Multi-modal data-based online supermarket unmanned warehouse intelligent sorting path optimization method

Abstract

The invention relates to an online supermarket unmanned warehouse intelligent sorting path optimization method based on multi-modal data, and belongs to the technical field of unmanned warehouse sorting, and the method comprises the following steps: S1, data real-time collection: deploying and installing sensing equipment in each area of a warehouse to collect commodity stock; s2, path calculation: constructing a dynamic electronic map, dividing a warehouse into grid units, and marking the state of each area with colors; an elastic priority path algorithm is executed, and paths are selected in different modes for conventional orders and urgent orders; s3, instruction issuing and execution: scheduling management is carried out, passing quantity and time limits are set at a main road intersection where the sorting robot passes, so that congestion at the main road intersection is avoided, and a one-way circulation mode is started in a peak period; and S4, abnormal problem processing, wherein different coping strategies are adopted according to faults of the single sorting robot, channel blockage and system faults.

Description

Technical Field

[0001] The present invention relates to the field of unmanned warehouse sorting technology, and in particular to an intelligent sorting path optimization method for an unmanned warehouse in an online supermarket based on multimodal data. Background Art

[0002] The existing unmanned warehouse sorting system has the following problems: path planning relies on manually preset routes and cannot respond to order changes in real time; it does not take into account the actual operating status of the equipment, such as robot congestion in a certain area and the occupancy of charging piles; environmental factors affect processing delays, such as still navigating according to the old map after the shelf is moved. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent sorting path optimization method for unmanned warehouses in online supermarkets based on multimodal data to solve the problems raised in the background technology.

[0004] In order to achieve the above-mentioned object of the invention, the present invention provides an intelligent sorting path optimization method for an online supermarket unmanned warehouse based on multimodal data, the method comprising the following steps:

[0005] S1: Real-time data collection

[0006] Sensor equipment is deployed and installed in various areas of the warehouse. Pressure sensors are installed on shelves to measure the inventory of goods at corresponding locations and report the inventory once. Sorting robots are equipped with dual-mode positioning modules to transmit coordinates and remaining battery power in real time. Infrared counters are set up at aisle intersections to count the number of sorting robots passing through in each time period.

[0007] Connect to the business system to obtain dynamic data, synchronize with the order system every 30 seconds, identify expedited orders, and receive promotional product lists pushed by the marketing system;

[0008] S2: Path Calculation

[0009] Build a dynamic electronic map, divide the warehouse into grid units, and use colors to mark the status of each area. The status corresponds to the color, including unimpeded, slow, and congested.

[0010] A flexible priority path algorithm is implemented. For regular orders, the comprehensive cost of all feasible paths is calculated: comprehensive cost = distance weight × 60% + congestion coefficient × 30% + robot battery charge × 10%. For expedited orders, a straight-line distance priority mode is enabled, automatically clearing other sorting robots on the path to alternate channels. If obstacles are encountered, the goods are relayed by adjacent sorting robots. For promotional pre-orders, batch collection paths are generated in advance, using a closed circular route, and a dedicated sorting robot is reserved for each promotional item.

[0011] S3: Instruction issuance and execution

[0012] Implement scheduling management and set traffic volume and time limits at intersections on main roads where sorting robots pass to avoid congestion. During peak hours, a one-way circulation mode will be activated.

[0013] Execute predictive path planning. When the sorting robot completes 80% of the current task, it calculates the return path in advance. Set up dedicated channels for high-frequency picking areas to improve the picking channels for high-frequency goods.

[0014] S4: Abnormal problem handling

[0015] If a single sorting robot fails, the robots within 3 meters of the fault point will be paused, the nearest idle robot will be controlled to take over the task, and the electronic map will be updated to mark the fault area; if the channel is blocked, a circular detour path will be automatically generated to avoid congestion of the picking robots, the speed of the sorting robots in the area will be controlled to be reduced, and an audible and visual alarm will be triggered to prompt manual inspection; if the system fails, it will first switch to the backup communication frequency band, enable the last valid path in the offline path memory, and start the manual takeover interface.

[0016] Furthermore, when performing path calculation in step S2, after receiving the order, the system first determines whether the order is an expedited order. If it is an expedited order, it scans the straight path from the target shelf to the packaging area, and sends a pull-over instruction to all sorting robots on the path to make way for the picking robot of the current expedited order, and then unlocks the speed of the picking robot of the current expedited order to allow it to quickly complete the picking of the expedited order.

[0017] Furthermore, when performing path calculation in step S2, after receiving the order, the system first determines whether the order is an expedited order. If it is not an expedited order but a regular order, the historical path database is called to select the three paths with the highest success rate in the past hour. Then, congested paths are eliminated based on real-time congestion data, and the optimal path is calculated according to the comprehensive cost value, so that the sorting robot can go to the target shelf to pick up goods according to the optimal path.

[0018] Furthermore, when the instruction in step S3 is issued and executed, the number of sorting robots passing in one direction is detected at the main road intersection. When the number reaches a certain value, the sorting robots in that direction are temporarily stopped to allow the sorting robots in the other direction of the main road intersection to pass. After a certain number of sorting robots in the other direction of the main road intersection have passed, the sorting robots in that direction are stopped again, and the above steps are repeated to avoid congestion at the main road intersection.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] Compared with the existing technology, the technical solution of the present application can improve sorting efficiency, reduce the idle rate of sorting robots, efficiently process expedited orders, and reduce collision accidents of sorting robots. DETAILED DESCRIPTION

[0021] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention are clearly and completely described below. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. The embodiments of the present invention are described below.

[0022] A method for optimizing intelligent sorting paths in an online supermarket unmanned warehouse based on multimodal data includes the following steps:

[0023] Step 1: Real-time data collection

[0024] Sensing equipment is deployed in various areas of the warehouse, and pressure sensors are installed on unmanned warehouse shelves. These sensors can be set to report product inventory every 10 seconds, such as five boxes of paper towels remaining in section A3 of a shelf. Sorting robots are equipped with dual-mode positioning modules, including UWB ultra-wideband (UWB) and visual recognition, which transmit the coordinates and remaining battery power of the sorting robots in real time. Infrared counters are installed at the intersections of sorting robot aisles to count the number of sorting robots passing through every five minutes.

[0025] Specifically, a high-precision pressure sensor can be installed under each shelf. For example, when the number of mineral water bottles on the third layer of area B of the shelf decreases from 20 boxes to 15 boxes, the sensor immediately sends a data packet B3-Mineral Water-15 through the LoRa wireless module.

[0026] Specifically, a coin-sized UWB positioning tag is installed on the top of each sorting robot, operating at a frequency of 6.5GHz. A positioning base station is arranged every 10 meters on the warehouse ceiling to form a triangulated positioning network. At the same time, a bottom camera is enabled to scan the QR code on the ground. A unique code is printed in the center of each floor tile to achieve double-safety positioning.

[0027] Specifically, channel monitoring can be achieved by hanging an infrared array sensor above the intersection, similar to the automatic door sensor in supermarkets. If more than three robots are detected passing through within five consecutive seconds, the area can be marked as a yellow warning, i.e. a slow-moving zone.

[0028] Connect to the business system to obtain dynamic data, synchronize with the order system every 30 seconds, and identify expedited orders; receive promotional product lists pushed by the marketing system, such as orders for buy-one-get-one-free promotions;

[0029] Step 2: Path calculation phase

[0030] By building a dynamic electronic map, the warehouse can be divided into 1m×1m or 2m×2m grid units, and the status of each area can be marked with a color. For example, green indicates a smooth area, yellow indicates a slow-moving area, and red indicates a congested area.

[0031] Specifically, for example, a 5,000 m2 warehouse floor can be divided into 50×100 grids, with each grid measuring 1m×1m in size. These grids are named using numerical codes, such as G37, which represents the grid in the 3rd row and 7th column.

[0032] The elastic priority path algorithm is used to calculate the comprehensive cost of all feasible paths for regular orders. This comprehensive cost is used to determine the picking path for the sorting robot for regular orders. The comprehensive cost = distance weight × 60% + congestion coefficient × 30% + robot battery power × 10%. For example, if path A has a distance of 15 meters, a congestion coefficient of 0.2, and a remaining battery power of 80%, the score for path A is: 15 × 0.6 + 0.2 × 30 + 80 × 0.1 = 9 + 6 + 8 = 23. If another path has a lower comprehensive cost, the other path is selected.

[0033] Specifically, the regular order processing process can be:

[0034] Enter the target shelf number, such as D12, and the current sorting robot position, for example, the sorting robot is located at G24, and the system generates three alternative routes, such as:

[0035] Route 1: G24→G25→G26...→G37. This route is the shortest but passes through a congested area.

[0036] Route 2: Go around the outer edge of the charging area, walking 8 meters longer but remaining unobstructed throughout;

[0037] Route 3: Use the reverse channel and wait 15 seconds for the traffic light to switch;

[0038] The score of each route is calculated according to the formula: comprehensive cost value = distance weight × 60% + congestion coefficient × 30% + robot power × 10%; in other implementations of this embodiment, the comprehensive cost value score can be: distance weight (40%) + smooth traffic score (30%) + power score (20%) + task urgency (10%). For example, the total score of route 2 is 82 × 0.4 + 95 × 0.3 + 70 × 0.2 + 60 × 0.1 = 79.1;

[0039] For expedited orders, the straight-line distance priority mode is enabled, automatically clearing other robots on the path to the backup channel. If an obstacle is encountered, the adjacent robot will relay the goods to complete the sorting.

[0040] Specifically, a special channel can be set up for expedited orders, triggering a red alarm light. The LED light strip on the top of the warehouse will turn flashing red, clearing the straight path from the target shelf to the packaging area. A pull-over instruction will be sent to all robots on the path, and the shelf position will be temporarily adjusted. For example, the electric shelf will automatically move 20cm to make room. The sorting robot will be accelerated, for example, the speed will be increased from the normal 1.5m / s to 2.2m / s.

[0041] For promotional pre-orders, batch collection routes can be generated 12 hours in advance, and circular closed routes can be used for sorting promotional pre-orders to avoid sorting congestion caused by the excessive volume of promotional pre-orders. Three dedicated robots can be reserved for each promotional item to sort separately from other orders, improving the sorting efficiency of the order.

[0042] Step 3: Instruction issuance and execution

[0043] Implement scheduling management and set traffic volume and time limits at main road intersections where sorting robots pass to avoid congestion. For example, after five sorting robots pass through in a row in one direction, traffic will be switched to the other direction for a certain period of time, or five sorting robots will pass through the other direction as well. If the time is up, the robot will automatically switch to avoid congestion of sorting robots in the other direction due to a small number of robots passing in one direction. During peak hours, a one-way circulation mode will be activated, such as when the order volume exceeds 500 orders per hour, to improve sorting efficiency and avoid congestion.

[0044] Execute predictive path planning. When the sorting robot completes 80% of the current task, it calculates the return route in advance. Set up dedicated channels for high-frequency picking areas to improve the picking channels for high-frequency goods. For example, the east channel is open from 8:00 a.m. to 10:00 a.m., and the west channel is switched from 8:00 p.m. to 10:00 p.m.

[0045] Step 4: Exception handling phase

[0046] Set up three-level emergency response:

[0047] A single robot failure can be a Level 1 response. In this case, the sorting robots within 3 meters of the fault point can be controlled to pause, and the nearest idle sorting robot will take over the task. The electronic map will be updated to mark the fault area.

[0048] Channel blockage can be a secondary response. In this case, a circular detour path can be automatically generated to avoid congestion, reduce the robot speed in the area to 0.5m / s, and trigger an audible and visual alarm to prompt manual inspection.

[0049] System failure can be set to a third-level response, at which time you can switch to the backup communication band, enable the last valid path in the offline path memory, and start the manual takeover interface.

[0050] In some embodiments, when a new order is received, it is determined whether it is an expedited order. If it is an expedited order, the straight path from the target shelf to the packaging area is scanned, and a pull-over instruction is sent to all robots on the path, unlocking the maximum speed permission for the current robot, for example, increasing the speed from 1.5m / s to 2.2m / s;

[0051] If it is not an urgent order, the historical path database is called to select the three paths with the highest success rate in the past hour. Unavailable paths are eliminated based on real-time congestion data, and the optimal path is automatically selected according to the calculation formula of the comprehensive cost value.

[0052] The technical solutions of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the above descriptions are merely for the purpose of explaining the solutions of the present invention and are not to be construed in any way as limiting the scope of protection of the invention. Based on the explanations herein, those skilled in the art can conceive of other specific embodiments of the present invention or equivalent replacements without inventive effort, and these will fall within the scope of protection of the present invention.

Claims

1. A method for optimizing intelligent sorting paths in online supermarket unmanned warehouses based on multimodal data, characterized in that: The method comprises the following steps: S1: Real-time data collection Sensor equipment is deployed and installed in various areas of the warehouse. Pressure sensors are installed on shelves to measure the inventory of goods at corresponding locations and report the inventory once. Sorting robots are equipped with dual-mode positioning modules to transmit coordinates and remaining battery power in real time. Infrared counters are set up at aisle intersections to count the number of sorting robots passing through in each time period. Connect to the business system to obtain dynamic data, synchronize with the order system every 30 seconds, identify expedited orders, and receive promotional product lists pushed by the marketing system; S2: Path Calculation Build a dynamic electronic map, divide the warehouse into grid units, and use colors to mark the status of each area. The status corresponds to the color, including unimpeded, slow, and congested. A flexible priority path algorithm is implemented. For regular orders, the comprehensive cost of all feasible paths is calculated: comprehensive cost = distance weight × 60% + congestion coefficient × 30% + robot battery charge × 10%. For expedited orders, a straight-line distance priority mode is enabled, automatically clearing other sorting robots on the path to alternate channels. If obstacles are encountered, the goods are relayed by adjacent sorting robots. For promotional pre-orders, batch collection paths are generated in advance, using a closed circular route, and a dedicated sorting robot is reserved for each promotional item. S3: Instruction issuance and execution Implement scheduling management and set traffic volume and time limits at intersections on main roads where sorting robots pass to avoid congestion. During peak hours, a one-way circulation mode will be activated. Execute predictive path planning. When the sorting robot completes 80% of the current task, it calculates the return path in advance. Set up dedicated channels for high-frequency picking areas to improve the picking channels for high-frequency goods. S4: Abnormal problem handling If a single sorting robot fails, the robots within 3 meters of the fault point will be paused, the nearest idle robot will be controlled to take over the task, and the electronic map will be updated to mark the fault area; if the channel is blocked, a circular detour path will be automatically generated to avoid congestion of the picking robots, the speed of the sorting robots in the area will be controlled to be reduced, and an audible and visual alarm will be triggered to prompt manual inspection; if the system fails, it will first switch to the backup communication frequency band, enable the last valid path in the offline path memory, and start the manual takeover interface.

2. The method for optimizing intelligent sorting paths in online supermarket unmanned warehouses based on multimodal data according to claim 1 is characterized in that: When performing path calculation in step S2, after receiving the order, the system first determines whether the order is an expedited order. If it is an expedited order, it scans the straight path from the target shelf to the packaging area, sends a pull-over instruction to all sorting robots on the path, makes way for the picking robot of the current expedited order, and then unlocks the speed of the picking robot of the current expedited order to allow it to quickly complete the picking of the expedited order.

3. The method for optimizing intelligent sorting paths in online supermarket unmanned warehouses based on multimodal data according to claim 1 is characterized in that: When performing path calculation in step S2, after receiving the order, the system first determines whether the order is an expedited order. If it is a regular order, it calls the historical path database and selects the three paths with the highest success rate in the past hour. Then, it eliminates congested paths based on real-time congestion data, calculates the optimal path based on the calculation of the comprehensive cost value, and lets the sorting robot go to the target shelf to pick up the goods according to the optimal path.

4. The method for optimizing intelligent sorting paths in online supermarket unmanned warehouses based on multimodal data according to claim 1 is characterized in that: When the instruction in step S3 is issued and executed, the number of sorting robots passing in one direction is detected at the main road intersection. When the number reaches a certain value, the sorting robots in that direction are temporarily stopped to allow the sorting robots in the other direction of the main road intersection to pass. After a certain number of sorting robots in the other direction of the main road intersection have passed, the sorting robots in that direction are stopped again. The above steps are repeated to avoid congestion at the main road intersection.