Intelligent warehousing multi-layer robot system based on cloud-edge collaboration

Abstract

The present disclosure relates to an intelligent warehousing multi-layer robot system based on cloud-edge collaboration, comprising: multiple three-dimensional shelves, a storage robot, an order sorting device and a cloud-edge warehouse management server; different three-dimensional layers of the multiple three-dimensional shelves are used to store storage goods of different specifications and sizes, and the multiple three-dimensional shelves are respectively used to store storage goods of different types; the cloud-edge warehouse management server is used to respond to the handling request of the order sorting device, determine the target storage information of the storage goods to be handled, determine the target storage robot according to the target storage information and the robot information of the storage robot, estimate the position information of the target storage robot that transports other storage goods based on the adaptive Monte Carlo positioning method, and plan the driving path of the target warehouse robot that transports the storage goods to be handled according to the position information and the target storage information; the warehouse robot is used to transport the storage goods to be handled according to the received target storage information and the driving path.

Description

Technical Field

[0001] The present disclosure relates to the field of intelligent logistics technology, and in particular to an intelligent warehousing multi-layer robot system based on cloud-edge collaboration. Background Art

[0002] With the advent of the intelligent era, unmanned or less-manned operations have become the trend and direction of the development of warehousing logistics and factory logistics. As the cost of warehouse land continues to increase, how to achieve high-level storage and increase the storage capacity of warehouses has become an important consideration in the warehousing industry. At the same time, how to analyze traditional and intelligent warehousing logistics through six dimensions such as space utilization, storage volume, and storage form, reduce the labor demand for warehousing logistics, reduce the error rate of goods picking, improve the level and efficiency of warehousing management, and improve space utilization and storage efficiency has become an urgent problem to be solved. Summary of the invention

[0003] In order to overcome the technical problems in related technologies such as high labor costs in warehousing logistics, high error rates in cargo picking, low levels and efficiency of warehousing management, and low space utilization and storage efficiency, the present disclosure provides an intelligent warehousing multi-layer robot system based on cloud-edge collaboration.

[0004] According to an embodiment of the present disclosure, a smart warehousing multi-layer robot system based on cloud-edge collaboration is provided, including:

[0005] Multiple three-dimensional shelves, storage robots, order sorting devices and cloud-edge warehouse management servers;

[0006] Wherein, different three-dimensional layers of the plurality of three-dimensional shelves are used to store warehouse goods of different specifications and sizes, and the plurality of three-dimensional shelves are respectively used to store the warehouse goods of different types;

[0007] The cloud edge warehouse management server is used to determine the target storage information of the stored goods to be transported in response to the transport request sent by the order sorting device to the cloud edge warehouse management server for the stored goods, determine the target storage robot used to transport the stored goods to be transported according to the target storage information and the robot information of the storage robot, and estimate the position information of the target storage robot corresponding to the other stored goods for positioning and transporting based on the adaptive Monte Carlo positioning method, and plan the driving path of the target storage robot corresponding to the stored goods to be transported according to the position information and the target storage information;

[0008] The warehouse robot is used to, when being determined as the target warehouse robot, receive the target warehouse information and the driving path sent by the cloud edge warehouse management server, and transport the warehouse goods to be transported according to the target warehouse information and the driving path.

[0009] In a possible implementation, the cloud edge warehouse management server is used to estimate the position information of the target warehouse robot corresponding to the other warehouse goods based on the adaptive Monte Carlo positioning method, and plan the driving path of the target warehouse robot corresponding to the warehouse goods to be transported according to the position information and the target warehouse information, including:

[0010] The cloud edge warehouse management server is used to determine the number of path collision points that exist during the transportation process of the target storage robot that transports other stored goods according to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, wherein the path collision point is a coordinate point where the driving path of the storage robot intersects;

[0011] Determine the number of collision points on the path as the number of particles in the adaptive Monte Carlo positioning method used to estimate the position of each of the warehouse robots, and estimate the position information of the target warehouse robots corresponding to the other warehouse goods based on the number of particles;

[0012] According to the target warehouse information and the location information of the order sorting device, a transport route of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing the number of collision points on the path, and a transport speed of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing collisions with the particles;

[0013] According to the transport route and the transport speed, a driving path of the target storage robot corresponding to the warehouse goods to be transported is obtained.

[0014] In one possible implementation, the cloud edge warehouse management server is used to plan the transport route of the target warehouse robot corresponding to the warehouse goods to be transported based on the target warehouse information and the location information of the order sorting device, with the principle of minimizing the number of path collision points, including:

[0015] The cloud edge warehouse management server is used to construct a graph theory model corresponding to the warehouse space according to the first occupation location information of the plurality of three-dimensional shelves and the second occupation location information of the order sorting device, wherein the nodes represent possible path points, the edges represent path segments, and the weights on the edges represent path costs;

[0016] According to the target storage information and the location information of the order sorting device, using the shortest path algorithm, calculate the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the warehouse goods to be transported;

[0017] Performing collision detection between the preliminary shortest path and particles corresponding to the target storage robot that is transporting other stored goods during the transport process;

[0018] If the preliminary shortest path collides with the particles corresponding to the target storage robot carrying other stored goods during the transportation process, the nodes or edges in the preliminary shortest path are adjusted and the path is recalculated until the path with the least number of collision points is found as the final transportation route.

[0019] In one possible implementation, the cloud edge warehouse management server is used to calculate the preliminary shortest path from the transport starting point to the transport end point of the target warehouse robot corresponding to the warehouse goods to be transported using the shortest path algorithm according to the target warehouse information and the location information of the order sorting device, including:

[0020] The cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request, which is used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, and according to the target warehouse information and the location information of the order sorting device;

[0021] The transport starting point is set to the node with the known shortest path, and the shortest path lengths of the remaining nodes are set to infinity;

[0022] Repeat the following steps: select the node with the smallest current shortest path estimate from the unprocessed nodes as the current node; for each neighbor node of the current node, check whether the path from the current node to the neighbor node is shorter than the known shortest path; if so, update the shortest path length of the neighbor node and the predecessor node of the shortest path until all nodes have been processed, and obtain the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the storage goods to be transported.

[0023] In a possible implementation, the cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request and used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, according to the target warehouse information and the location information of the order sorting device, including:

[0024] The cloud edge warehouse management server is used to determine the position corresponding to the target storage information as the transportation starting point, and the position corresponding to the position information of the order sorting device as the transportation end point, when the transportation identifier carried in the transportation request and used to characterize the warehoused goods to be transported is outbound goods;

[0025] When the transport identifier carried in the transport request and used to characterize the warehouse goods to be transported is incoming goods, the position corresponding to the position information of the order sorting device is determined as the transport starting point, and the position corresponding to the target warehouse information is determined as the transport end point.

[0026] In one possible implementation, the cloud edge warehouse management server is used to plan the transport speed of the target storage robot corresponding to the warehouse goods to be transported based on the principle of minimum collision with the particles, including:

[0027] The cloud edge warehouse management server is used to dynamically predict the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process according to the number of the particles;

[0028] Determine whether there is a collision with the particle according to the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process and the current speed of the target storage robot corresponding to the stored goods to be transported;

[0029] If there is a collision, the current speed or acceleration of the target storage robot corresponding to the warehouse goods to be transported is adjusted based on the principle of minimum collision with the particles to obtain the planned transport speed of the target storage robot corresponding to the warehouse goods to be transported.

[0030] In a possible implementation, the cloud edge warehouse management server is used to determine the number of path collision points that exist during the transportation process of the target storage robot corresponding to the other stored goods according to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, including:

[0031] The cloud edge warehouse management server is used to determine the distance between two adjacent three-dimensional shelves according to the first occupied position information of the plurality of three-dimensional shelves, and determine the greatest common divisor of the distances as the side length of the voxel grid;

[0032] According to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, the space in the warehouse space not occupied by the three-dimensional shelves and the order sorting device is divided into a plurality of voxel grids according to the side length, each of the voxel grids having a unique identifier;

[0033] Mapping the driving path of each storage robot to the corresponding voxel grid, and recording the number of robots passing through each voxel grid;

[0034] All the voxel grids are traversed, and the voxel grids with more than one robot are used as collision voxel grids, and the number of the collision voxel grids is determined as the number of path collision points in the transportation process of the target storage robot corresponding to the other storage goods.

[0035] In a possible implementation, the storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, including:

[0036] The storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, and to monitor the relative position relationship between the target storage robot and the surrounding environment in real time by using sensors and image processing technology according to the physical size and turning radius of the storage robot and the size of the goods to be transported, and to determine whether there is a collision;

[0037] When it is determined that there is a risk of collision, the speed, direction or path of the storage robot is adjusted to avoid the collision until the task of transporting the goods to be transported is completed.

[0038] In a possible implementation, the target storage information includes: a target three-dimensional shelf and a target three-dimensional layer.

[0039] In a possible implementation, the robot information includes: robot position information, robot maximum carrying weight information, robot size information, and robot turning radius.

[0040] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects:

[0041] 1. By combining the order sorting device with the cloud-edge warehouse management server, the system can quickly respond to handling requests and accurately determine the target warehouse information of the goods to be transported. This automated process replaces the traditional manual sorting and transportation, greatly improving the efficiency of warehousing operations and reducing inaccurate classification problems caused by human errors. The cloud-edge warehouse management server uses advanced computing power and data storage capabilities to acquire and process a large amount of warehouse data in real time, including the storage location, specifications, dimensions, types and other information of the goods, which helps to achieve refined warehouse management and improve overall operational efficiency.

[0042] 2. Adopt the adaptive Monte Carlo positioning method to accurately estimate the position of the warehouse robot. This method can dynamically plan the optimal driving path according to the real-time position of the robot and the target warehouse information, avoid obstacles, improve the navigation accuracy and driving efficiency of the robot, support multi-layer stereoscopic shelves, and store goods of different specifications, sizes and types to meet diverse warehousing needs. At the same time, the warehouse robot can autonomously complete the cargo handling task according to the received target warehouse information and driving path, which enhances the flexibility and adaptability of the system.

[0043] 3. Under the cloud-edge collaborative architecture, the processing of some sensitive data can be completed on the edge side without the need to transmit all of it to the cloud. This helps reduce the risk of data leakage and improve the system's data security and privacy protection capabilities. Automated and intelligent warehouse management reduces dependence on manpower and reduces labor costs. At the same time, since the system can accurately and efficiently handle large amounts of goods, it reduces losses and waste caused by human errors, further reducing operating costs. By real-time updating of cargo information and robot status in the warehouse, it helps managers optimize warehouse layout and resource allocation, and improve warehouse space utilization and cargo turnover. In this way, through automated, intelligent and refined management methods, the warehouse efficiency, accuracy and flexibility are significantly improved, and operating costs are reduced.

[0044] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0046] Figure 1 It is a schematic diagram of an intelligent warehousing multi-layer robot system based on cloud-edge collaboration according to an exemplary embodiment.

[0047] Figure 2 It is a schematic diagram of a three-dimensional shelf 1 according to an exemplary embodiment.

[0048] Figure 3 is a schematic diagram of an order sorting device 3 according to an exemplary embodiment.

[0049] Figure 4 It is a schematic diagram of a storage / inloading workstation 4 according to an exemplary embodiment. DETAILED DESCRIPTION

[0050] Here, exemplary embodiments will be described in detail, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present disclosure.

[0051] Figure 1 is a schematic diagram of an intelligent warehousing multi-layer robot system based on cloud-edge collaboration according to an exemplary embodiment. Figure 1 As shown, including:

[0052] Multiple three-dimensional shelves 1, storage robots 2, order sorting devices 3 and cloud-edge warehouse management servers;

[0053] Among them, different three-dimensional layers of the multiple three-dimensional shelves are used to store warehouse goods of different specifications and sizes, and the multiple three-dimensional shelves are respectively used to store warehouse goods of different types; through the setting of multiple temporary storage locations, the access efficiency is improved, and the problems of low space utilization and low storage efficiency of existing warehousing solutions are solved.

[0054] In one embodiment, if Figure 1 As shown, the three-dimensional shelf 1 is planned to have two horizontal running channels 12 and three vertical running channels 13. Figure 2 As shown, in order to improve the space utilization rate of the warehouse space, the intelligent storage multi-layer robot system 10 constructed above has a double-row three-dimensional shelf 1, and the double-row three-dimensional shelf 1 is a multi-layer structure, and the height of the highest shelf can reach up to 10 meters, matching the height capability of the storage robot to pick up goods. The three-dimensional shelf 1 is constructed by a steel structure or aluminum profile structure such as a column sheet 100, a beam 101, and a layer plate 102. A lifting mechanism 11 for carrying warehouse goods up and down can be built between the three-dimensional shelves 1. During the construction process, the beam 101 is fixed on the column sheet 100, and the distance between the column sheets 100 can be adjusted according to the storage requirements of different material boxes; the layer plate 102 is fixed on the two beams 101, and there is a whole layer plate 102 between the two column sheets 100, which can store four material boxes, wherein the material box can be used to store warehouse goods.

[0055] The storage robot 2 is a robot that moves autonomously or under control in a warehouse environment, and is used to perform tasks such as handling, storing, and retrieving goods to improve warehouse operation efficiency. Specifically, the storage robot 2 is used to complete the automatic storage and retrieval tasks of material boxes, and includes a chassis, a storage and retrieval fork, and a plurality of temporary storage positions, wherein the lifting mechanism 11 is used to lift the fork to dock the material boxes and temporary storage positions of the three-dimensional shelf 1, and a maximum of eight temporary storage positions are set. During the warehousing operation, the storage robot 2 clamps the material boxes from the warehousing workstation and can temporarily store them in the temporary storage position. It can take up to nine material boxes at the same time, and then put them into the corresponding layers of the three-dimensional shelf 1 in sequence, which fully improves the storage and retrieval efficiency; the same applies to the outbound operation.

[0056] In this embodiment, the order sorting device 3 is an automated or semi-automated device or system, which is used to select specified goods from the warehouse according to order requirements and may trigger subsequent transportation requests, such as Figure 3 As shown, the order sorting device 3 is used for automated picking of orders of the same batch, so as to pick the material boxes to different order pallets according to different orders. The order sorting device 3 comprises a plurality of belt conveyor line platforms 30, a code scanning device 31, a linear belt sorting device 32, a grid chute 33 and an order pallet 34. The whole sorting process is as follows: all the material boxes of the same batch of orders are transported to the belt conveyor line platform 30 by the storage robot 2, and after being scanned by the code scanning device 31, they are allocated according to the information of the customer order by the linear belt sorting device 32, and transferred to the grid chute 33. The grid chute corresponds to a certain order, and then manually transferred from the chute to the order pallet next to it. Through the above modules, the same batch of orders are shipped out in a centralized manner, and the automatic sorting equipment completes the material box picking of the order, realizing the goods-to-person picking mode and improving the sorting efficiency of customer orders.

[0057] In this implementation example, Figure 1 and 4 As shown, the in / out storage workstation 4 includes two conveyor line platforms 40, an appearance detection device 41, a code scanning device 42, an elevator 43 and a multi-layer conveyor line platform 44, wherein the above two conveyor line platforms 40 are independent power sections, corresponding to the upper and lower line positions and cache positions of the material box respectively, and the function of the elevator 43 and the multi-layer conveyor line platform 44 used in conjunction is to provide multiple material box cache positions, and cooperate with the loading capacity of the storage robot 2 to fully improve the efficiency of in / out storage.

[0058] In this implementation example, the cloud edge warehouse management server is deployed on a warehouse management system server at the edge of cloud computing (i.e., close to the data source or user end) to process real-time data requests related to warehouse management, such as order sorting, cargo handling instructions, etc., to improve data processing speed and response efficiency. In addition, the cloud edge warehouse management server can be used to receive and process batch orders sent by the order system, as well as mobilize the three-dimensional shelf 1 and the in / out workstation 4 to carry out the in / out of the material box, and the order sorting device 3 to carry out order sorting. At the same time, the cloud edge warehouse management server can also have the function of monitoring and managing the above modules, such as being responsible for processing batch order information, mobilizing the work of various automated equipment, monitoring the working status of equipment, managing material box and cargo information, and managing cargo space allocation.

[0059] The cloud edge warehouse management server is used to determine the target storage information of the stored goods to be transported in response to the transport request sent by the order sorting device to the cloud edge warehouse management server for the stored goods, determine the target storage robot used to transport the stored goods to be transported according to the target storage information and the robot information of the storage robot, and estimate the position information of the target storage robot corresponding to the other stored goods for positioning and transporting based on the adaptive Monte Carlo positioning method, and plan the driving path of the target storage robot corresponding to the stored goods to be transported according to the position information and the target storage information;

[0060] Among them, the transport request is an instruction or request issued by the order sorting device to move specific warehouse goods from the current location to the specified location. The target warehouse information is used to describe the detailed location information where the warehouse goods to be transported should be transported, including but not limited to the shelf number, number of layers, grid positions, etc. The robot information can be a data set of the status, position, and capabilities (such as load-bearing capacity, speed) of the warehouse robot, which is used to decide which tasks to assign to which robots. The driving path is calculated based on the target warehouse information and the current location information. It is the optimal or feasible moving route for the warehouse robot from the current location to the target location.

[0061] In one embodiment, the cloud edge warehouse management server continuously monitors requests from the order sorting device through its network interface. When receiving a transport request, the server parses the data in the request, including the unique identifier of the warehoused goods to be transported (such as SKU, storage location number, etc.). For example, the order sorting device detects that an order requires a certain model of electronic components, and the component is currently stored in the B layer C grid in the A area. The sorting device sends a transport request containing the storage location information of the component to the cloud edge warehouse management server.

[0062] Furthermore, the cloud edge warehouse management server queries its database or cache based on the cargo information in the transport request to determine the target storage location (such as a shelf in another area, a shipping port, etc.) to which the cargo should be transported. For example, the cloud edge warehouse management server queries that the electronic component should be transported to the F grid of the E layer in the D area of ​​the shipping area for subsequent packaging and shipment.

[0063] Furthermore, the cloud-edge warehouse management server comprehensively considers the status of all current warehouse robots (such as idle, busy, power, location, etc.) and the distance of the target warehouse information, obstacles and other factors, and determines the target warehouse robot that is most suitable for performing this handling task through algorithms (such as optimization strategies such as the closest distance and the least time). For example, the cloud-edge warehouse management server finds through calculation that the H warehouse robot located in area G is close to both the starting position and the target position and is currently idle, so it is selected as the target warehouse robot.

[0064] Furthermore, in order to ensure the safety and efficiency of path planning, the cloud edge warehouse management server adopts an adaptive Monte Carlo positioning method to update the location information of all storage robots in the warehouse in real time or regularly. By simulating the movement of multiple particles in the warehouse, the weights of the particles are adjusted according to sensor data (such as RFID, LiDAR, etc.), and finally the precise position of each robot is estimated. For example, the position is updated every 5 minutes, the particle filter algorithm is used to simulate the movement of each robot, and the particle distribution is adjusted according to the latest sensor data to ensure accurate estimation of the positions of the two storage robots K and L in areas I and J.

[0065] Furthermore, based on the target warehouse information, the current location information of the target warehouse robot, and the map data of the warehouse (including shelf layout, aisle width, obstacle location, etc.), the cloud edge warehouse management server uses a path planning algorithm (such as A* algorithm, Dijkstra algorithm, etc.) to calculate the optimal driving path. For example, the cloud edge warehouse management server plans a shortest path for the H warehouse robot to pick up the goods from the current location to the C grid on the B floor of the A area, and then go to the F grid on the E floor of the D area to put the goods.

[0066] The warehouse robot is used to, when being determined as the target warehouse robot, receive the target warehouse information and the driving path sent by the cloud edge warehouse management server, and transport the warehouse goods to be transported according to the target warehouse information and the driving path.

[0067] In one embodiment, the warehouse robot receives the target warehouse information and driving path sent by the cloud edge warehouse management server through its communication module. After parsing this information, the robot uses its navigation system and drive system to autonomously move to the starting position according to the planned path, perform the pickup operation, and then move to the target position to place the goods. During the whole process, the robot may also need to use sensors (such as cameras, ultrasonic sensors) to avoid collisions and maintain the correct navigation direction.

[0068] For example, after receiving a handling task, the H warehouse robot first locates itself at the C grid on the B layer in area A through its built-in navigation system, uses a robotic arm or suction cup to remove the electronic components, and then drives along the planned path to the F grid on the E layer in area D, places the components at the designated location, and completes the handling task. Throughout the process, the robot continuously uses a laser radar to scan the surrounding environment to ensure safe driving.

[0069] By combining the order sorting device with the cloud-edge warehouse management server, the above solution enables the system to quickly respond to handling requests and accurately determine the target warehouse information of the goods to be transported. This automated process replaces the traditional manual sorting and transportation, greatly improving the efficiency of warehousing operations and reducing the problem of inaccurate classification caused by human errors. The cloud-edge warehouse management server uses advanced computing power and data storage capabilities to acquire and process a large amount of warehouse data in real time, including the storage location, specifications, size, type and other information of the goods. This helps to achieve refined warehouse management and improve overall operational efficiency.

[0070] Furthermore, the adaptive Monte Carlo positioning method is used to accurately estimate the position of the storage robot. This method can dynamically plan the optimal driving path according to the real-time position of the robot and the target storage information, avoid obstacles, and improve the navigation accuracy and driving efficiency of the robot. It supports multi-layer stereoscopic shelves and can store goods of different sizes and types to meet diverse storage needs. At the same time, the storage robot can autonomously complete the cargo handling task according to the received target storage information and driving path, which enhances the flexibility and adaptability of the system.

[0071] Furthermore, under the cloud-edge collaborative architecture, the processing of some sensitive data can be completed on the edge side without the need to transmit all of it to the cloud, which helps reduce the risk of data leakage and improves the system's data security and privacy protection capabilities. Automated and intelligent warehouse management reduces dependence on manpower and reduces labor costs. At the same time, since the system can accurately and efficiently handle large amounts of goods, it reduces losses and waste caused by human errors, further reducing operating costs. By updating the cargo information and robot status in the warehouse in real time, it helps managers optimize warehouse layout and resource allocation, and improve warehouse space utilization and cargo turnover. In this way, through automated, intelligent and refined management methods, the efficiency, accuracy and flexibility of warehousing are significantly improved, and operating costs are reduced.

[0072] In a possible implementation, the cloud edge warehouse management server is used to estimate the position information of the target warehouse robot corresponding to the other warehouse goods based on the adaptive Monte Carlo positioning method, and plan the driving path of the target warehouse robot corresponding to the warehouse goods to be transported according to the position information and the target warehouse information, including:

[0073] The cloud edge warehouse management server is used to determine the number of path collision points that exist during the transportation process of the target storage robot that transports other stored goods according to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, wherein the path collision point is a coordinate point where the driving path of the storage robot intersects;

[0074] In one embodiment, the cloud edge warehouse management server first obtains the precise location information of all three-dimensional shelves and order sorting devices in the warehouse (i.e., the first occupied location information and the second occupied location information). This information is usually expressed in coordinate form and stored in the server's database. Then, based on the current task allocation of all warehouse robots, their respective driving paths are simulated. During the simulation, it is checked whether these paths have intersections, i.e., path collision points. The number of path collision points reflects the complexity of robot activities in the warehouse and the potential risk of collision.

[0075] For example, suppose there are three three-dimensional shelves A, B, and C in the warehouse, located at (x1, y1), (x2, y2), and (x3, y3), respectively, and the order sorting device D is located at (xd, yd). Currently, two storage robots R1 and R2 are assigned handling tasks. The server simulates the driving paths of R1 and R2 and finds that they cross paths at (x_cross, y_cross), that is, there is a path collision point.

[0076] Determine the number of collision points on the path as the number of particles in the adaptive Monte Carlo positioning method used to estimate the position of each of the warehouse robots, and estimate the position information of the target warehouse robots corresponding to the other warehouse goods based on the number of particles;

[0077] In one embodiment, after determining the number of path collision points, the cloud edge warehouse management server dynamically adjusts the number of particles used in the adaptive Monte Carlo positioning method based on this number. Generally speaking, the more path collision points there are, the more complex the robot activities in the warehouse are, and the greater the positioning difficulty is, so more particles are needed to more accurately estimate the robot's position. The server sets a basic number of particles as a starting point, and then increases linearly or nonlinearly based on the number of path collision points.

[0078] For example, suppose the server sets the number of basic particles to 100, and for each additional path collision point, the number of particles increases by 50. In the above embodiment, since there is a path collision point, the server will use 150 particles to estimate the positions of R1 and R2 (especially the robot currently being transported).

[0079] After determining the number of particles, the cloud edge warehouse management server will start the adaptive Monte Carlo positioning algorithm. The algorithm first randomly generates a specified number of particles, each of which represents a possible position of the robot. Then, based on the robot's motion model (such as speed, acceleration, etc.) and sensor data (such as odometer, lidar, etc.), the algorithm updates the position of each particle and calculates the weight corresponding to each particle. The weight reflects the similarity between the particle position and the true position. Finally, the algorithm retains high-weight particles and eliminates low-weight particles through the resampling process, thereby gradually approaching the true position of the robot. In this way, the cloud edge warehouse management server can effectively determine the number of path collision points, and adjust the number of particles in the adaptive Monte Carlo positioning method accordingly, so as to more accurately estimate the position information of the target warehouse robot corresponding to the handling of other warehouse goods.

[0080] For example, taking R1 as an example, the server uses 150 particles to estimate its position. During the iteration of the algorithm, the server continuously updates the positions of these particles based on R1's motion model and sensor data, and calculates the weight of each particle. After many iterations, the server finds that a group of particles are gathered near (xr1_est, yr1_est), and these particles have higher weights. Therefore, the server believes that (xr1_est, yr1_est) is a good estimate of R1's current position.

[0081] According to the target warehouse information and the location information of the order sorting device, a transport route of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing the number of collision points on the path, and a transport speed of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing collisions with the particles;

[0082] In one embodiment, after determining the target storage location (i.e., the location of the three-dimensional shelf to which the stored goods need to be transported) and the location of the order sorting device, the cloud edge warehouse management server needs to plan a transportation route from the sorting device to the target storage location. The design of this route should follow the principle of "minimizing the number of path collision points", that is, avoiding intersections with existing or predicted paths of other storage robots as much as possible.

[0083] The server uses the warehouse map data, the location of existing warehouse robots and their movement trends (derived through prediction algorithms), and possible obstacle information in the warehouse to build an optimized path planning model. The model takes into account multiple factors, such as path length, number of turns, safety, etc., and uses algorithms (such as A* algorithm, Dijkstra algorithm or its variants) to find the optimal solution.

[0084] For example, suppose the target warehouse goods need to be transported from order sorting device D to three-dimensional shelf C, and there are already multiple warehouse robots in the current warehouse performing their respective tasks. By analyzing the warehouse map and real-time data, the server finds that if it goes directly from D to C, it will cross the path of the R2 robot that is heading to the three-dimensional shelf A. To avoid this collision, the server plans a slightly longer but safe route, so that R1 (assuming it is the robot carrying the target warehouse goods) can avoid the driving path of R2 while minimizing additional turns and distance.

[0085] In one embodiment, in the adaptive Monte Carlo positioning method, particles represent possible locations of the robot. Since the distribution and movement of particles are based on the robot's motion model and sensor data, they will reflect the locations of other dynamic objects in the warehouse to a certain extent (although probabilistically). For safety reasons, the cloud edge warehouse management server will consider the potential collision risk with these "particles" when planning the handling speed of the target warehouse robot.

[0086] Specifically, the server simulates the robot's trajectories at different speeds and calculates the intersection of these trajectories with the particle distribution. By adjusting the speed, the server can find a balance point, that is, while ensuring the handling efficiency, it can minimize the risk of collision with particles (that is, potential obstacles or robots).

[0087] For example, when planning R1's transport speed, the server found that if it drove at the maximum speed, its trajectory would overlap with multiple high-weight particles (i.e., particles that are more likely to represent real objects). For safety reasons, the server reduced R1's transport speed and re-simulated its trajectory. After reducing the speed, the overlap between R1's trajectory and the particles was significantly reduced, thereby reducing the risk of collision.

[0088] According to the transport route and the transport speed, a driving path of the target storage robot corresponding to the warehouse goods to be transported is obtained.

[0089] In one embodiment, after completing the planning of the transport route and transport speed, the cloud edge warehouse management server will combine these two elements to generate a complete driving path for the target storage robot (such as R1). This path includes all necessary information from the starting point (order sorting device D) to the target point (three-dimensional shelf C), such as direction, distance, speed change point, etc. The server generates a detailed driving path for R1. The path starts from the order sorting device D, follows the planned route to bypass areas that may collide with other robots, and moves towards the three-dimensional shelf C at a safe speed. During the driving process, R1 will adjust the direction and speed according to the instructions in the path to ensure that the task can be completed accurately and efficiently.

[0090] In one possible implementation, the cloud edge warehouse management server is used to plan the transport route of the target warehouse robot corresponding to the warehouse goods to be transported based on the target warehouse information and the location information of the order sorting device, with the principle of minimizing the number of path collision points, including:

[0091] The cloud edge warehouse management server is used to construct a graph theory model corresponding to the warehouse space according to the first occupation location information of the plurality of three-dimensional shelves and the second occupation location information of the order sorting device, wherein the nodes represent possible path points, the edges represent path segments, and the weights on the edges represent path costs;

[0092] In one embodiment, in the graph theory model, the warehouse space is abstracted as a graph consisting of nodes and edges. Each node represents a possible path point in the warehouse, which can be the entrance, exit, location of the order sorting device of the three-dimensional shelf, and other key turning points in the warehouse. The edge represents the path segment connecting these nodes, and each edge has its corresponding weight, which can represent the path cost, such as distance, time, energy consumption, etc. By constructing such a graph theory model, path planning and optimization can be conveniently performed.

[0093] Assume that there are three three-dimensional shelves A, B, and C and an order picking device D in the warehouse. The entrances and exits of A, B, and C, as well as the location of D, can be regarded as nodes, and the path segments connecting these nodes can be regarded as edges. For example, there is an edge from D to the entrance of A, another edge from the exit of A to the entrance of B, and so on. Each edge is assigned a weight value according to its actual distance.

[0094] According to the target storage information and the location information of the order sorting device, using the shortest path algorithm, calculate the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the warehouse goods to be transported;

[0095] In one embodiment, after the graph theory model is constructed, a shortest path algorithm (such as Dijkstra algorithm, A* algorithm, etc.) can be used to calculate the initial shortest path of the target storage robot from the transport starting point (usually the order sorting device D) to the transport end point (the entrance or designated location of the target storage shelf). This path is calculated based on the nodes and edges in the current graph theory model and their corresponding weights.

[0096] Assume that the goal is to move a warehouse cargo from D to the entrance of C. Use the Dijkstra algorithm to search for the shortest path in the graph model, starting from D and ending at the entrance of C. The algorithm will gradually traverse the nodes in the graph and calculate the shortest distance to each node based on the weight of the edge, and finally find the shortest path from D to the entrance of C.

[0097] Performing collision detection between the preliminary shortest path and particles corresponding to the target storage robot that is transporting other stored goods during the transport process;

[0098] In one embodiment, after obtaining the preliminary shortest path, it is necessary to perform collision detection between the path and the particles corresponding to the target storage robot carrying other stored goods during the handling process. The particles here represent the possible positions and movement trajectories of other robots. By simulating the movement of these particles, it can be determined whether the preliminary shortest path has a collision risk with them.

[0099] Assume that on the preliminary shortest path, a particle (representing another robot carrying something) will appear at a certain point on the path at the same time. This means there is a risk of collision, and the situation at this collision point can be analyzed to determine whether the path needs to be adjusted.

[0100] If the preliminary shortest path collides with the particles corresponding to the target storage robot carrying other stored goods during the transportation process, the nodes or edges in the preliminary shortest path are adjusted and the path is recalculated until the path with the least number of collision points is found as the final transportation route.

[0101] In one embodiment, if the preliminary shortest path collides with the particle corresponding to the target storage robot carrying other storage goods during the transportation process, the path is adjusted to avoid the collision point. This can be achieved by changing the nodes or edges in the path. Specifically, you can try to find alternative path segments, or adjust the order of nodes to bypass the collision point. After each adjustment, you need to recalculate the cost of the path and ensure that the new path is still feasible and has a low cost.

[0102] After a collision point is found, the path can be adjusted to avoid it. For example, try to find an alternative path near the preliminary shortest path that bypasses the collision point and has a relatively low cost. After finding an alternative path, recalculate its cost and compare it to the cost of the preliminary shortest path. If the cost of the alternative path is still low and there is no risk of collision with other particles, it can be used as the final transport route. If there is still a problem of collision or high cost, you can continue to adjust the path and recalculate until you find a satisfactory solution.

[0103] In one possible implementation, the cloud edge warehouse management server is used to calculate the preliminary shortest path from the transport starting point to the transport end point of the target warehouse robot corresponding to the warehouse goods to be transported using the shortest path algorithm according to the target warehouse information and the location information of the order sorting device, including:

[0104] The cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request, which is used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, and according to the target warehouse information and the location information of the order sorting device;

[0105] In one embodiment, the cloud edge warehouse management server first parses the transport identifier in the transport request, which is used to clarify whether the warehouse goods to be transported are outbound goods or inbound goods. Based on this identifier and the location information of the target warehouse information and the order sorting device, the server can determine the starting point (i.e., the current location of the goods) and the end point (i.e., the location where the goods need to arrive) of the transport. For outbound goods, the starting point is usually the exit or designated location of the three-dimensional shelf, and the end point is the order sorting device; for inbound goods, the starting point is the order sorting device, and the end point is the entrance or designated location of the three-dimensional shelf.

[0106] Assume that the transport identifier carried in the transport request indicates that the warehouse goods to be transported are outbound goods, and the target storage information points to an exit position of the three-dimensional shelf A, and the order sorting device D is located on the other side of the warehouse. Based on this information, the cloud edge warehouse management server determines that the transport starting point is the exit of the three-dimensional shelf A, and the transport end point is the order sorting device D.

[0107] The transport starting point is set to the node with the known shortest path, and the shortest path lengths of the remaining nodes are set to infinity;

[0108] In one embodiment, after determining the transport start point and end point, the cloud edge warehouse management server starts to execute the initialization step of the shortest path algorithm (such as the Dijkstra algorithm). It sets the transport start point to the node with the known shortest path, and sets the shortest path length of the node to 0 (indicating that the distance from the start point to the start point is 0). At the same time, it sets the shortest path length of all other nodes to infinity (or a very large number), indicating that the shortest paths from these nodes to the start point have not yet been determined.

[0109] Continuing with the example of outbound goods, the cloud edge warehouse management server sets the exit of the three-dimensional shelf A as the node of the known shortest path and sets its shortest path length to 0. At the same time, it sets the shortest path length of the order sorting device D and all other possible path points (such as other shelves in the warehouse, aisle intersections, etc.) to infinity.

[0110] Repeat the following steps: select the node with the smallest current shortest path estimate from the unprocessed nodes as the current node; for each neighbor node of the current node, check whether the path from the current node to the neighbor node is shorter than the known shortest path; if so, update the shortest path length of the neighbor node and the predecessor node of the shortest path until all nodes have been processed, and obtain the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the storage goods to be transported.

[0111] In one embodiment, the cloud edge warehouse server repeatedly performs the following steps until all nodes have been processed:

[0112] The node with the smallest current shortest path estimate is selected from the unprocessed nodes as the current node. This node is the currently known node closest to the starting point of the transport.

[0113] For each neighbor node of the current node (i.e., a node directly connected to the current node), check whether the path from the current node to the neighbor node is shorter than the known shortest path. This is usually achieved by comparing "the shortest path length of the current node + the weight of the edge from the current node to the neighbor node" with "the current shortest path length of the neighbor node".

[0114] If the path from the current node to the neighbor node is shorter, the shortest path length of the neighbor node and the predecessor node of the shortest path are updated. The predecessor node is used to backtrack and construct the shortest path after the algorithm ends.

[0115] For example, in the example of outbound goods, the cloud edge warehouse management server first selects the exit of the three-dimensional shelf A as the current node (because it is the starting point and the shortest path length is 0). Then, it checks the nodes directly connected to the exit (such as aisles in the warehouse, entrances to other shelves, etc.) and tries to reach these nodes through the exit. For each neighbor node, if the path to it through the exit is shorter than its current shortest path, update its shortest path length and predecessor node. This process is repeated until the order sorting device D has been processed and the server obtains the preliminary shortest path from the exit of the three-dimensional shelf A to the order sorting device D.

[0116] In a possible implementation, the cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request and used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, according to the target warehouse information and the location information of the order sorting device, including:

[0117] The cloud edge warehouse management server is used to determine the position corresponding to the target storage information as the transportation starting point, and the position corresponding to the position information of the order sorting device as the transportation end point, when the transportation identifier carried in the transportation request and used to characterize the warehoused goods to be transported is outbound goods;

[0118] When the transport identifier carried in the transport request and used to characterize the warehouse goods to be transported is incoming goods, the position corresponding to the position information of the order sorting device is determined as the transport starting point, and the position corresponding to the target warehouse information is determined as the transport end point.

[0119] In one embodiment, when the transport identifier carried in the transport request indicates that the warehouse goods to be transported are outbound goods, the primary task of the cloud edge warehouse management server is to determine the starting point and end point of the transport. In this case, the transport starting point is naturally the current location of the goods, that is, the location corresponding to the target storage information (usually an exit or a designated location of the three-dimensional shelf). The transport end point is the location where the goods need to be transported, that is, the location of the order sorting device, so as to carry out subsequent packaging, distribution and other operations.

[0120] For example, suppose there is a request for transporting outbound goods, where the target storage information points to an exit location of the three-dimensional shelf A, and the order sorting device D is located on the other side of the warehouse. After the cloud edge warehouse management server parses the transport request, it recognizes that the transport is marked as outbound goods. Therefore, it determines the exit location of the three-dimensional shelf A as the transport starting point and the location of the order sorting device D as the transport end point. Subsequently, the server will use the shortest path algorithm (such as the Dijkstra algorithm) to calculate the preliminary shortest path from the transport starting point to the transport end point.

[0121] In one embodiment, when the transport identifier carried in the transport request indicates that the warehoused goods to be transported are incoming goods, the determination of the starting point and end point of the transport is opposite to that of outgoing goods. In this case, the transport starting point is the location of the order sorting device, because the goods will first be transported here for pre-processing such as sorting and packaging. The transport end point is the location where the goods need to be stored, that is, the location corresponding to the target storage information (usually a certain entrance or designated location of the three-dimensional shelf).

[0122] Suppose there is a request for transporting incoming goods, where the order sorting device D is located on one side of the warehouse, and the target storage information points to a certain entrance position of the three-dimensional shelf B. After the cloud edge warehouse management server parses the transport request, it recognizes that the transport mark is incoming goods. Therefore, it determines the location of the order sorting device D as the transport starting point and the entrance position of the three-dimensional shelf B as the transport end point. Subsequently, the server will also use the shortest path algorithm to calculate the preliminary shortest path from the transport starting point to the transport end point.

[0123] In both cases, once the starting and ending points of the transportation are determined, the cloud edge warehouse management server can use graph theory models and shortest path algorithms to plan the optimal transportation path. This path will guide the target storage robot to start from the starting point and transport the goods to the end point along the shortest path, thereby improving the operational efficiency and accuracy of the warehouse.

[0124] In one possible implementation, the cloud edge warehouse management server is used to plan the transport speed of the target storage robot corresponding to the warehouse goods to be transported based on the principle of minimum collision with the particles, including:

[0125] The cloud edge warehouse management server is used to dynamically predict the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process according to the number of the particles;

[0126] In one embodiment, in a warehouse environment, multiple target storage robots may be performing handling tasks at the same time. In order to avoid collisions, it is necessary to predict the movement trajectories of these robots in the future. This is usually based on the current position, speed, acceleration of the robot and their respective task goals (i.e., the starting and end points of the handling). The prediction algorithm takes into account the dynamic characteristics of the robot (such as acceleration limit, maximum speed, etc.) and the limitations of the warehouse environment (such as shelf location, aisle width, etc.).

[0127] The number of particles can represent the number of robots operating simultaneously in the warehouse or the density of trajectory points that need to be predicted. By increasing or decreasing the number of particles, the accuracy and amount of calculation of the prediction can be adjusted. Each particle can be regarded as a possible position or state, and the overall motion trajectory of the robot can be predicted by simulating the movement of these particles.

[0128] Assume that there are three target storage robots working at the same time in the warehouse, marked as R1, R2 and R3. The storage goods being transported by R1 may have a path conflict with the storage goods currently to be transported. The cloud edge warehouse management server uses particle filtering or Monte Carlo simulation methods to predict the movement trajectory of R1 in the future based on the current position, speed, acceleration and target position of R1. At the same time, the rationality of the movement trajectory is ensured according to the layout and restrictions of the warehouse environment.

[0129] Determine whether there is a collision with the particle according to the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process and the current speed of the target storage robot corresponding to the stored goods to be transported;

[0130] In one embodiment, after obtaining the motion trajectory of the target storage robot that transports other stored goods, it is necessary to compare these trajectories with the current state (including position, speed, and acceleration) of the target storage robot (denoted as R0) corresponding to the stored goods to be transported. By calculating the intersection or proximity between the possible position of R0 within a predetermined time and the motion trajectory, it can be determined whether there is a collision risk.

[0131] Continuing with the above example, the cloud edge warehouse management server compares the current state of R0 (including speed and acceleration) with the motion trajectory of R1. If R0 continues to move at the current speed, its position at a certain point in the future will fall on the motion trajectory of R1 or very close to it, then there is a risk of collision.

[0132] If there is a collision, the current speed or acceleration of the target storage robot corresponding to the warehouse goods to be transported is adjusted based on the principle of minimum collision with the particles to obtain the planned transport speed of the target storage robot corresponding to the warehouse goods to be transported.

[0133] In one embodiment, if it is determined that there is a risk of collision, the cloud edge warehouse management server needs to adjust the speed or acceleration of R0 to avoid collision. The principle of adjustment is to minimize the impact on the original handling plan of R0 while ensuring that the risk of collision with R1 (or other robots) is minimized. This usually involves fine-tuning the speed or acceleration and re-evaluating whether the adjusted motion trajectory meets the collision-free requirement.

[0134] After discovering that there was a risk of collision between R0 and R1, the cloud edge warehouse management server calculated a series of possible speed adjustment schemes and evaluated the impact of each scheme on R0's handling efficiency and collision risk. In the end, the server selected a speed adjustment scheme that could reduce the risk of collision while maintaining high handling efficiency, and sent the new speed or acceleration command to R0. After receiving the command, R0 continued to perform the handling task according to the new speed or acceleration, thus avoiding a collision with R1.

[0135] Through such dynamic prediction and collision avoidance mechanisms, the cloud-edge warehouse management server can ensure that multiple target warehouse robots can work together safely and efficiently in the warehouse environment.

[0136] In a possible implementation, the cloud edge warehouse management server is used to determine the number of path collision points that exist during the transportation process of the target storage robot corresponding to the other stored goods according to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, including:

[0137] The cloud edge warehouse management server is used to determine the distance between two adjacent three-dimensional shelves according to the first occupied position information of the plurality of three-dimensional shelves, and determine the greatest common divisor of the distances as the side length of the voxel grid;

[0138] In one embodiment, in order to efficiently discretize the warehouse space and facilitate the detection of subsequent path collisions, we introduce the concept of voxel grids. The selection of the side length of the voxel grid is crucial, as it directly affects the number of grids, the accuracy of collision detection, and the computational efficiency. Here, we use the first occupancy information of multiple stereoscopic shelves and order sorting devices in the warehouse (i.e., their projected areas on the warehouse plane) to determine the side length of the voxel grid by calculating the greatest common divisor of the distances between adjacent stereoscopic shelves. The advantage of this is that it ensures that the voxel grid can better fit the actual layout of the warehouse, reduces unnecessary grid divisions, and improves computational efficiency.

[0139] Suppose there are three three-dimensional shelves A, B, and C in the warehouse, and their first location information is (x1, y1), (x2, y2), and (x3, y3) respectively. We first calculate the horizontal distance between adjacent shelves (ignoring the height factor). For example, the distance between AB is sqrt((x2-x1) 2 + (y2-y1) 2 ), and the same is true for the distance between BC. Then, we find the greatest common divisor (GCD) of these distances and use that value as the side length of the voxel grid. For example, if the distance between AB is 6 meters and the distance between BC is 9 meters, their GCD is 3 meters, so the side length of the voxel grid is determined to be 3 meters.

[0140] According to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, the space in the warehouse space not occupied by the three-dimensional shelves and the order sorting device is divided into a plurality of voxel grids according to the side length, each of the voxel grids having a unique identifier;

[0141] In one embodiment, after determining the side length of the voxel grid, we can divide the portion of the warehouse space not occupied by the three-dimensional shelves and order picking devices into multiple voxel grids according to the overall size of the warehouse (including length, width and height, but in this step we mainly focus on the plane size). Each voxel grid represents a small cube in three-dimensional space and has a unique identifier (such as encoded by its row, column number and height layer in the grid).

[0142] Assume that the plane dimensions of the warehouse are 100 meters long and 80 meters wide, and the height is not considered in this step. We have determined that the side length of the voxel grid is 3 meters. Then, in the horizontal direction, the warehouse can be divided into grids of approximately 33 rows (100 / 3 rounded up) and 27 columns (80 / 3 rounded up). For each grid, we can uniquely identify it by its row and column number (such as (i, j)). In the vertical direction, although this step does not directly involve the division of height, the height layer can be further subdivided as needed in subsequent processing.

[0143] Mapping the driving path of each storage robot to the corresponding voxel grid, and recording the number of robots passing through each voxel grid;

[0144] In one embodiment, in a warehouse environment, the path of each storage robot is a continuous curve. To associate these paths with voxel grids, we need to map each point on the path to the nearest voxel grid. Then, we record the number of robots passing through each voxel grid, which helps us identify potential path collision points.

[0145] Assume that there are n warehouse robots running in the warehouse at the same time, and their driving paths are known. We can use spatial interpolation or path discretization methods to map the points on each path to the nearest voxel grid. For example, for a point P(x, y, z) on the path, we can calculate the row and column number of the grid to which it belongs (by dividing by the length of the grid edge and rounding down), and update the counter of the number of robots passing through the grid.

[0146] All the voxel grids are traversed, and the voxel grids with more than one robot are used as collision voxel grids, and the number of the collision voxel grids is determined as the number of path collision points in the transportation process of the target storage robot corresponding to the other storage goods.

[0147] In one embodiment, after completing the mapping of all robot paths, we traverse all voxel grids. If more than one robot passes through a grid, it means that these robots have crossed paths in the grid area and there is a risk of collision. We call these grids collision voxel grids, and count their number as the number of path collision points in the handling process.

[0148] Traverse all voxel grids and check the robot count counter on each grid. If the counter of a grid is greater than 1, it means that at least two robots have crossed paths in the grid area. We mark such grids as collision voxel grids and accumulate their counts as the final number of collision points. This number can be used as an important indicator to evaluate the quality of warehouse path planning and collision risk.

[0149] In a possible implementation, the storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, including:

[0150] The storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, and to monitor the relative position relationship between the target storage robot and the surrounding environment in real time by using sensors and image processing technology according to the physical size and turning radius of the storage robot and the size of the goods to be transported, and to determine whether there is a collision;

[0151] In one embodiment, after determining the target storage information and driving path, the storage robot starts to perform the handling task. In order to ensure the safety and smooth progress of the handling process, it is necessary to monitor the relative position relationship between the robot and the surrounding environment in real time. This is usually achieved through a variety of sensors (such as lidar, ultrasonic sensors, cameras, etc.) and image processing technology installed on the robot. Sensors are used to collect environmental data, such as distance, speed, obstacle information, etc., while image processing technology is used to analyze the images captured by the camera and identify obstacles, road signs, etc.

[0152] For example, warehouse robot R starts to move the goods to be moved based on the target warehouse information (such as shelf location) and the planned driving path. During the driving process, the laser radar on the robot continuously scans the surrounding environment and measures the distance to the surrounding objects. At the same time, the camera captures real-time video images and identifies obstacles in front (such as other robots, shelves, people, etc.) through image processing algorithms. The robot integrates these sensor data and image processing results to form a comprehensive perception of the surrounding environment.

[0153] In one embodiment, after monitoring the relative position relationship between the robot and the surrounding environment in real time, the system needs to evaluate whether there is a collision risk based on this information. This usually involves comparing the real-time data with the physical size of the robot, the turning radius, and the size of the goods to be transported. If it is predicted that the robot may collide with the surrounding environment when traveling at the current speed, direction, or path, then it is determined that there is a collision risk.

[0154] For example, based on the distance data measured by the LiDAR, the image recognition results of the camera, and the physical parameters of the robot (such as body width, height, turning radius, etc.), the collision detection algorithm evaluates the safety of the current driving path. For example, if an obstacle is found ahead and the distance is less than the robot's braking distance plus the safety margin, or if it is predicted that the robot will scrape against the shelf next to it when turning, the system determines that there is a collision risk.

[0155] When it is determined that there is a risk of collision, the speed, direction or path of the storage robot is adjusted to avoid the collision until the task of transporting the goods to be transported is completed.

[0156] In one embodiment, when a collision risk is determined, immediate measures are taken to avoid the collision. This typically includes adjusting the robot's speed, direction, or path. The choice of adjustment strategy depends on the specific collision situation and environmental constraints. For example, if the collision risk is caused by excessive speed, the robot can slow down or stop; if it is caused by the wrong direction or unreasonable path planning, the path can be replanned or the driving direction can be changed.

[0157] For example, after discovering a collision risk, the system first tries to reduce the possibility of collision by slowing down. If the collision still cannot be avoided after slowing down, the system replans the path based on the current environmental information and the robot's capabilities. For example, if there is an obstacle ahead that cannot be bypassed, the system may plan a long but safe path to avoid the obstacle. Alternatively, if the robot has the ability to move sideways, it may also consider moving sideways to avoid the obstacle while maintaining the forward direction.

[0158] In one embodiment, after adjusting the speed, direction or path and successfully avoiding collisions, the warehouse robot continues to move the goods to be moved according to the new driving plan. During the entire moving process, the system will continue to monitor the environment and be prepared to make further adjustments if necessary. The moving task is not completed until the goods are safely moved to the target location and the unloading operation is completed.

[0159] After successfully avoiding all collision risks, the storage robot R continues to drive along the adjusted path and finally safely places the goods to be transported at the target storage location. After completing the unloading, the robot returns to the idle state and waits for the next handling task. During the entire handling process, the system always maintains real-time monitoring of the environment and dynamic adjustment capabilities to ensure the safe and efficient completion of the handling task.

[0160] In a possible implementation, the target storage information includes: a target three-dimensional shelf and a target three-dimensional layer.

[0161] In a possible implementation, the robot information includes: robot position information, robot maximum carrying weight information, robot size information, and robot turning radius.

[0162] The robot position information refers to the current position coordinates of the robot in the warehouse, usually including X, Y coordinates (and possibly Z coordinates, indicating height). It is used to determine the robot's starting position and plan the driving path from the starting position to the target position. Before performing a handling task, the robot will first obtain its current position information through GPS, LiDAR or other positioning technologies, such as (X=10, Y=20).

[0163] The robot's maximum handling weight information refers to the maximum weight limit of the cargo that the robot can carry. It is used to determine whether the robot can carry the currently specified cargo to avoid damage or safety issues caused by overloading. If the robot's maximum handling weight information is 50kg, and the weight of the cargo to be carried is 60kg, the robot cannot perform the handling task.

[0164] Robot size information refers to the physical size of the robot, including length, width, and height. It is used to evaluate the robot's ability to move in the warehouse and its ability to pass through narrow passages or complex environments. When planning a driving path, the robot needs to consider its own size information to ensure that it can pass through passages and doorways in the warehouse smoothly.

[0165] The robot turning radius refers to the minimum turning radius that the robot can complete during driving. It is used to plan the turning path of the robot in the warehouse to ensure that the robot can smoothly bypass obstacles or drive along the predetermined route. When planning the driving path, if the path includes a part that needs to turn, the robot will calculate the appropriate turning path based on its own turning radius to avoid collision with obstacles.

[0166] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the present disclosure. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0167] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An intelligent warehousing multi-layer robot system based on cloud-edge collaboration, characterized in that: include: Multiple three-dimensional shelves, storage robots, order sorting devices and cloud-edge warehouse management servers; Wherein, different three-dimensional layers of the plurality of three-dimensional shelves are used to store warehouse goods of different specifications and sizes, and the plurality of three-dimensional shelves are respectively used to store the warehouse goods of different types; The cloud edge warehouse management server is used to determine the target storage information of the stored goods to be transported in response to the transport request sent by the order sorting device to the cloud edge warehouse management server for the stored goods, determine the target storage robot used to transport the stored goods to be transported according to the target storage information and the robot information of the storage robot, and estimate the position information of the target storage robot corresponding to the other stored goods for positioning and transporting based on the adaptive Monte Carlo positioning method, and plan the driving path of the target storage robot corresponding to the stored goods to be transported according to the position information and the target storage information; The storage robot is used to, when being determined as the target storage robot, receive the target storage information and the driving path sent by the cloud edge warehouse management server, and transport the storage goods to be transported according to the target storage information and the driving path; The cloud edge warehouse management server is used to estimate the position information of the target warehouse robot corresponding to the other warehouse goods based on the adaptive Monte Carlo positioning method, and plan the driving path of the target warehouse robot corresponding to the warehouse goods to be transported according to the position information and the target warehouse information, including: The cloud edge warehouse management server is used to determine the number of path collision points that exist during the transportation process of the target storage robot that transports other stored goods according to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, wherein the path collision point is a coordinate point where the driving path of the storage robot intersects; Determine the number of collision points on the path as the number of particles in the adaptive Monte Carlo positioning method used to estimate the position of each of the warehouse robots, and estimate the position information of the target warehouse robots corresponding to the other warehouse goods based on the number of particles; According to the target warehouse information and the location information of the order sorting device, a transport route of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing the number of collision points on the path, and a transport speed of the target warehouse robot corresponding to the warehouse goods to be transported is planned based on the principle of minimizing collisions with the particles; According to the transport route and the transport speed, a driving path of the target storage robot corresponding to the warehouse goods to be transported is obtained.

2. The system according to claim 1, characterized in that The cloud edge warehouse management server is used to plan the transport route of the target warehouse robot corresponding to the warehouse goods to be transported based on the target warehouse information and the location information of the order sorting device, with the principle of minimizing the number of path collision points, including: The cloud edge warehouse management server is used to construct a graph theory model corresponding to the warehouse space according to the first occupation location information of the plurality of three-dimensional shelves and the second occupation location information of the order sorting device, wherein the nodes represent possible path points, the edges represent path segments, and the weights on the edges represent path costs; According to the target storage information and the location information of the order sorting device, using the shortest path algorithm, calculate the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the warehouse goods to be transported; Performing collision detection between the preliminary shortest path and particles corresponding to the target storage robot that is transporting other stored goods during the transport process; If the preliminary shortest path collides with the particles corresponding to the target storage robot carrying other stored goods during the transportation process, the nodes or edges in the preliminary shortest path are adjusted and the path is recalculated until the path with the least number of collision points is found as the final transportation route.

3. The system according to claim 2, characterized in that The cloud edge warehouse management server is used to calculate the preliminary shortest path from the transport starting point to the transport end point of the target warehouse robot corresponding to the warehouse goods to be transported using the shortest path algorithm according to the target warehouse information and the location information of the order sorting device, including: The cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request, which is used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, and according to the target warehouse information and the location information of the order sorting device; The transport starting point is set to the node with the known shortest path, and the shortest path lengths of the remaining nodes are set to infinity; Repeat the following steps: select the node with the smallest current shortest path estimate from the unprocessed nodes as the current node; for each neighbor node of the current node, check whether the path from the current node to the neighbor node is shorter than the known shortest path; if so, update the shortest path length of the neighbor node and the predecessor node of the shortest path until all nodes have been processed, and obtain the preliminary shortest path from the transport starting point to the transport end point of the target storage robot corresponding to the storage goods to be transported.

4. The system according to claim 3, characterized in that The cloud edge warehouse management server is used to determine the transport starting point and the transport end point according to the transport identifier carried in the transport request and used to characterize that the warehouse goods to be transported are outbound goods or inbound goods, according to the target warehouse information and the location information of the order sorting device, including: The cloud edge warehouse management server is used to determine the position corresponding to the target storage information as the transportation starting point, and the position corresponding to the position information of the order sorting device as the transportation end point, when the transportation identifier carried in the transportation request and used to characterize the warehoused goods to be transported is outbound goods; When the transport identifier carried in the transport request and used to characterize the warehouse goods to be transported is incoming goods, the position corresponding to the position information of the order sorting device is determined as the transport starting point, and the position corresponding to the target warehouse information is determined as the transport end point.

5. The system according to claim 2, characterized in that The cloud edge warehouse management server is used to plan the transport speed of the target storage robot corresponding to the warehouse goods to be transported based on the principle of minimum collision with the particles, including: The cloud edge warehouse management server is used to dynamically predict the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process according to the number of the particles; Determine whether there is a collision with the particle according to the motion trajectory of the target storage robot corresponding to the other stored goods during the transportation process and the current speed of the target storage robot corresponding to the stored goods to be transported; If there is a collision, the current speed or acceleration of the target storage robot corresponding to the warehouse goods to be transported is adjusted based on the principle of minimum collision with the particles to obtain the planned transport speed of the target storage robot corresponding to the warehouse goods to be transported.

6. The system according to any one of claims 1 to 5, characterized in that: The cloud edge warehouse management server is used to determine the number of path collision points that exist in the transportation process of the target storage robot corresponding to the other storage goods according to the first occupation position information of the plurality of three-dimensional shelves and the second occupation position information of the order sorting device, including: The cloud edge warehouse management server is used to determine the distance between two adjacent three-dimensional shelves according to the first occupied position information of the plurality of three-dimensional shelves, and determine the greatest common divisor of the distances as the side length of the voxel grid; According to the first occupied position information of the plurality of three-dimensional shelves and the second occupied position information of the order sorting device, the space in the warehouse space not occupied by the three-dimensional shelves and the order sorting device is divided into a plurality of voxel grids according to the side length, each of the voxel grids having a unique identifier; Mapping the driving path of each storage robot to the corresponding voxel grid, and recording the number of robots passing through each voxel grid; All the voxel grids are traversed, and the voxel grids with more than one storage robots are used as collision voxel grids, and the number of the collision voxel grids is determined as the number of path collision points in the transportation process of the target storage robots corresponding to the other storage goods.

7. The system according to any one of claims 1 to 5, characterized in that: The storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, including: The storage robot is used to transport the storage goods to be transported according to the target storage information and the driving path, and to monitor the relative position relationship between the target storage robot and the surrounding environment in real time by using sensors and image processing technology according to the physical size and turning radius of the storage robot and the size of the storage goods to be transported, and to determine whether there is a collision; When it is determined that there is a risk of collision, the speed, direction or path of the storage robot is adjusted to avoid the collision until the task of transporting the warehouse goods to be transported is completed.

8. The system according to any one of claims 1 to 5, characterized in that: The target storage information includes: a target three-dimensional shelf and a target three-dimensional layer.

9. The system according to any one of claims 1 to 5, characterized in that: The robot information includes: robot position information, robot maximum handling weight information, robot size information and robot turning radius.

Images