor or gravity sensor built into the storage basket, or by operator scanning a code). When a basket position remains in the pre-occupied state for more than a preset time T_timeout without transitioning to the occupied state, it indicates that the package may be delayed during transportation or handover. The central dynamic addressing layer proactively pushes a verification and re-labeling task to the labeling and printing all-in-one machine at the terminal station execution layer. Here, T_timeout is the pre-occupied timeout threshold, which can be dynamically adjusted based on factors such as the distance between the station and the sorting center and transportation timeliness. The pushed task information includes the waybill number, the original assigned basket position ID, and the timeout duration, guiding the operator to proactively find the package and re-label it or intervene manually, thereby avoiding resource waste caused by long-term unoccupied basket positions.

[0037] The physical storage basket has a built-in gravity sensor. The central dynamic addressing layer is configured to automatically trigger a pickup notification to the recipient when the gravity sensor detects that the total weight of the items in the basket reaches a preset weight threshold W_notify, where W_notify is the weight threshold that triggers the notification. This proactively reminds the recipient when all packages arrive or the weight accumulates to the expected value, avoiding frequent fragmented notifications and improving the user experience. Simultaneously, the gravity sensor can also assist in determining whether there are packages in the basket (real-time verification) and whether any packages have been mistakenly taken or left behind, providing hardware verification for state transitions.

[0038] The atomic operations are implemented using a distributed lock mechanism, with the lock granularity at the individual basket position level. Before performing a state change on a target basket position, the distributed lock for that basket position is acquired first. Only after verifying that the position is still idle under lock protection is the state updated to pre-occupied. This ensures that in high-concurrency sorting scenarios, the same idle basket position will not be simultaneously allocated to two different packages. The distributed lock can employ Redis's SETNX command with automatic timeout release, or be based on ZooKeeper's ephemeral sequential nodes, guaranteeing an immediate failure return and attempt to find a suboptimal basket position when lock contention fails, thereby constructing a highly available addressing and allocation service.

[0039] A logistics sorting, package collection and transportation, and last-mile storage method based on cross-space dynamic mapping includes the following steps: S1. At the sorting center, the moment a package leaves the sorting chute, a "streetlight"-style multi-dimensional sensing unit collects the waybill information. If the visual sensor captures a clear image, the waybill number is obtained. If the visual sensor fails to capture a clear image, the waybill number is retrieved from the chute allocation record of the sorting system using the timestamp recorded by the infrared grating. During the retrieval, the system queries the chute allocation log table and matches the allocation record closest to that moment to extract the waybill number. Based on the obtained waybill number, the cloud-based central dynamic addressing layer determines the target end station (matching the preset station routing table according to the waybill address) and runs a multi-constraint greedy addressing algorithm to allocate the target physical basket ID: first, hard constraint trimming is performed from the available baskets in the target station, including weight constraint w≤W_max, size constraint l≤L and d ≤D and h ≤ H, and locking constraint excluding baskets under maintenance. For packages marked as large or irregularly shaped, the size constraint must also check whether the longest side of the package satisfies 2×max(L, D, H) redundancy space conditions; for the candidate set pruned through hard constraints, the weighted score is calculated according to the following formula: ; Select the basket with the smallest score as the target physical basket ID, update the basket status from idle to pre-occupied through distributed lock atomic operations, and establish a dynamic mapping relationship between the unique identifier of the package and the basket ID. S2. Within the same communication cycle, the micro-thermal labeling unit prints an auxiliary mapping label containing the target physical basket ID issued by the central dynamic addressing layer, and immediately attaches it to the non-label area on the package surface, completing the physical-information mapping. The printing and labeling actions are coordinated by a PLC or motion controller to ensure that the label is firmly attached before the package enters the collection container, preventing it from falling off during subsequent transportation. S3. Parcels are transported to the corresponding end-point station via the collection container. Operators quickly visually locate the corresponding physical storage bin based on the bin ID on the auxiliary mapping label, and place the parcel into the bin with the unique physical code, completing the warehousing process. After warehousing, a confirmation event can be triggered via the confirmation button on the bin or by automatic sensing, changing the system status to "occupied". S4. When a parcel is found to have a missing or unreadable auxiliary mapping label at the station, the original waybill number of the parcel is scanned by the end-point labeling and printing all-in-one machine. The system retrieves the dynamic mapping relationship corresponding to the waybill number from the central dynamic addressing layer and verifies the validity of the mapping. Only when the mapping relationship is valid is a new auxiliary mapping label allowed to be reprinted. After the reprint is completed, the old label mapping relationship corresponding to the parcel is marked as used. The system rejects any secondary basket binding requests based on the used mapping relationship, ensuring that the same parcel always corresponds to only one valid basket binding in the system, preventing the problem of one parcel being bound to multiple baskets due to concurrent operations.

[0040] In step S1, when the central dynamic addressing layer detects multiple packages for the same recipient within a preset time window Δt, it triggers an aggregation mode. During score calculation, the aggregation score weight coefficient α is temporarily amplified, causing subsequent packages to be preferentially clustered in the vicinity of already assigned basket positions. For example, after the first batch of packages falls into basket position (2,3,1), subsequent packages in the selectable basket positions will have smaller Score_dist values ​​for candidate basket positions with smaller Manhattan distances. Under the amplified α', the overall score advantage is significant, thus increasing the probability of selection. This results in packages for the same recipient ultimately being concentrated in one or two adjacent basket groups. Simultaneously, when the visual sensor fails to capture a clear image, the waybill number is retrieved from the sorting system using the timestamp recorded by the infrared grating, ensuring the completeness of the waybill information collection and continuously completing labeling and mapping establishment. This dual-mode redundancy design makes the waybill collection success rate close to 100%, laying a solid data foundation for subsequent fully automatic mapping and seamless delivery.

[0041] Example 2

[0042] The present invention also provides another way to trigger the pickup notification, namely, by having the delivery person scan the delivery confirmation mark after completing the delivery to trigger the notification, so as to reduce hardware costs.

[0043] Delivery confirmation identifiers (such as QR codes or barcodes) are placed at last-mile delivery stations or parcel lockers. After the delivery person completes the operation of transferring the item from the logistics container to the physical collection bin, they scan the delivery confirmation identifier using a handheld terminal. The handheld terminal then sends a confirmation signal containing the collection bin number and the delivery completion time to the central dynamic addressing layer. Upon receiving the confirmation signal, the central dynamic addressing layer triggers a pickup notification to the recipient. Optionally, the central dynamic addressing layer can aggregate multiple delivery confirmation signals within a preset time window and send pickup notifications to the corresponding recipients in batches.

[0044] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping, characterized in that, include: The central dynamic addressing layer, deployed in the cloud, is used to maintain the occupancy status matrix of all physical storage bins in the last-mile delivery stations in real time. The occupancy status matrix records the current status of each storage bin and the unique identifier of the package it is bound to. The occupancy status includes at least an idle state, a pre-occupied state, and a fully occupied state. The central dynamic addressing layer is configured to: respond to a package sorting request, run a multi-constraint greedy addressing algorithm to dynamically allocate target physical bin IDs, establish a dynamic mapping relationship between the package unique identifier and the target physical bin ID, and persistently store the mapping relationship. While establishing the mapping relationship, atomic operations are used to update the state of the selected basket in the occupancy state matrix from idle to pre-occupied; The sorting center mapping execution layer, located below the sorting line, includes a "streetlight"-style multi-dimensional sensing unit and a micro-thermal labeling unit. The "streetlight"-style multi-dimensional sensing unit is an inverted U-shaped structure spanning above the sorting conveyor line, integrating a vision sensor and an infrared grating. It is configured to collect waybill information the instant the package leaves the sorting chute. When the vision sensor fails to capture a clear image, the system uses the package's timestamp recorded by the infrared grating to retrieve the corresponding waybill number from the chute allocation record of the sorting system, ensuring reliable acquisition of waybill information. The micro-thermal labeling unit is synchronized with the sensing unit via an industrial communication network. Within the same communication cycle, based on the target physical basket ID issued by the central dynamic addressing layer, it synchronously prints and attaches auxiliary mapping labels, affixing the labels to the non-waybill area of ​​the package surface. The last-mile delivery station execution layer includes multiple storage baskets with unique physical codes and a labeling and printing machine. The storage baskets are used to hold packages and correspond one-to-one with physical basket IDs. The labeling and printing machine is configured to, when a package arrives at the station, if the auxiliary mapping label attached to its surface is missing or unreadable, query the central dynamic addressing layer by scanning the original package tracking number to obtain the dynamic mapping relationship corresponding to that tracking number, acquire the assigned target physical basket ID, and regenerate the auxiliary mapping label on-site. When reprinting the label, the labeling and printing machine verifies the validity of the dynamic mapping relationship with the central dynamic addressing layer, and only allows the reprinting operation if the mapping relationship is valid. After reprinting, the original label mapping relationship corresponding to the package is marked as used, and the system rejects secondary basket binding requests based on the used mapping relationship.

2. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 1, characterized in that, The multi-constraint greedy addressing algorithm includes a hard constraint pruning layer and a multi-dimensional weighted scoring layer; The hard constraint pruning layer is used to sequentially perform the following filtering from all available baskets at the current target station to obtain candidate set C1: Weight constraints: Remove baskets whose package weight w exceeds the maximum load capacity W_max. Size constraints are applied to exclude baskets whose three-dimensional dimensions (l, d, h) exceed the basket's capacity (L, D, H), where l, d, and h are the length, width, and height of the package, respectively, and L, D, and H are the length, width, and height of the basket. Lock constraints and remove baskets marked as under maintenance or reserved. If the candidate set C1 is empty after trimming, the no-solution exception handling process is triggered, and a manual intervention instruction is sent to the sorting center. The multi-dimensional weighted scoring layer is used to calculate a weighted score (Score) for each candidate basket in the candidate set C1. The scoring function is: ； Wherein, α, β, γ, and δ are configurable weight coefficients, with default values ​​of 0.3, 0.2, 0.1, and 0.4, respectively; Score_dist is the aggregation score, reflecting the degree of aggregation of packages; It is defined as: when the same recipient already has packages in a pre-occupied or occupied state at the target station, obtain the set of coordinates of the locations of these existing packages in the designated bins. Let the coordinates of the candidate basket position be (x, y, z), where x, y, and z are the column coordinates, height coordinates, and depth coordinates of the candidate basket position, respectively, and x_q, y_q, and z_q are the column coordinates, height coordinates, and depth coordinates of the existing basket position, respectively. ； The smaller the value, the higher the physical aggregation with the assigned package; if the recipient currently has no other packages, then Score_dist = 0; Score_weight is the weight fit score, defined as: Score_weight = (w / W_max) × 100. The smaller the value, the better the weight fit. Score_heat is the thermal balance score, defined as: Score_heat=f / T, where f is the historical usage frequency of the basket within the preset time window T, and T is the length of the preset time window. The smaller the value, the more idle the basket is recently. Score_ergo is the ergonomic score, defined as: Score_ergo=|y-H_center|, where y is the height coordinate of the candidate basket position, and H_center is the center value of the preset optimal operating height interval [H_min, H_opt], ​​H_center=(H_min + H_opt) / 2, the smaller the value, the closer the height is to the optimal operating center; The system selects the basket with the smallest score value as the target physical basket ID.

3. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 2, characterized in that, The central dynamic addressing layer also maintains a time sliding window Δt. When n packages (n>1) belonging to the same recipient are detected to request basket allocation in sequence within Δt, an aggregation mode is triggered. In the aggregation mode, for the second and subsequent packages of the same recipient, the weight coefficient α of the aggregation degree score is temporarily amplified to α', α' = α×K, where K is an aggregation gain factor greater than 1, with a default value of 6.

67. This allows subsequent packages to be preferentially allocated to baskets that are closer to the Manhattan distance of the baskets where the already allocated packages are located, thereby strengthening the spatial clustering of packages belonging to the same recipient.

4. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 2, characterized in that, The size constraints further include: if the package type is marked as large or irregular, in addition to checking the basic three-dimensional dimensions, it is also necessary to check whether the longest side of the package satisfies the redundancy space condition of 2×max(L,D,H), where max(L,D,H) represents the maximum value among the three dimensions of the basket position: length, width, and height; if the redundancy space condition is not met, the basket position is removed from the candidate set.

5. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 1, characterized in that, The dynamic mapping relationship is time-sensitive, and its state transition logic is uniformly controlled by the central dynamic addressing layer. Specifically, the idle state is converted to the pre-occupied state by the labeling event, the pre-occupied state is converted to the occupied state by the basket entry confirmation event, and the occupied state is converted to the idle state by the item retrieval confirmation event. When a basket is in the pre-occupied state for more than a preset time T_timeout and has not been converted to the occupied state, the central dynamic addressing layer actively pushes a verification and reprint task to the labeling and printing all-in-one machine of the terminal station execution layer. Here, T_timeout is the pre-occupied timeout threshold.

6. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 1, characterized in that, The physical storage basket has a built-in gravity sensor; the central dynamic addressing layer is configured to automatically trigger a pickup notification to the recipient when the gravity sensor detects that the total weight of the items in the basket reaches a preset weight threshold W_notify, where W_notify is the weight threshold that triggers the notification.

7. The logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 1, characterized in that, The atomic operations are implemented through a distributed lock mechanism, with the lock granularity being a single basket position. Before performing a state change on a target basket position, the distributed lock of that basket position is acquired first, and the state is updated to pre-occupied only after verifying that the state is still idle under the protection of the lock. This ensures that the same idle basket position will not be assigned to two different packages at the same time in a high-concurrency sorting scenario.

8. A logistics sorting, package collection and transportation, and last-mile storage method based on cross-space dynamic mapping, characterized in that, A logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping as described in any one of claims 1-7 includes the following steps: S1. At the sorting center, the "streetlight" type multi-dimensional sensing unit collects the waybill information the instant the package leaves the sorting chute. If the visual sensor captures a clear image, the waybill number is obtained. If the visual sensor fails to capture a clear image, the waybill number is retrieved from the chute allocation record of the sorting system using the timestamp recorded by the infrared grating. Based on the obtained waybill number, the cloud-based central dynamic addressing layer determines the target end station and runs a multi-constraint greedy addressing algorithm to allocate the target physical basket ID: First, hard constraint pruning is performed from the available baskets in the target station, including weight constraints w≤W_max, size constraints l≤L and d ≤D and h ≤ H, and locking constraints to exclude baskets under maintenance. For packages marked as large or irregularly shaped, the size constraint must also check whether the longest side of the package satisfies the redundancy space condition of 2×max(L, D, H). For the candidate set that passes the hard constraint pruning, the weighted score is calculated according to the following formula: ； Wherein, α, β, γ, and δ are weighting coefficients, and the definitions of each parameter are the same as in claim 2; the basket with the smallest score value is selected as the target physical basket ID, and the status of the basket is updated from idle to pre-occupied through distributed lock atomic operations, and a dynamic mapping relationship between the unique identifier of the package and the basket ID is established; S2. Within the same communication cycle, the micro-thermal labeling unit prints an auxiliary mapping label containing the target physical basket ID issued by the central dynamic addressing layer, and immediately attaches it to the non-sheet area of ​​the package surface to complete the physical-information mapping marking. S3. The parcel is transported to the corresponding terminal station via the collection container. The operator puts the parcel into the storage basket with the corresponding unique physical code according to the basket ID on the auxiliary mapping label, and completes the warehousing. S4. When a parcel is found to have a missing or unreadable auxiliary mapping label at the station, the original waybill number of the parcel is scanned by the end-point labeling and printing all-in-one machine. The system retrieves the dynamic mapping relationship corresponding to the waybill number from the central dynamic addressing layer and verifies the validity of the mapping. Only when the mapping relationship is valid is a new auxiliary mapping label allowed to be reprinted. After the reprint is completed, the old label mapping relationship corresponding to the parcel is marked as used, and the system rejects any secondary basket binding requests based on the used mapping relationship.

9. A logistics sorting, package collection and transportation, and last-mile storage method based on cross-space dynamic mapping according to claim 8, characterized in that, In step S1, when the central dynamic addressing layer detects multiple packages for the same recipient within a preset time window Δt, it triggers an aggregation mode. When calculating the score, the aggregation degree score weight coefficient α is temporarily amplified, so that the subsequent package basket allocation results are preferentially clustered in the adjacent area of ​​the already allocated baskets. At the same time, when the visual sensor fails to capture a clear image, the waybill number is retrieved from the sorting system using the timestamp recorded by the infrared grating to ensure the completeness of the waybill information collection, thereby continuously completing the labeling and mapping establishment.

10. A logistics sorting, package collection and transportation, and last-mile storage system based on cross-space dynamic mapping according to claim 1, characterized in that, A delivery confirmation icon is set up at the last-mile delivery station. After completing the delivery and unloading, the deliveryman scans the delivery confirmation icon to generate a confirmation signal and sends it to the central dynamic addressing layer. Upon receiving the confirmation signal, the central dynamic addressing layer triggers a pickup notification to the recipient.

Images