Novel packaging box tracking, transporting and recycling system

Abstract

The invention discloses a novel packaging box tracking, transporting and recycling system, relates to the field of logistics intelligent management, and solves the problems of storage information isolation, low verification efficiency, monitoring report splitting and single recycling judgment. The system comprises a three-dimensional storage module, a verification module before transportation, a transportation monitoring module and a recovery module after transportation, closed-loop management is carried out through a core algorithm, the box selection efficiency is improved by 83% through a three-dimensional storage association algorithm, and the storage error is reduced to 0.8%; a multi-dimensional verification algorithm enables the verification error to be 0.3%; the abnormal feedback delay is shortened to 1.8 minutes through a transportation dynamic adaptation algorithm; the recovery priority algorithm enables the extraction misjudgment rate to be 1.5% and the recovery efficiency to be improved by 75%. The system is suitable for e-commerce promotion, multi-category retail and other scenes, both innovation and practicability are considered, the logistics cost is reduced, and the customer satisfaction is improved.

Description

Technical Field

[0001] This system achieves closed-loop management of the entire packaging box process—from box selection to storage, verification, transportation, and recycling—through deep collaboration of four functional modules: 3D storage management, pre-transport cargo verification, transportation dynamic monitoring, and post-transport inspection and recycling. Driven by a quantitative algorithm, this version does not currently include a flowchart, but it reserves flowchart logic through a "step-by-step workflow" (which can be directly mapped to flowchart nodes later). The system focuses on a detailed explanation of algorithmic details (parameter definition, formula derivation, threshold basis) and the interconnected logic of all modules (data interaction, parameter linkage, and anomaly propagation) to ensure the rigor and feasibility of the technical solution. Background Technology

[0002] Packaging boxes are the core carriers of the entire logistics process, and their management efficiency directly affects cargo turnover costs and customer experience. However, existing technologies have bottlenecks in multiple aspects: the storage stage only records "packaging box ID + cargo location" without linking cargo attributes and box status, requiring manual matching for box selection (each box takes ≥30 seconds), and the storage statistics error rate reaches 8%; pre-transport verification relies on manual checking of orders and goods without combining weight verification, resulting in a mispacking and omission rate of ≥5%, with even lower efficiency during e-commerce promotions; transportation monitoring uses fixed-frequency reporting (e.g., every 30 minutes), failing to adapt to progress deviations and road conditions, with abnormal feedback delays of ≥10 minutes, and redundant information interfering with customers during normal transportation; post-transport recycling relies solely on appearance and rough weighing to determine the status of goods taken out, making it difficult to detect lightweight residues, with a misjudgment rate of 12%, and it is not sorted by recycling value, resulting in low resource utilization.

[0003] Furthermore, the existing system lacks key algorithm support: there is no storage information association algorithm to realize the linkage of "location-goods-box", no efficient verification algorithm to match multi-dimensional information, no dynamic reporting algorithm to balance timeliness and redundancy, and no multi-dimensional recycling algorithm to accurately determine the status, making it difficult to meet the efficient management needs of e-commerce promotions, multi-category retail and other scenarios. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a novel packaging box tracking, transportation and recycling system to solve the problems mentioned in the background art.

[0005] 3D storage management 1.1 Workflow S1: Receive box selection request (triggering condition: the warehouse management system issues the instruction of "goods category / weight / volume + outbound port location"); S2: Candidate Packaging Box Parameter Collection (including 3 sub-steps: ① RFID reading of box ID and pre-adapted goods type; ② UWB positioning to obtain cargo location coordinates; ③ Warehouse management system to call the current status of the box (available / occupied / to be cleaned)). S3: Core parameter calculation (calculate L correction, M correction, S value); S4: Calculation and screening of storage correlation index (final SI) (filtering low-fit candidate boxes with final SI < 0.5). S5: Box Selection Recommendation and Confirmation (The top 10 candidate boxes are pushed to the industrial tablet in descending order of SI. After confirmation by warehouse personnel, a "box ID-location coordinates-cargo information" association record is generated). S6: Storage statistics update (automatically generates "Category Distribution Table, Box Status Percentage Table, and Storage Location Utilization Rate Table" every 5 minutes, triggering anomaly alerts: storage location utilization rate <30% or >90%) 1.2 Core Algorithm (1) Position Distance Coefficient: To address the actual path lengthening caused by warehouse aisle obstruction, a passageway smoothness correction factor (C) is introduced. a The formula is: in: C a Value selection rules (based on warehouse type validation): Flat warehouse with unobstructed channel C a =1.0 (actual path = straight-line distance), partially obstructed passage (e.g., temporary storage of goods) C a =0.8 (Actual path = straight-line distance × 1.1), Fully obstructed passage (e.g., equipment maintenance) C a =0.5 (Actual path = straight-line distance × 1.25); Maximum storage location distance: preset according to warehouse layout (50-80m for flat warehouses and 100-150m for automated warehouses). For example, flat warehouses are often set to 50m during e-commerce promotions. Basis for the value: Data from 100 sets of warehouse tests showed that without correction of L, the actual handling path after box selection was 15%-40% longer than expected, and after correction, the path deviation was ≤5%, and the box selection efficiency was improved by 30%.

[0007] Goods matching is based on a three-dimensional approach: category, weight, and volume. The formula is as follows: in: Category matching degree (M) p ): Completely identical (e.g., "600L refrigerator box is compatible with 600L refrigerator") M p =1; Products of the same category but with a specification deviation ≤5% (e.g., "50L box is suitable for 52L goods") M p =0.5; Inconsistent product category (e.g., "food box compatible with industrial parts") M p =0; Weight fit factor (W) ): W = Actual weight of goods / Rated load capacity of the container, ranging from 0.3 to 1.0 (<0.3 indicates "too light, wasted space", >1.0 indicates "overweight, structural risk to the container", both are set with W). =0); Volume fit coefficient (V) ):V = Actual volume of goods / rated volume of box, range 0.3-1.0 (<0.3 indicates "too small, difficult to secure", >1.0 indicates "overcapacity, cannot be sealed", both are set at V) =0); Weighting criteria: Industry data shows that the damage rate (12%) caused by category mismatch is more than 1.5 times that of weight mismatch (8%) and volume mismatch (7%), therefore M p The highest weight (40%) is used to balance space utilization and cargo security.

[0008] Packaging box usability Classified by "Ready to Use - Pending Processing - Unavailable": Available (Undamaged, unbound goods) S=1; To be cleaned (Not cleaned after use, no structural damage) S=0.3; Occupied (bound goods) or scrapped (structural damage) S=0; Status update trigger conditions: S changes from 1 to 0 after box selection confirmation; S changes from 0.3 to 1 after goods are removed and cleaned; S changes from 0.3 to 0 (scrap) when appearance inspection A correction is <70%. ②SI final formula and threshold setting

[0009] Formula meaning: 1-L correction reflects "the closer the location, the higher the score" (when L correction = 0, 1-L correction = 1, the location is optimal); M correction directly reflects the suitability of the goods; S replenishment box status affects (usable boxes score 0.7 points higher than boxes to be cleaned). Threshold filtering logic (based on 1000 bin selection tests): SI final > 0.8: Highly adaptable box (optimal location + full adaptation + usable), handling efficiency after box selection ≥ 15 boxes / minute, adaptation success rate 100%; 0.5 < SI final ≤ 0.8: Medium-sized adapter box (e.g., if the location is good but needs cleaning, or if it is fully compatible but the location is far away), manual confirmation is required to determine whether to use it; SI final value ≤ 0.5: Low fit box (such as product category mismatch or location too far away) should be filtered directly to avoid waste of resources; Anomaly Handling: If the SI of all candidate boxes is ≤0.5, the system will automatically trigger "Location Adjustment Suggestion", analyze the reasons for low compatibility (such as "high-frequency cargo boxes are distributed in remote locations" or "insufficient box types"), and recommend moving the high-frequency cargo boxes to within 50m of the outbound gate (L correction <0.2). After adjustment, the compatibility rate will be improved to over 95%.

[0010] 2. Pre-transport cargo inspection: By automating hash matching (category / quantity / specification) and accurately verifying weight, an verification efficiency of "10 boxes / minute" is achieved, and the error rate is reduced to below 0.3%.

[0011] 2.1 Step-by-step workflow of flowchart logic: S1: Verification initialization (Triggering condition: The packaging box is loaded and placed on the intelligent weighing platform; the verification terminal obtains "Order ID-Goods List-Theoretical Total Weight-Box Weight W8" from the order system). S2: Goods information hash matching (① The barcode / RFID of each item is scanned by the barcode scanner to generate a "goods information string"; ② The terminal calculates the actual H; ③ The H list stored in the order system is called to compare whether they are consistent). S3: Weight verification (① The intelligent weighing platform reads the actual total weight of the fully loaded box and filters out abnormal data with fluctuations > 0.5%; ② Calculate ΔW correction). S4: Calculation and result determination of verification confidence (CC) (① Substitute into the CC formula; ② Classify according to CC value: pass / second verification / fail); S5: Result execution (if successful, a transport order with a QR code is generated; if the second verification is successful, discrepancies are marked for manual review; if unsuccessful, the shipment is returned for adjustment, and steps 2-4 are executed again).

[0012] 2.2 Core Algorithm This invention employs a multi-dimensional cargo verification algorithm (including hash rules and weight correction). The cargo information string format (fixed length 64 bits, to avoid ambiguity):

[0013] Hash algorithm selection: The SHA-256 encryption algorithm is used to generate a 256-bit fixed-length hash value, with a brute-force difficulty of [insert difficulty here]. (The collision rate is 0 in actual applications); Matching criteria: If the H list is completely consistent with the actual H (each of the 256 bits is the same), the "category / quantity / specification / order uniqueness" is determined to match; if they are inconsistent, the difference fields are automatically marked (such as "quantity digit 0010 ≠ 0009") to avoid manual full review.

[0014] Weight deviation rate (ΔW correction) calculation (eliminating box self-weight interference) This calculation needs to deduct the box's own weight (W8), otherwise it may lead to misjudgment due to missing small items (e.g., if a 0.5kg accessory is missing, the box's own weight is 5kg, and the full load weight is 120kg, without correction ΔW=0.42%, it is easy to be judged as qualified). Therefore, it needs to be corrected.

[0015] W8 acquisition method: Pre-set in the box RFID tag (written at the factory, accuracy ±0.01kg), automatically read during verification; Value range and meaning: ΔW correction ∈ [-5%, 5%] means "normal weight" (including cargo self-weight error and weighing error); >5% means "overweight" (may be overloaded or mixed with foreign objects); <-5% means "underweight" (may be missing from the shipment).

[0016] 3. GPS data filtering for transportation status uses Q=0.01 (process noise) and R=0.1 (measurement noise). Since the fluctuation of location data in logistics transportation mainly comes from vehicle bumps (low-frequency noise), this parameter combination can reduce the positioning error from 1m to within 0.5m, while avoiding lag caused by excessive smoothing.

[0017] TD Correction Stage Weights (W) j )set up: Last-mile delivery (W) j =2.0) has the highest weight because customers have the lowest willingness to wait at the last stage (studies show that a 10-minute delay at the last stage increases customer complaint rates by 40%); the outbound stage (W j =1.5) The weight is second, because the delay in outbound delivery will have a chain reaction effect on trunk transportation.

[0018] T reports final boundary constraints (5-60 minutes): Minimum interval of 5 minutes: to avoid high-frequency push notifications that could overload customer information (tests show that when the interval is less than 5 minutes, the customer open rate drops from 60% to 25%); maximum interval of 60 minutes: to ensure timely feedback on abnormal traffic conditions (such as congestion) (route adjustments can be completed within 60 minutes).

[0019] Goods Retrieval Visual matching degree (V-correction) region weights: The bottom of the container (W_q=0.4) has the highest weight because most of the cargo residue is concentrated there (accounting for 75%, such as fresh juice and parts debris). Focusing on this area can increase the residue detection rate by 60%. The corners (W_q=0.3) are the next most important because small items (such as screws and washers) are easily hidden in the corners.

[0020] Appearance integrity rate (A correction) and weighting of vulnerable areas: Corner (W) s=2.0) has the highest weight because the damage rate of the corners of the packaging box is 3 times that of the box wall (after 1000 cycles of testing, the corners account for 65%). Prioritizing the evaluation of the corners can help identify boxes with structural risks in advance.

[0021] RI Adjusted Weight Allocation: The remaining lease duration (50%) has the highest weighting because the recycling income of packaging boxes with high remaining lease value (such as T remaining lease = 25 days, cycle 30 days) is 5 times that of boxes with low remaining value. Prioritizing recycling can increase the number of annual cycles from 8 to 12.

[0022] Full module algorithm correlation logic Algorithm parameter linkage: The RC correction of the transportation module (such as RC=1.0 in case of an accident) will be synchronized to the recycling module. If an accident occurs during transportation, the recycling module will automatically improve the appearance inspection accuracy (the corner damage threshold is reduced from 2cm² to 1cm² when calculating A correction) to avoid missing hidden structural damage caused by the accident.

[0023] The "cargo-packing box association data" of the 3D storage management module is transferred to the pre-transport cargo verification module as the basis for generating the H_ manifest (ensuring that the manifest is consistent with the actual stored cargo). The "full load weight and order planned route" from the pre-transport cargo verification module are transmitted to the transportation dynamic monitoring module as the basis for TD calculation (planned travel time is determined based on the route). The "transportation duration" of the transportation dynamic monitoring module is transmitted to the post-transportation detection and recovery module as the basis for calculating T_remaining rental (T_remaining rental = total rental period - transportation duration). The "packaging box status (to be cleaned / discarded)" of the post-transportation inspection and recycling module is transmitted to the three-dimensional storage management module to update the storage statistics table (the status of boxes to be cleaned is set to S=0.3, and boxes to be discarded are removed from the candidate list). Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0025] Figure 1 Flowchart of pre-shipment cargo verification process Figure 2 Flowchart of the 3D storage management module; Detailed Implementation The application process of this system is further illustrated below with reference to two specific embodiments. It should be noted that the scope of protection of this invention is not limited to the following embodiments.

[0026] Example 1: E-commerce promotion scenario (transportation of home appliances) 1. Preprocessing stage Parameter settings: Maximum distance between storage locations (outlet as origin) 50m, rental period for packaging boxes for home appliances 30 days, allowable weighing deviation ±5%, customer level CR=1.5 (VIP customer). Hardware deployment: UWB tags are affixed to each storage location in the warehouse, RFID tags (storage ID: HM-2025-001, specification: compatible with 600L refrigerators) are affixed to appliance packaging boxes, barcode scanners and smart weighing platforms are deployed in the verification area, GPS and 4G modules are installed on transport vehicles, and vision cameras and weighing platforms are deployed in the recycling area.

[0027] 2. 3D storage box selection (stocking period starts November 10th) Box selection requirements: Outlet A3, 20 boxes suitable for 600L refrigerators need to be selected; Data extraction: The location distance of candidate container HM-2025-001 is 10m (L=10 / 50=0.2), the cargo matching degree is M=1 (suitable for 600L refrigerator), and the status is S=1 (available). SI calculation: SI = 0.4 × (1 - 0.2) + 0.4 × 1 + 0.2 × 1 = 0.32 + 0.4 + 0.2 = 0.92; Results: 20 boxes were recommended in descending order of SI, reducing the box selection time from the traditional 30 seconds / box to 4 seconds / box. The storage statistics table was updated in real time: "Location A3-10 → HM-2025-001 → 600L refrigerator × 1 → Available", with a statistical error rate of 0.8%.

[0028] 3. Pre-shipment verification (outbound on November 11th) Order information: HM-2025-001 corresponds to the order "600L refrigerator × 1, theoretical weight 120kg", H_list = SHA256("600L refrigerator-1-120kg"); Verification process: (1) Scan the RFID tag inside the refrigerator and generate the string "600L refrigerator-1-120kg", H_actual = H_list; (2) The actual weight is 120.5 kg, ΔW = (120.5 - 120) / 120 × 100% ≈ 0.42%; (3) CC = 1 - 0.5 × 0.42 ≈ 0.979 ≥ 0.8 → Verification passed, transport order generated; Efficiency comparison: Verification of 200 boxes took 60 minutes (traditional manual verification takes 100 minutes), with an error rate of 0.3%.

[0029] 4. Transportation dynamic monitoring (November 11-12, Hangzhou → Beijing) Basic parameters: Planned driving time 20 hours, CR=1.5; Dynamic adaptation: (1) Driving to Jinan section (actual time 8 hours, planned time 7 hours → TD=(7-8) / 7×100%≈-14.3%, road condition RC=0.8 (slow traffic)); (2) T_report = 30 × (1 - 0.3 × 14.3 / 100) / (0.8 × 1.5) = 30 × 0.957 / 1.2 ≈ 23.9 minutes (take 24 minutes); (3) Report content: "Current location: Jinan Ring Expressway K15, estimated delay of 1.5 hours, driver has been contacted to speed up"; Customer feedback: Abnormal feedback is delayed by 1.8 minutes (traditionally 12 minutes), and invalid information is reduced by 65%.

[0030] 5. Recycling after transportation (after arrival on November 12th) Test data: ΔW_take = (15-120) / 120×100% = -87.5%, V = 0.95 (slight residue), T_lease remaining = 25 days, A = 92%, recovery distance 0.5km; Algorithm calculation: (1) EC = 0.6 × (1 - 12.5 / 100) + 0.4 × 0.95 = 0.525 + 0.38 = 0.905 ≥ 0.9 → completely removed; (2) RI=0.5×(25 / 30)+0.3×(92 / 100)+0.2×(1 / 0.5)≈0.417+0.276+0.4=1.093; Result: Assigned to the high-priority recycling channel, cleaning is completed within 1 hour, and the bins are re-entered into the storage pool, with a recycling efficiency of 35 bins / hour (compared to 20 bins / hour in the traditional system).

[0031] Example 2: Multi-category retail scenario (supermarket food + daily necessities transportation) 1. Preprocessing stage Parameter settings: Maximum distance between storage locations (outlet as origin) 30m, packaging boxes are divided into two categories (food: suitable for snacks / beverages; daily necessities: suitable for laundry detergent / paper towels), rental period 15 days, allowable weighing deviation ±3% (food weight is highly sensitive), customer level CR=1.0 (ordinary supermarket customer). Hardware deployment: Same as in Example 1, but with the addition of a "Food / Daily Necessities" label to distinguish the type of packaging box.

[0032] 2. Three-dimensional storage box selection (routine replenishment period) Box selection requirements: Outlet B2 requires 10 food packaging boxes (suitable for snacks) and 5 daily necessities packaging boxes (suitable for laundry detergent). Data extraction: Food category box ID-SP001: Storage distance 6m (L=6 / 30=0.2), M=1 (suitable for snacks), S=1→SI=0.4×0.8+0.4×1+0.2×1=0.92; Daily Necessities Box ID-RI001: Storage distance 9m (L=9 / 30=0.3), M=1 (suitable for laundry detergent), S=1→SI=0.4×0.7+0.4×1+0.2×1=0.88; Results: The selection time for 15 packaging boxes was 75 seconds (compared to 450 seconds under traditional methods). The storage statistics table was displayed according to the categories of "food" and "daily necessities". The statistical error rate was 0.6%.

[0033] 3. Pre-transport verification Order information: SP001: Snacks (10 bags of potato chips, theoretical weight 2kg), H_list = SHA256("Potato chips - 10 - 2kg"); RI001: Laundry detergent (2kg x 3 bottles, theoretical weight 6kg), H_list = SHA256("Laundry detergent - 3 - 6kg"); Verification process: SP001: H_actual = H_list, actual weight 2.05kg, ΔW = 2.5% → CC = 1 - 0.5 × 2.5 = 0.875 ≥ 0.8 → Pass; RI001: H_actual = H_list, actual weight 5.9kg, ΔW = -1.67% → CC = 1 - 0.5 × 1.67 ≈ 0.916 ≥ 0.8 → Pass; Efficiency comparison: 15 boxes took 45 seconds to verify (traditional 150 seconds), with an error rate of 0.2% (no mispacking or missing boxes).

[0034] 4. Transportation dynamic monitoring (short-distance transportation within 30km to supermarket stores) Basic parameters: Planned driving time 1 hour, CR=1.0; Dynamic adaptation: (1) Driving to the urban section (actual time 0.4 hours, planned time 0.3 hours → TD=(0.3-0.4) / 0.3×100%≈-33.3%, RC=0.8 (congestion)); (2) T_report = 30 × (1 - 0.3 × 33.3 / 100) / (0.8 × 1.0) = 30 × 0.9 / 0.8 ≈ 33.75 minutes (take 34 minutes); (3) Report content: "Current location: XX intersection, estimated delay of 15 minutes, the store has been contacted to adjust the delivery time"; Customer feedback: Abnormal feedback is delayed by 1.5 minutes and there is no redundant information.

[0035] 5. Recycling after transportation Test data: SP001: ΔW_ = (0.3-2) / 2 × 100% = -85%, V = 0.9 (snack bag residue), T_remaining rental = 12 days, A = 95%, recycling distance 0.3km; RI001: ΔW_take = (0.5-6) / 6×100%≈-91.7%, V=0.98 (no residue), T_lease remaining = 10 days, A=90%, recycling distance 0.3km; Algorithm calculation: SP001: EC = 0.6 × (1 - 15 / 100) + 0.4 × 0.9 = 0.51 + 0.36 = 0.87 → Manual verification (no residue), RI = 0.5 × (12 / 15) + 0.3 × 0.95 + 0.2 × (1 / 0.3) ≈ 0.4 + 0.285 + 0.667 = 1.352; RI001:EC=0.6×(1-8.3 / 100)+0.4×0.98≈0.55+0.392=0.942; RI=0.5×(10 / 15)+0.3×0.9+0.2×(1 / 0.3)≈0.333+0.27+0.667=1.27; Results: SP001 was prioritized for recycling and cleaning was completed within 2 hours (food containers required disinfection), followed by RI001. The recycling efficiency was 32 containers / hour, with a recycling rate of 88%.

Claims

1. A novel packaging box tracking transportation recycling system, characterized by, The application relates to a three-dimensional storage management system. The three-dimensional storage management module is used for realizing the association management of packing boxes, stored goods and storage locations, executing box selection recommendation, storage statistics and warehouse in-out state updating; The pre-transport goods verification module is used for verifying the consistency of the types, quantity and weight of goods in the packing boxes, and executing goods information matching and weight verification; The transport dynamic monitoring module is used for collecting transport location, road condition and progress data in real time, dynamically adjusting customer report strategies and pushing transport information; The post-transport detection and recovery module is used for judging the goods taking-out state, calculating recovery priority and allocating recovery channels; The three-dimensional storage management module, the pre-transport goods verification module, the transport dynamic monitoring module and the post-transport detection and recovery module realize data interaction through a cloud database, and form a whole-process closed loop of "box selection-storage-verification-transport-recovery".

2. The system of claim 1, wherein, The three-dimensional storage management module embeds a three-dimensional storage association algorithm, calculates a storage association degree index SI=0.4x(1-L)+0.4xM+0.2xS, wherein L is a location distance coefficient, M is a goods matching degree and S is a packing box available state, and the module recommends optimal packing boxes in descending order of SI.

3. The system of claim 2, wherein, The goods matching degree M satisfies M=1 when the type and specification are completely consistent, M=0.5 when the type is consistent and the specification deviation is less than or equal to 5%, and M=0 when the type is inconsistent.

4. The system of claim 1, wherein, The pre-transport goods verification module embeds a multi-dimensional verification algorithm, calculates a verification confidence CC, and judges CC based on the SHA-256 hash matching result of an order and actual goods and a weight deviation rate AW. When CC is greater than or equal to 0.8, the goods are passed; when 0.5<=CC<0.8, the goods are subjected to secondary verification; and when CC is less than 0.5, the goods are failed.

5. The system of claim 4, wherein, The SHA-256 hash value is generated based on a "goods type-quantity-specification" string. When the hash values are completely consistent, the goods information is matched.

6. The system of claim 1, wherein, The transport dynamic monitoring module embeds a transport state-report adaptation algorithm, calculates a report interval T_report=30x(1-0.3xmax(0,-TD / 100)) / (RCxCR), wherein TD is a transport progress deviation, RC is a road condition coefficient and CR is a customer level, and the module pushes information according to T_report.

7. The system of claim 6, wherein, The road condition coefficient RC satisfies RC=0.2 when the road is smooth, RC=0.5 when the road is slow, RC=0.8 when the road is congested and RC=1.0 when the road is in an accident, and the road condition coefficient is obtained through a third-party map API.

8. The system of claim 1, wherein, The post-transport detection and recovery module embeds a taking-out judgment algorithm, calculates a taking-out confidence EC=0.6xmin(1,1-absolute value of AW_take+100) / 10)+0.4xV, wherein AW_take is a weight change rate and V is a visual matching degree, and EC is greater than or equal to 0.9 when the goods are completely taken out.

9. The system of claim 8, wherein, The post-transport detection and recovery module also calculates a recovery priority RI=0.5x(T_rentleft / rent period)+0.3x(A / 100)+0.2x(1 / recovery distance), and allocates recovery channels in descending order of RI.

10. The system of claim 1, wherein, The three-dimensional storage management module contains passive RFID and UWB positioning tags, the pre-transport verification module contains a bar code / RFID scanning gun and an intelligent weighing table, the transport monitoring module contains a vehicle-mounted GPS and a 4G module, and the recovery module contains a visual camera and a recovery weighing table.

Images