A multi-robot cooperative unattended express post station system and express consignment access method

Abstract

The application discloses a multi-robot cooperative unattended express post station system and express delivery item storage and taking method, and the system comprises an express delivery item rough inspection unit, a visual scanning unit, a delivery robot unit, a taking robot unit, a cabinet grid coding system and an autonomous path learning module. The visual scanning unit calculates the volume of the express delivery item through a multi-phase camera and identifies a bar code, and sends information to the delivery robot; the delivery robot matches the optimal idle cabinet grid based on the cabinet grid coding and volume information, and completes rail-free dynamic path planning through the autonomous path learning module; the taking robot receives an external trigger signal and matches a target cabinet grid. When warehousing, batch express delivery items are stored in the cabinet grid after rough inspection, scanning and matching; when taking out, the taking robot receives a trigger signal to position the cabinet grid and sequentially takes and delivers. The application realizes the full-process unmanned storage and taking of the express delivery items, gets rid of the constraint of fixed tracks, and improves the space utilization efficiency and site adaptability.

Description

Technical Field

[0001] This invention relates to the field of express delivery technology, specifically to an unmanned express delivery station system with multi-robot collaboration and a method for storing and retrieving express packages. Background Technology

[0002] With the continuous growth in last-mile delivery demand, community stations have become core terminal nodes in the logistics network, undertaking key functions such as temporary storage, sorting, delivery, and self-pickup of parcels. However, most mainstream express delivery stations currently operate manually, relying on staff to complete all aspects of operations, including unloading, preliminary sorting, shelving, parcel verification, and inventory management. This model not only leads to high labor costs for station operations but also limits processing efficiency due to staff workload. Especially during peak e-commerce promotional periods such as "Double Eleven" and "618," insufficient processing capacity can easily result in problems such as long queues, parcel backlogs, and compromised user experience.

[0003] To address the aforementioned reliance on manual labor, some unmanned or semi-automated parcel locker solutions have emerged in the existing technology field. These solutions typically employ a structure combining fixed conveyor belts with smart parcel lockers. Parcels are transported via the conveyor belt to the vicinity of designated lockers, where robotic arms or pusher mechanisms then secure them for storage. However, such fixed equipment systems are structurally complex, have high initial deployment costs, and once built, their site layout is difficult to adjust flexibly, resulting in poor adaptability to different sizes or irregularly shaped parcel locations. Furthermore, the accompanying transfer robots mostly travel along pre-set physical tracks or magnetic guide lines, with completely fixed operating paths. This makes them unable to handle dynamic scenarios such as adjustments to shelf layouts, temporarily stacked goods, or moving obstacles, leading to severely insufficient environmental adaptability and fault tolerance.

[0004] On the other hand, the existing locker management system is rather crude. Lockers are typically distinguished only by simple numerical or alphanumeric numbers, without integrating structural attributes such as spatial coordinates and internal dimensions (length, width, height, and volume). This deficiency leads to the system often only roughly allocating packages to lockers based on the availability of space, lacking the ability to accurately match package size with locker volume. This makes it difficult to effectively improve storage space utilization. Furthermore, it significantly increases the algorithmic complexity of robot locker positioning and scheduling, hindering efficient multi-machine collaborative operations and restricting the overall performance improvement of unmanned locker systems. Summary of the Invention

[0005] To address this, the present invention provides a multi-robot collaborative unmanned express delivery station system and a method for storing and retrieving express packages, solving the problems of low automation in the entire process of existing stations, poor scene adaptability due to the reliance of transfer robots on fixed tracks, low spatial matching efficiency, and insufficient positioning accuracy.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-robot collaborative unmanned express delivery station system, comprising: a package preliminary inspection unit, a visual scanning unit, a delivery robot unit, a pickup robot unit, a cabinet coding system, and an autonomous path learning module;

[0007] The express mail preliminary inspection unit is located at the entrance of the station. The express mail preliminary inspection unit receives a batch of express mail and picks them up one by one and transfers them to the scanning station.

[0008] The visual scanning unit is deployed above the scanning station. The visual scanning unit collects images of the express package and calculates the volume of the express package. Based on the volume, it matches the appropriate cabinet specifications, generates express package information, and sends it to the delivery robot unit.

[0009] The delivery robot unit locally stores all cabinet codes and real-time occupancy status data. The delivery robot unit receives package information, matches an available target cabinet with suitable specifications, plans a path to store the package in the target cabinet, and then updates the local data.

[0010] The pickup robot unit locally stores the correspondence data between cabinet codes and parcel barcodes. The pickup robot unit receives the user's pickup instruction, matches the corresponding cabinet code where the parcel is located, plans the route, grabs the parcel, and transports it to the pickup port for delivery to the user.

[0011] The cabinet coding system assigns a unique code to all cabinets in the station, and each code contains the spatial location coordinates and volume specifications of the corresponding cabinet.

[0012] The autonomous path learning module is mounted on the delivery robot and the pickup robot. The autonomous path learning module perceives the environment and dynamically plans the travel path.

[0013] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the pre-inspection unit includes an unloading area and a pre-inspection robotic arm. The pre-inspection robotic arm is equipped with a package sensing module. After detecting the arrival of a package, the pre-inspection robotic arm picks up the packages one by one and transfers them to the carrier rack at the scanning station.

[0014] As a preferred solution for an unmanned express delivery station system with multi-robot collaboration, the visual scanning unit includes a multi-phase camera array arranged in a ring above the scanning station support frame, used to collect images of express packages from multiple angles and identify the barcode information on the express package label.

[0015] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the formula for calculating the three-dimensional volume of the package by the visual scanning unit is:

[0016]

[0017] In the formula, Indicates the volume of the package. This indicates the maximum length of the package as measured by 3D reconstruction using a multi-phase camera. Indicates the maximum width. Indicates the maximum height; the visual scanning unit will It is sent to the delivery robot unit as part of the package information.

[0018] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the formula for matching the delivery robot unit with an idle and suitable target cabinet is as follows:

[0019]

[0020] In the formula, This is the set of all currently available cabinets. For cabinet index, For the first The internal volume specifications of each locker; among all available lockers with sufficient volume to accommodate parcels, the locker with the smallest volume difference is selected as the target locker. .

[0021] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the autonomous path learning module performs path planning using the A* algorithm model, with the cost function being:

[0022]

[0023] In the formula, For the current search node, From the starting point to the current node The actual cost of movement, To start from the current node Heuristic cost estimation to the target node; the autonomous path learning module perceives environmental obstacles in real time and dynamically updates the search space. and parameter.

[0024] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the autonomous path learning module updates the path cost through a dynamic weight adjustment formula when it detects temporary obstacles:

[0025]

[0026]

[0027] In the formula, For path segments, For the current moment, This represents the geometric distance of the path segment. For a moment Obstacle density coefficient on the next path segment, The preset obstacle avoidance weight coefficients; P represents a candidate path; Indicates the starting point from the current robot position. To the end of the target cabinet The set of all connected paths; This is the optimal path that minimizes the total cost after the update.

[0028] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, when the same pickup instruction corresponds to multiple packages, the pickup robot unit calculates the optimal pickup order:

[0029]

[0030] In the formula, This represents the total number of packages awaiting pickup. The coordinates of the cabinet that the robot has currently reached or is about to reach. The coordinates of the next cabinet the robot should go to. Let be the Euclidean distance function between two points. The set of all possible pickup paths. The optimal order for picking up items is the one that minimizes the total moving distance.

[0031] As a preferred solution for a multi-robot collaborative unmanned express delivery station system, the pickup robot unit is equipped with an instruction parsing module for receiving and parsing externally input pickup instructions. The pickup instructions include at least one type of identification information, such as barcode data, QR code data, or the last four digits of a mobile phone number. The pickup robot unit retrieves and matches the corresponding target cabinet based on the identification information in the locally stored correspondence data between cabinet codes and express barcodes. Each code in the cabinet coding system includes the row and column coordinates of the cabinet and the internal length, width, and height volume parameters of the cabinet.

[0032] The present invention also provides a method for storing and retrieving express mail using the above system, including an express mail inbound process and an express mail outbound process;

[0033] The express mail receiving process includes:

[0034] S11. The delivery vehicle arrives at the station entrance and unloads the batch of parcels into the unloading area;

[0035] S12. After the rough inspection robotic arm senses the package, it picks up the package one by one and places it on the carrier frame at the scanning station.

[0036] S13. The visual scanning unit collects images of the package, calculates its volume, determines the appropriate cabinet size, and transmits the package information to the delivery robot.

[0037] S14. After receiving the package information, the delivery robot matches the available target cabinet with the local cabinet data, plans the route through the autonomous path learning module, stores the package in the target cabinet and updates the local status.

[0038] The express delivery outbound process includes:

[0039] S21. The pickup robot unit receives an external pickup trigger signal containing package identification information;

[0040] S22. The package pickup robot unit matches the corresponding cabinet code of the package based on the package identification information;

[0041] S23. The pickup robot plans its path through the autonomous path learning module, reaches the target cabinet, and picks up the package; if the package identification information corresponds to multiple packages, the pickup is completed sequentially according to the optimal path.

[0042] S24. The pickup robot transports the grabbed package to the pickup port and triggers the pickup port to open to execute the package output action.

[0043] The present invention has the following advantages:

[0044] This invention achieves fully automated processing of express parcels from unloading, preliminary inspection, scanning into the locker, to user pickup through the collaboration of multiple devices including a preliminary inspection robotic arm, a vision scanning unit, a delivery robot, and a pickup robot. This significantly reduces the labor costs of the parcel station while greatly improving the efficiency of parcel processing and the throughput capacity during peak hours.

[0045] This invention's robot is equipped with an autonomous path learning module, enabling dynamic path planning without the need for magnetic strips or guide rails. It can perceive real-time adjustments to the shelf layout and changes in temporary obstacles within the station, possessing a high degree of scene adaptability and greatly improving the system's deployment flexibility in irregular sites and dynamic environments.

[0046] This invention establishes a cabinet coding system that integrates spatial location coordinates and volume specifications, enabling precise matching between package volume and cabinet capacity and avoiding space waste. At the same time, this coding system is pre-stored locally on the robot, significantly simplifying the robot's cabinet positioning and scheduling logic and improving system response speed.

[0047] This invention employs a modular architecture with multiple robots working collaboratively and each stage operating in parallel, ensuring strong independence for each unit. The number of devices can be flexibly configured according to the actual size of the delivery station and the volume of parcels delivered, adapting to different scales of last-mile delivery scenarios. It possesses excellent scalability and industrialization potential. Attached Figure Description

[0048] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0049] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0050] Figure 1 This is a schematic diagram of the multi-robot collaborative unmanned express delivery station system architecture provided in an embodiment of the present invention;

[0051] Figure 2 This is a schematic diagram illustrating an application scenario of the multi-robot collaborative unmanned express delivery station system provided in this embodiment of the invention.

[0052] Figure 3 This is a flowchart of a multi-robot collaborative unmanned express delivery station parcel storage and retrieval method provided in an embodiment of the present invention. Detailed Implementation

[0053] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] See Figure 1 and Figure 2This invention provides a multi-robot collaborative unmanned express delivery station system, including: a package preliminary inspection unit, a visual scanning unit, a delivery robot unit, a pickup robot unit, a cabinet coding system, and an autonomous path learning module. The package preliminary inspection unit is located at the station entrance, receiving batches of packages and picking them up one by one for transfer to the scanning station. The visual scanning unit is deployed above the scanning station, acquiring package images and calculating package volume. Based on the volume, it matches the appropriate cabinet specifications, generates package information, and sends it to the delivery robot unit. The delivery robot unit locally stores all cabinet codes and real-time occupancy status data. The robot unit receives package information, matches it with an available and suitable locker, plans a path to store the package in the locker, and updates its local data. The pickup robot unit locally stores the correspondence between locker codes and package barcodes. It receives user pickup instructions, matches the corresponding locker code, plans a path, grabs the package, and delivers it to the pickup point for user delivery. The locker coding system uniquely encodes all lockers in the station, with each code containing the corresponding locker's spatial coordinates and volume specifications. The autonomous path learning module is mounted on both the delivery robot and the pickup robot. This module senses the environment and dynamically plans its path.

[0055] In its implementation, the collaborative relationships between the parcel pre-inspection unit, visual scanning unit, delivery robot unit, pickup robot unit, locker coding system, and autonomous path learning module are built upon a fully localized data architecture. The locker coding system pre-assigns a unique code containing spatial coordinates and volume specifications to each locker within the station. This code library is synchronously stored in the local storage of both the delivery and pickup robot units, and the corresponding robot unit is responsible for updating the occupancy status in real time after each parcel is placed in or removed from a locker. The parcel volume information generated by the visual scanning unit is packaged and directly sent to the delivery robot unit. The delivery robot unit can independently complete target locker matching and path planning based on the locally stored real-time occupancy status data without needing to access a remote central server. The autonomous path learning module continuously senses changes in the physical environment within the station, providing the delivery and pickup robot units with dynamic path planning capabilities that do not rely on fixed physical tracks or magnetic strip guidance. This architecture fundamentally breaks free from the constraints of existing transfer robots that rely on preset tracks, while avoiding the risk of single point of failure, giving the system the physical basis for continuous and stable operation in dynamic scenarios such as shelf layout adjustments and the appearance of temporary obstacles.

[0056] In one possible embodiment, the pre-inspection unit for express mail includes an unloading area and a pre-inspection robotic arm. The pre-inspection robotic arm is equipped with an express mail sensing module. After detecting the arrival of express mail, the pre-inspection robotic arm picks up the express mail one by one and transfers it to the carrier rack at the scanning station.

[0057] Specifically, during operation, the pre-inspection unit handles a large number of unordered parcels unloaded from delivery vehicles in the unloading area, located at the station entrance for easy vehicle unloading. A parcel sensing module on the pre-inspection robotic arm continuously monitors changes in the spatial contour of the stacked parcels, identifying the boundary positions of individual parcels to be grasped and generating trigger signals. Upon receiving these signals, the robotic arm, based on a pre-set spatial trajectory control algorithm, sequentially grasps, lifts, and horizontally moves each parcel from the unloading area, ultimately placing it smoothly onto a transparent support frame at the scanning station. This continuous physical process transforms the stacked, disordered parcels into an ordered sequence of individual parcels side-by-side, effectively eliminating the adverse effects of overlapping or obstruction during visual scanning and providing a favorable physical foundation for accurate identification by the visual scanning unit.

[0058] In one possible embodiment, the delivery robot unit and the pickup robot unit have the same hardware structure, both equipped with a mobile chassis, a multi-degree-of-freedom robotic arm, a carrying basket, a local storage unit, and an environmental perception component mounted on the autonomous path learning module; the system is also equipped with a dynamic task scheduling module, which is used to monitor the geographical location, working status and task queue length of each robot in real time, and dynamically allocate the job roles to be performed by each robot according to the monitoring results.

[0059] When the number of parcel receiving tasks exceeds the number of currently idle delivery robots, the dynamic task scheduling module selects the idle pickup robot unit closest to the scanning station and sends it a temporary task switching instruction. Upon receiving the instruction, the selected pickup robot unit switches its locally stored working mode from "pickup mode" to "delivery mode," loads the same cabinet code and real-time occupancy status data mapping table as the delivery robot unit, then receives parcel information transmitted by the visual scanning unit, completes target cabinet matching, path planning, and parcel placement, and updates its local status data after placement. Similarly, when pickup tasks surge, idle delivery robot units can also receive instructions to switch to "pickup mode," load the correspondence data between cabinet codes and parcel barcodes, and perform pickup and delivery operations.

[0060] Furthermore, in addition to a fixed pre-inspection robotic arm, the express mail pre-inspection unit also includes a mobile pre-inspection station. When the fixed pre-inspection robotic arm malfunctions or express mail accumulates in the unloading area, the dynamic task scheduling module can assign the nearest delivery robot or retrieval robot to the unloading area. These robots will then use their onboard robotic arms to perform package sensing and individual package grabbing actions, directly transferring the packages to the carrier rack at the scanning station, thus temporarily replacing the function of the express mail pre-inspection unit. Once the malfunction is resolved or the backlog is alleviated, the scheduling module will restore the robot to its original delivery or retrieval role.

[0061] In this embodiment, the role interchange between the delivery robot, the pickup robot, and the express delivery preliminary inspection unit is realized based on a unified hardware platform and flexible task scheduling logic. This enables the system to automatically adjust the functional division of each device according to the instantaneous workload, effectively avoiding the overall processing efficiency decline caused by the overload of a certain unit, while improving the system's fault tolerance and uninterrupted operation capability in the event of equipment malfunction.

[0062] In one possible embodiment, the visual scanning unit includes a multi-phase camera array arranged in a ring above the scanning station support frame, used to acquire images of express packages from multiple angles and identify the barcode information on the express package label.

[0063] Specifically, the visual scanning unit employs a multi-phase camera array arranged in a ring above the scanning station's support frame, forming a comprehensive visual acquisition field. Since the orientation of the packaging surface of a package cannot be pre-controlled when it is placed on the support frame by the robotic arm, monocular vision solutions struggle to effectively identify packages with the barcode facing down or obscured by its own packaging. The ring array, however, acquires images simultaneously from different spatial orientations, enabling it to capture the reflective area of ​​the barcode on any placement, thus ensuring a high degree of tolerance for package orientation during scanning. Furthermore, the visual parallax information contained in the multi-view image data also forms the necessary spatial data foundation for 3D modeling, providing a reliable source of data for the visual scanning unit to perform subsequent volume measurement calculations.

[0064] In one possible embodiment, the formula used by the visual scanning unit to calculate the three-dimensional volume of the package is:

[0065]

[0066] In the formula, Indicates the volume of the package. This indicates the maximum length of the package as measured by 3D reconstruction using a multi-phase camera. Indicates the maximum width. Indicates the maximum height; the visual scanning unit will It is sent to the delivery robot unit as part of the package information.

[0067] Specifically, the visual scanning unit inputs the synchronous images captured by the multi-phase camera array into the stereo vision processing module. Through point matching and triangulation, it reconstructs the contour point cloud of the package surface in three orthogonal directions. Then, it extracts the maximum dimensions of the package's outer contour in the three axes from the point cloud data: maximum length, maximum width, and maximum height. The visual scanning unit performs a direct multiplication operation on these three axial dimensions to obtain a value representing the actual physical volume of the package's outer envelope. This volume value, together with the package label barcode information synchronously identified by the visual scanning unit, constitutes a complete package information data packet. This packet is transmitted via the local area network communication link to the local cache of the delivery robot unit, serving as input parameters for the delivery robot unit's cabinet selection and target matching operations.

[0068] In one possible embodiment, the formula for the delivery robot unit to match an idle and suitable target cabinet is:

[0069]

[0070] In the formula, This is the set of all currently available cabinets. For cabinet index, For the first The internal volume specifications of each locker; among all available lockers with sufficient volume to accommodate parcels, the locker with the smallest volume difference is selected as the target locker. .

[0071] Specifically, when matching a target locker, the delivery robot unit first retrieves all locker codes and corresponding real-time occupancy status data from its local storage, and then performs a two-layer logical filtering based on the package volume value attached to the package information data packet. The first layer of constraint is to check the set of available lockers. The first constraint is to check whether the internal volume of each locker is greater than or equal to the volume of the package. This physically eliminates invalid lockers that are too small to accommodate the target package, ensuring that the package can be placed smoothly without jamming or deformation. The second constraint is to calculate the difference between the internal volume of each locker and the volume of the package among all candidate lockers that have passed the volume screening, and select the locker with the smallest difference as the final target locker. By minimizing this difference, the delivery robot unit can ensure that each package occupies a locker with a volume closest to its own, avoiding the fragmentation and waste of locker space caused by loading small packages into large lockers, thereby maximizing the use of the limited high-value storage space resources within the station.

[0072] In one possible embodiment, the autonomous path learning module performs path planning using the A* algorithm model, with the cost function being:

[0073]

[0074] In the formula, For the current search node, From the starting point to the current node The actual cost of movement, To start from the current node Heuristic cost estimation to the target node; the autonomous path learning module perceives environmental obstacles in real time and dynamically updates the search space. and parameter.

[0075] Specifically, during path planning, the autonomous path learning module abstracts the map information within the station into a node network with several connecting edges, marking insurmountable shelf boundaries or virtual partitions as impassable areas. The path planning is executed using the A* algorithm, which achieves efficient convergence of the search process by introducing a heuristic cost estimation term. The actual movement cost term reflects the cumulative physical distance traveled by the delivery or retrieval robot unit from the starting node to the current node, while the heuristic cost estimation term is estimated based on the straight-line Euclidean distance between the current position and the target shelf position. The weighted sum of these two terms forms the total estimated value of the current node. As the delivery or retrieval robot unit moves along the planned path, the autonomous path learning module uses onboard LiDAR and visual sensors to detect the distribution of surrounding obstacles in real time, incorporating corresponding changes in space occupancy into the search space and dynamically updating the cost value of each connecting edge in the node network. This continuous feedback mechanism allows the delivery and retrieval robot units to continuously adjust their routes without stopping, improving the system's real-time adaptability to transient environments such as temporarily stacked goods or moving obstacles.

[0076] In one possible embodiment, when the autonomous path learning module detects a temporary obstacle, it updates the path cost using a dynamic weight adjustment formula:

[0077]

[0078]

[0079] In the formula, For path segments, For the current moment, This represents the geometric distance of the path segment. For a moment Obstacle density coefficient on the next path segment, The preset obstacle avoidance weight coefficients; P represents a candidate path; Indicates the starting point from the current robot position. To the end of the target cabinet The set of all connected paths; This is the optimal path that minimizes the total cost after the update.

[0080] Specifically, when the autonomous path learning module detects temporary obstacles in the currently planned path using sensors, a dynamic weight adjustment mechanism intervenes. This mechanism operates by converting the conventional static path length into a dynamically changing passage cost over time. For each path segment in the network, the autonomous path learning module calculates the real-time passage cost, which consists of a fixed geometric distance and the obstacle density coefficient at the current moment. The obstacle density coefficient increases with the area occupied along the path segment. After calculating the cost of all connected path segments, the autonomous path learning module traverses the set of all feasible connected paths from the current node to the target cabinet node and sorts the accumulated passage costs of each path segment as the total cost of the entire path. Finally, the autonomous path learning module selects the path with the minimum total cost from the set as the updated optimal path. This calculation process can quickly generate new path planning results when the local or global environment changes, enabling the delivery robot and retrieval robot units to autonomously navigate in dynamic scenarios without human intervention.

[0081] In one possible embodiment, when the same pickup instruction corresponds to multiple packages, the pickup robot unit calculates the optimal pickup order:

[0082]

[0083] In the formula, This represents the total number of packages awaiting pickup. The coordinates of the cabinet that the robot has currently reached or is about to reach. The coordinates of the next cabinet the robot should go to. Let be the Euclidean distance function between two points. The set of all possible pickup paths. The optimal order for picking up items is the one that minimizes the total moving distance.

[0084] Specifically, when the pickup robot unit identifies a pickup request involving multiple packages using a locally stored "cabinet code - package barcode" mapping table, its internal path optimization module treats the spatial coordinates of the target cabinets containing the multiple packages as a set of multiple nodes. The path optimization module uses an enumeration strategy to traverse all possible access sequences within this set, using the coordinate differences between adjacent nodes in each sequence as a calculation term. It then sums all adjacent distances for each sequence to obtain the total travel distance for that sequence. After generating the calculation results for all sequences, the path optimization module determines the sequence with the minimum travel distance as the optimal pickup order and uses this as the actual path for the pickup robot unit to perform the grasping action. This optimization method based on enumeration and comparison effectively avoids repeated backtracking due to an unreasonable pickup order, significantly reducing the execution time and robot energy consumption of the entire pickup process when a single user has multiple packages to pick up.

[0085] In one possible embodiment, the pickup robot unit is equipped with an instruction parsing module for receiving and parsing externally input pickup instructions. The pickup instructions include at least one type of identification information, such as barcode data, QR code data, or the last four digits of a mobile phone number. The pickup robot unit retrieves and matches the corresponding target cabinet based on the identification information in a locally stored database of correspondences between cabinet codes and parcel barcodes. Each code in the cabinet coding system includes the row and column coordinates of the cabinet and the internal length, width, and height volume parameters of the cabinet.

[0086] Specifically, the pickup robot unit includes an instruction parsing module for signal interface conversion. This module uniformly converts and adapts various pickup instruction formats from external input. In actual deployment, users may use barcode scanning results from handheld terminals, QR code images on mobile phone screens, or directly enter the last three digits of their mobile phone number on the device's keypad. These three signal sources each have different physical carriers and encoding protocols. The instruction parsing module can simultaneously support these three input forms and convert any form of input content into a unified internal query identifier. This identifier is directly sent to a locally pre-stored "cabinet code - parcel barcode" correspondence table for precise retrieval. Since the entire retrieval process only involves reading and writing operations within the pickup robot unit's internal storage space and does not rely on a remote server communication link, the target cabinet can be locked and located instantly upon user interaction. At the same time, since each code in the cabinet coding system has integrated the row and column coordinates of the cabinet and its internal length, width, height and volume parameters, the retrieval robot unit can directly obtain the three-dimensional spatial position information of the cabinet when it retrieves the identity code of the target cabinet, providing a complete and accurate spatial coordinate reference for the path planning of the autonomous path learning module and the grasping action of the robotic arm.

[0087] See Figure 3 The present invention also provides a method for storing and retrieving express mail using the system of the above embodiments, including an express mail inbound process and an express mail outbound process;

[0088] The express mail receiving process includes:

[0089] S11. The delivery vehicle arrives at the station entrance and unloads the batch of parcels into the unloading area;

[0090] S12. After the rough inspection robotic arm senses the package, it picks up the package one by one and places it on the carrier frame at the scanning station.

[0091] S13. The visual scanning unit collects images of the package, calculates its volume, determines the appropriate cabinet size, and transmits the package information to the delivery robot.

[0092] S14. After receiving the package information, the delivery robot matches the available target cabinet with the local cabinet data, plans the route through the autonomous path learning module, stores the package in the target cabinet and updates the local status.

[0093] The express delivery outbound process includes:

[0094] S21. The pickup robot unit receives an external pickup trigger signal containing package identification information;

[0095] S22. The package pickup robot unit matches the corresponding cabinet code of the package based on the package identification information;

[0096] S23. The pickup robot plans its path through the autonomous path learning module, reaches the target cabinet, and picks up the package; if the package identification information corresponds to multiple packages, the pickup is completed sequentially according to the optimal path.

[0097] S24. The pickup robot transports the grabbed package to the pickup port and triggers the pickup port to open to execute the package output action.

[0098] The inbound process begins with the delivery vehicle unloading a batch of packages into the unloading area of ​​the pre-inspection unit. This is followed by the pre-inspection robotic arm picking up and smoothly transferring the unordered packages one by one; the visual scanning unit identifying the label barcode and measuring the outer envelope volume of each package; and the delivery robot unit matching the target cabinet and planning the path based on the locally stored cabinet coding system and real-time occupancy data, until the packages are stored in the target cabinet. The outbound process begins with the pickup robot unit receiving an external trigger signal containing package identification information. This is followed by the instruction parsing module normalizing multi-format input signals and retrieving them from the local database; the pickup robot unit then plans and executes the path using its autonomous path learning module; finally, the package is transported to the pickup port, triggering an output action. These two work sequences are fully allowed to run concurrently on the timeline. The delivery robot unit and the pickup robot unit each execute independent storage and retrieval tasks based on the locally stored cabinet coding system and the dynamic navigation information provided by the autonomous path learning module. The two do not occupy each other's computing resources nor do they need to share any centralized scheduling nodes, thus ensuring that the unattended express station can still maintain continuous and stable all-weather operation capability under high-frequency concurrent storage and retrieval requests.

[0099] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A multi-robot coordinated unattended express post station system, characterized in that, include: The system includes a package pre-inspection unit, a visual scanning unit, a delivery robot unit, a pickup robot unit, a cabinet coding system, and an autonomous path learning module. The express mail preliminary inspection unit is located at the entrance of the station. The express mail preliminary inspection unit receives a batch of express mail and picks them up one by one and transfers them to the scanning station. The visual scanning unit is deployed above the scanning station. The visual scanning unit collects images of the express package and calculates the volume of the express package. Based on the volume, it matches the appropriate cabinet specifications, generates express package information, and sends it to the delivery robot unit. The delivery robot unit locally stores all cabinet codes and real-time occupancy status data. The delivery robot unit receives package information, matches an available target cabinet with suitable specifications, plans a path to store the package in the target cabinet, and then updates the local data. The pickup robot unit locally stores the correspondence data between cabinet codes and parcel barcodes. The pickup robot unit receives the user's pickup instruction, matches the corresponding cabinet code where the parcel is located, plans the route, grabs the parcel, and transports it to the pickup port for delivery to the user. The cabinet coding system assigns a unique code to all cabinets in the station, and each code contains the spatial location coordinates and volume specifications of the corresponding cabinet. The autonomous path learning module is mounted on the delivery robot and the pickup robot. The autonomous path learning module perceives the environment and dynamically plans the travel path.

2. The multi-robot coordinated unattended delivery kiosk system of claim 1, wherein, The pre-inspection unit for express mail includes an unloading area and a pre-inspection robotic arm. The pre-inspection robotic arm is equipped with an express mail sensing module. After detecting the arrival of an express mail, the pre-inspection robotic arm picks up the express mail one by one and transfers it to the carrier rack at the scanning station.

3. The multi-robot coordinated unattended delivery kiosk system of claim 1, wherein, The visual scanning unit includes a multi-phase camera array, arranged in a ring above the scanning station support frame, used to acquire images of express packages from multiple angles and identify the barcode information on the express package label.

4. The multi-robot collaborative unmanned express delivery station system according to claim 1, characterized in that, The formula used by the visual scanning unit to calculate the three-dimensional volume of the package is: ， In the formula, Indicates the volume of the package. This indicates the maximum length of the package as measured by 3D reconstruction using a multi-phase camera. Indicates the maximum width. Indicates the maximum height; The visual scanning unit will It is sent to the delivery robot unit as part of the package information.

5. The multi-robot collaborative unmanned express delivery station system according to claim 4, characterized in that, The formula for matching the delivery robot unit to an idle and suitable target cabinet is as follows: ， In the formula, This is the set of all currently available cabinets. For cabinet index, For the first The internal volume specifications of each locker; among all available lockers with sufficient volume to accommodate parcels, the locker with the smallest volume difference is selected as the target locker. .

6. The multi-robot collaborative unmanned express delivery station system according to claim 1, characterized in that, The autonomous path learning module performs path planning using the A* algorithm model, with the following cost function: ， In the formula, For the current search node, From the starting point to the current node The actual cost of movement, To start from the current node Heuristic cost estimation to the target node; the autonomous path learning module perceives environmental obstacles in real time and dynamically updates the search space. and parameter.

7. The multi-robot collaborative unmanned express delivery station system according to claim 6, characterized in that, When the autonomous path learning module detects a temporary obstacle, it updates the path cost using a dynamic weight adjustment formula: ， ， In the formula, For path segments, For the current moment, This represents the geometric distance of the path segment. For a moment Obstacle density coefficient on the next path segment, The preset obstacle avoidance weight coefficient; P represents a candidate path; Indicates the starting point from the current robot position. To the end of the target cabinet The set of all connected paths; This is the optimal path that minimizes the total cost after the update.

8. The multi-robot collaborative unmanned express delivery station system according to claim 1, characterized in that, When the same pickup instruction corresponds to multiple packages, the pickup robot unit calculates the optimal pickup order: ， In the formula, This represents the total number of packages awaiting pickup. The coordinates of the cabinet that the robot has currently reached or is about to reach. The coordinates of the next cabinet the robot should go to. Let be the Euclidean distance function between two points. The set of all possible pickup paths. The optimal order for picking up items is the one that minimizes the total moving distance.

9. The multi-robot collaborative unmanned express delivery station system according to claim 1, characterized in that, The pickup robot unit is equipped with an instruction parsing module for receiving and parsing externally input pickup instructions. The pickup instructions include at least one type of identification information, such as barcode data, QR code data, or the last four digits of a mobile phone number. The pickup robot unit retrieves and matches the corresponding target cabinet based on the identification information in the locally stored correspondence data between cabinet codes and parcel barcodes. Each code in the cabinet coding system includes the row and column coordinates of the cabinet and the internal length, width, and height volume parameters of the cabinet.

10. A method for storing and retrieving express mail implemented using the system described in any one of claims 1 to 9, characterized in that, This includes the parcel receiving process and the parcel dispatch process; The express mail receiving process includes: S11. The delivery vehicle arrives at the station entrance and unloads the batch of parcels into the unloading area; S12. After the rough inspection robotic arm senses the package, it picks up the package one by one and places it on the carrier frame at the scanning station. S13. The visual scanning unit collects images of the package, calculates its volume, determines the appropriate cabinet size, and transmits the package information to the delivery robot. S14. After receiving the package information, the delivery robot matches the available target cabinet with the local cabinet data, plans the route through the autonomous path learning module, stores the package in the target cabinet and updates the local status. The express delivery outbound process includes: S21. The pickup robot unit receives an external pickup trigger signal containing package identification information; S22. The package pickup robot unit matches the corresponding cabinet code of the package based on the package identification information; S23. The pickup robot plans its path through the autonomous path learning module, reaches the target cabinet, and picks up the package; if the package identification information corresponds to multiple packages, the pickup is completed sequentially according to the optimal path. S24. The pickup robot transports the grabbed package to the pickup port and triggers the pickup port to open to execute the package output action.

Images