Warehouse logistics grabbing method and sorting robot

Abstract

The invention discloses a warehouse logistics grabbing method and a sorting robot, and the warehouse logistics grabbing method comprises the following steps: determining a starting point according to the position information of a to-be-grabbed article, and entering an area where the starting point is located; obtaining a picture in an area where the starting point is located; feature information of the to-be-recognized article in the photo is obtained, the feature information at least comprises the shape, the color and the damage condition, and a judgment result is output according to the feature information; and under the condition that the judgment result meets the requirement, the corresponding grabbing mechanical arm is selected for grabbing according to the feature information. According to the grabbing method for warehouse logistics and the sorting robot, the to-be-grabbed objects are recognized, the types of the to-be-grabbed objects are respectively obtained, the corresponding mechanical arms are selected for grabbing according to the types, the grabbing stability is better, the grabbing error rate is lower, and the efficiency is also well improved.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics robots, and in particular to a grabbing method and a sorting robot for warehousing logistics. Background Art

[0002] Taking Suzhou Zhuoshengyu Intelligent Technology Co., Ltd. as an example, its patent "Automatic Sorting Method and System for Intelligent Logistics Robot" disclosed on February 23, 2024, the intelligent logistics robot it claims relies on the cooperation between the camera and the rotating mechanism platform. The main content involved is the cooperation between machine vision and automation equipment. The rotating mechanism platform is mainly composed of motors, and the camera generally uses an industrial camera.

[0003] JD's ground wolf sorting robot mainly adopts the mode of QR code recognition + cluster path planning. A large number of QR codes are printed on the ground. The QR codes are used for positioning, turning and moving. The dynamic planning path algorithm is mainly adopted.

[0004] Dart robot

[0005] The Libiao robot also uses a similar operating process to JD.com's Dilang, using a QR code for positioning, but it has an industrial computer built in to control the position, and the built-in graphics card is mainly Intel Atom The Z3735 processor has its own integrated graphics card. Similar to the JD Logistics workflow, items are placed manually and it does not have an automatic grabbing function.

[0006] Bastian Solutions SmartPick

[0007] The AI ​​vision platform guides the robot and EoAT to the pick location in the correct orientation to reliably and accurately pick up individual items. The AI ​​then directs the robot to activate the appropriate vacuum cups to grab the product. Once safe, the robot retracts from the bin and places the object at the secondary destination, ready for another pick and place sequence. As the robot picks up objects of different shapes, sizes, and materials, it continuously learns and improves how to identify and pick up objects in different locations, further improving efficiency.

[0008] A fully automated order fulfillment solution consists of a goods-to-person system, industrial pick-and-place robots, an AI vision system, auxiliary material handling equipment, and warehouse software (such as WES) that controls the entire process. Upon receipt of an order, the warehouse software communicates with the GTP (goods-to-person) system to deliver the box containing the ordered items to the GTP workstation operated by the robot. The robot's AI and vision platform identifies the SKU required for the order, then selects the "easiest to pick" product in the pile and releases the robot to pick it and place it at the designated destination. Once the order is completed, the container is transported for further outbound processing, such as checkweighing using an inline scale to ensure order accuracy and automatic labeling to attach shipping labels.

[0009] Bastian Solutions’ case picking robots are equipped with innovative AI and machine vision that can quickly adapt to your current products or integrate new products. As new products are introduced, vision automatically identifies pickable items during a dynamic evaluation process. Over time, the AI ​​software builds a library of objects it has seen and handled, along with optimal picking information. Based on previous experience, machine learning continuously reclassifies non-pickable items, increasing autonomy and SKU coverage. This knowledge is shared across the fleet, enabling them to continuously learn from each other so they can operate as a single, efficient order fulfillment operation. The AI ​​and vision platform connects directly with BastianSolutions’ Exacta® intralogistics software, transferring product data and sharing order fulfillment information. This integration enables the robotic picking cells to operate with increased autonomy.

[0010] The above robots all rely on a single grasping mechanical arm when grasping, which may lead to poor grasping stability for different types of objects, thus easily causing errors during sorting and reducing the grasping efficiency. Summary of the invention

[0011] In view of this, an object of the present invention is to provide a grasping method and a sorting robot for warehousing logistics, which have better grasping stability, lower grasping error rate, and greatly improved efficiency.

[0012] The present invention provides a method for grabbing warehousing logistics, comprising the following steps:

[0013] Determine the starting point according to the location information of the object to be grasped, and enter the area where the starting point is located;

[0014] Obtaining coordinate information within the area where the starting point is located;

[0015] Acquire characteristic information of the object to be identified in the photo, wherein the characteristic information at least includes shape, color and damage, and output a judgment result according to the characteristic information;

[0016] When the judgment result meets the requirements, a corresponding grasping robot arm is selected for grasping according to the feature information.

[0017] In one embodiment, determining the starting point according to the position information of the object to be grasped further includes:

[0018] Obtaining the location information of the object to be grasped;

[0019] Constructing a radar map around the object to be grasped;

[0020] The starting point is determined by calculation based on the position of the object to be grasped and the position information of the object to be grasped.

[0021] In one embodiment, the step of calculating and determining the starting point based on the position of the user and the position information of the object to be grasped further includes:

[0022] Determine the grasping range of the grasping robot arm;

[0023] The starting point is calculated according to the position information of the object to be grasped and the own position in combination with the grasping range.

[0024] In one embodiment, determining the starting point according to the position information of the object to be grasped further includes:

[0025] When at the position of said starting point;

[0026] Calculating the base angle according to the position information of the object to be grasped and the position information of the starting point;

[0027] Whether it is within the grasping range of the grasping robot arm is judged in combination with the base angle, the position information of the object to be grasped and the position information of the starting point.

[0028] In one embodiment, acquiring feature information of the object to be identified in the photo, wherein the feature information at least includes shape, color, and damage, and outputting a judgment result according to the feature information further includes:

[0029] Divide the input photo into an S×S grid and perform adaptive anchor box calculation to select a bounding box from the grid;

[0030] Extract and fuse features for each bounding box to form a feature map;

[0031] Predictions are made based on the feature map, and the world coordinates, size, and confidence of the bounding box are output to obtain an output result.

[0032] The present invention also provides a sorting robot for warehousing logistics, which adopts the above-mentioned grabbing method for warehousing logistics, including a trolley, a grabbing mechanical arm component, an image acquisition component, a map component module and a controller component. The grabbing mechanical arm component, the map component module and the controller component are all assembled on the trolley, the image acquisition component is assembled on the grabbing mechanical arm component, and the image acquisition component, the grabbing mechanical arm component and the map component module are all communicatively connected with the controller component.

[0033] In one embodiment, the trolley includes a bracket, at least four wheels and a corresponding number of drive motors, the drive motors and the wheels are all mounted on the bracket, the drive motor is used to drive the corresponding wheel movements, the drive motor is communicatively connected to the controller component, and the grabbing robot arm component, map component module and controller component are all mounted on the bracket.

[0034] In one embodiment, the grabbing robot arm component includes a servo, a linkage arm, a robot arm platform, a robot arm base, a large arm, a large arm connecting rod, a small arm, a posture retaining frame, a small arm connecting rod and a grabbing piece. The robot arm platform is mounted on the trolley, and the servo and the robot arm base are both mounted on the robot arm platform. The servo is linked to the posture retaining frame through the linkage arm, and the posture retaining frame is hingedly mounted on the large arm. The two ends of the small arm connecting rod are respectively hingedly connected to the posture retaining frame and the large arm connecting rod, and the large arm and the large arm connecting rod are both mounted on the robot arm base.

[0035] The grabbing method and sorting robot for warehousing logistics provided by the present invention can identify the objects to be grabbed, distinguish the types of the objects to be grabbed, and select the corresponding manipulator for grabbing according to the types. The grabbing stability is better, the grabbing error rate is lower, and the efficiency is greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 A schematic flow chart of a method for grabbing warehouse logistics provided in Embodiment 1 of the present invention.

[0038] Figure 2 This is a system block diagram corresponding to the warehouse logistics grabbing method provided in Example 1 of the present invention.

[0039] Figure 3 This is a schematic structural diagram of the warehouse logistics sorting robot from the first perspective provided in the second embodiment of the present invention.

[0040] Figure 4 This is a schematic structural diagram of the warehouse logistics sorting robot from the first perspective provided in the second embodiment of the present invention.

[0041] Figure 5 A schematic diagram of the structure of a grabbing component of a warehouse logistics sorting robot provided in the second embodiment of the present invention.

[0042] Figure 6 A schematic diagram of the structure of the grabbing mechanical arm component of the warehouse logistics sorting robot provided in the second embodiment of the present invention.

[0043] Figure 7 A schematic diagram of the structure of the trolley of the warehouse logistics sorting robot provided in the second embodiment of the present invention.

[0044] Figure 8 This is a schematic diagram of the abstract structure of the trolley chassis of the warehouse logistics sorting robot provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0045] The specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the description of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In the description of the present invention, unless otherwise clearly specified and limited, the terms "set", "install", "connection" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0047] The directions or positional relationships indicated by terms such as “upper”, “lower”, “left”, “right”, “front”, “back”, “top”, “bottom”, “inside” and “outside” are based on the directions or positional relationships shown in the accompanying drawings, or are the directions or positional relationships in which the inventive product is usually placed when used. They are only for the convenience of description and simplified description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as a limitation on the present invention.

[0048] The terms "first", "second", "third", etc. are merely used to distinguish elements of similar nature, and do not indicate or imply relative importance or a particular order.

[0049] The terms "comprises," "comprising," or any other variations thereof, are intended to cover a non-exclusive inclusion of the elements listed and may also include additional elements not expressly listed.

[0050] Embodiment 1

[0051] See also Figure 1 and Figure 2 The storage logistics grabbing method provided by the present invention comprises the following steps:

[0052] S1, determine the starting point according to the location information of the object to be grasped, and enter the area where the starting point is located.

[0053] It is understandable that the above steps may further include:

[0054] S101, obtaining the location information of the object to be grasped.

[0055] It is understandable that the location information of each object to be grasped is entered in the management system. When transfer is required, the location information of the object to be grasped can be obtained through the management system. The location information can be the three-dimensional coordinate data of the center point of the object to be grasped.

[0056] S102, constructing a radar map around the object to be grasped.

[0057] It can be understood that the three-dimensional coordinates of the center point of the object to be grasped can be used as the center point, and then the laser radar rplidar can be used to obtain various target information in the preset area around the object to be grasped, and the radar map can be constructed through the cartographer algorithm, and the coordinates can be constructed in the map to determine the coordinate positions of various targets.

[0058] S103, calculating and determining the starting point according to the position of the self and the position information of the object to be grasped.

[0059] The above steps may further include:

[0060] S103a, determining the grasping range of the grasping robot arm.

[0061] It can be understood that the grasping range of the grasping robot arm is known, and the grasping range can be a circular area with its own point as the origin. In the case that the grasping robot arm has a blind area, its grasping range can also be an arc-shaped area.

[0062] S103b, calculating the starting point according to the position information of the object to be grasped and the own position in combination with the grasping range.

[0063] It can be understood that the distance between the position of the object to be grasped and its own position is calculated. Its own position can be the center point of the end of the grasping robot arm away from the grasping member. A point is selected in an area where the distance is less than the radius within the grasping range. It can be randomly selected as long as there is no interference between the sorting robots.

[0064] It is understandable that in order to verify whether the position of the starting point is appropriate, the above steps may further include:

[0065] S104, when at the starting point, calculating the base angle according to the position information of the object to be grasped and the position information of the starting point.

[0066] It can be known that the base angle can be obtained by calculating the inverse tangent function. The base angle can be understood as whether the angle of the sorting robot is appropriate, that is, whether the rotation angle of the grabbing arm of the sorting robot is appropriate. This angle can well determine whether the object to be grabbed can be photographed. The sorting robot is equipped with a lidar, a grayscale sensor, an odometer, an inertial sensor and an encoder. The encoder returns data to the odometer. The odometer first calculates the mileage distance and calculates the posture in combination with the inertial sensor. Then the odometer and inertial sensor data are fused, and the extended Kalman filter is used to eliminate errors. The two sensor data after error elimination are combined with the lidar for navigation and positioning, and the posture data of the sorting robot can also be acquired in real time.

[0067] S105, judging whether it is within the grasping range of the grasping robot arm by combining the base angle, the position information of the object to be grasped and the position information of the starting point.

[0068] It is understandable that the straight-line distance between the sorting robot and the object to be grasped can be calculated based on the base angle, the position information of the object to be grasped and the position information of the starting point, and the straight-line distance is compared with the grasping range to determine whether it is within the grasping range of the grasping robot arm.

[0069] S2, obtaining photos in the area where the starting point is located.

[0070] It is understandable that after the sorting robot reaches the starting point, the angle it faces is the location of the item to be grasped, and the sorting robot takes pictures of the area where it is located. The pictures can be taken using a monocular camera as a data source to capture pictures of various scenes in real time.

[0071] S3, obtaining feature information of the object to be identified in the photo, wherein the feature information at least includes shape, color and damage condition, and outputting a judgment result based on the feature information.

[0072] It is understandable that the above steps may further include:

[0073] The extraction of features such as color and shape can be done by using the YOLOv5 model and the BPU chip of the Horizon X3 development board. The YOLOv5 model is embedded in the BPU chip to achieve efficient feature extraction and detection of whether the package is damaged. YOLOv5 is an advanced target detection model that can quickly and accurately identify and determine whether the target object is in the photo and its position in the photo in the image taken by a monocular camera. By utilizing the BPU (computing unit) on the Horizon X3 development board, the chip has powerful parallel computing capabilities and efficient processing performance, which further accelerates the image processing and rendering process and greatly reduces the time consumed by YOLOv5 in the post-processing process.

[0074] If you want to maximize the use of the BPU on the Horizon X3 development board to accelerate the model reasoning process, you need to convert the pt model trained by YOLOv5 into a bin model supported by the BPU of the Horizon X3 development board. At the same time, you need to modify the activation function of YOLOv5 so that all operators of the model run on the BPU. The YOLOv5 model is trained with a large number of data sets so that it can accurately detect the damage of the package under different lighting conditions and angles. The YOLOv5 model used can identify small objects through deep learning. This is obtained by collecting a large number of small objects (such as spheres, rings, cubes, and cuboids of different colors, shapes, and lighting conditions), annotating the data with Make Sense, and training with the configured training parameters.

[0075] The above steps may further include:

[0076] S301, dividing the input photo into an S×S grid, and performing adaptive anchor box calculation to select a bounding box from the grid;

[0077] S302, extracting and fusing features of each bounding box to form a feature map;

[0078] S303, predicting according to the feature map, and outputting the world coordinates, size and confidence of the bounding box to obtain an output result.

[0079] It can be understood that the input photo is divided into S×S grids, and each grid is responsible for detecting the objects falling within its range.

[0080] For each grid, the model predicts multiple bounding boxes and outputs the s-coordinates, sizes, and probability of object presence of these boxes.

[0081] The predictions include: the center position (x, y), width and height (w, h), target category (classscore) and confidence score of each bounding box.

[0082] Bounding Box: YOLOv5 predicts multiple candidate boxes (often called "anchor boxes" or anchorboxes) for each grid, which are obtained by learning the distribution of objects of different sizes and shapes. The model predicts the offset (relative to the grid) and size of the box.

[0083] Confidence score: Each predicted box has a confidence score, which indicates the probability of the existence of an object in the box and the degree of match (IoU) between the box and the actual object. If the confidence is low, the box may be a wrong prediction. YOLOv5 makes predictions at different scales in order to detect objects of different sizes. The structure of the network adopts a design similar to FPN (Feature Pyramid Network), which makes predictions on feature maps from shallow to deep layers. Shallow features are responsible for small objects, and deep features are responsible for large objects.

[0084] YOLOv5 uses a backbone network with a CSP (Cross Stage Partial) structure, such as CSPDarknet53, which can extract features more efficiently. This structure reduces the amount of computation while retaining rich feature expressions by gradually segmenting and fusing feature maps.

[0085] S4, judging the object to be identified according to the feature information, and selecting a corresponding grasping robot arm to grasp the object according to the feature information when the feature information meets the requirements.

[0086] It is understandable that the above steps may further include:

[0087] S401, determining the type of the object to be identified based on the color information.

[0088] It can be known that, in addition to using color information to determine the type of the object to be identified, shape features and the like may also be used to determine the type of the object to be identified.

[0089] S402, when the color information matches,

[0090] S403, selecting a corresponding grasping robot arm to grasp according to the shape information.

[0091] It is understandable that when the color information does not match, the grabbing instruction is skipped and the next grabbing instruction of the item to be grabbed is directly entered. During the grabbing process, the object to be identified can be grabbed according to its position and placed on the corresponding conveyor belt or at the designated storage area after grabbing.

[0092] Embodiment 2

[0093] See also Figures 3 to 7 This embodiment provides a sorting robot for warehousing logistics, which adopts the above-mentioned warehousing logistics grasping method, including a trolley 1, a grasping mechanical arm component 2, an image acquisition component 3, a map component module and a controller component. The grasping mechanical arm component 2, the map component module and the controller component are all assembled on the trolley 1, the image acquisition component 3 is assembled on the grasping mechanical arm component 2, and the image acquisition component 3, the grasping mechanical arm component 2 and the map component module are all communicatively connected with the controller component.

[0094] Understandably, see Figure 2 When controlling multiple sorting robots, a local area network can be established, and a server and multiple sorting robots can be deployed in the local area network as the core to form a robot cluster. The aim is to provide customers with an efficient, reliable and scalable logistics management system through the integration and application of advanced technologies, including:

[0095] The sorting robot host uses the Xuripi X3, and ROS2 and TROS are independently deployed on each host to support the robot's autonomous navigation and operation. At the same time, in order to realize system monitoring and data collection, each host also deploys the Docker image of Prometheus and node_exporter nodes, as well as the Kubernetes slave node. In this way, each robot host can run independently and maintain real-time communication with the server.

[0096] On the server side, Grafana, SQLite database, and K8S master node are deployed. Grafana is used to collect and display data from each Prometheus node to achieve unified monitoring and maintenance. The SQLite database stores the QR code data published by each robot host through socket.io, which is convenient for centralized management and analysis. The K8S master node is responsible for coordinating and managing the application deployment on each robot host through Kubernetes to ensure stable operation and rapid updates of the system.

[0097] The Kubernetes master node uses K8S to centrally manage and maintain each robot host. When there is a new Docker image or application update, the Master node can quickly distribute the update to each robot host to achieve unified update. This not only simplifies the maintenance of the system, but also ensures the consistency of the software version on each host, improving the reliability and security of the system.

[0098] Grafana integrates data from each Prometheus node to provide real-time monitoring and alarm functions to ensure the stability and reliability of the system under high load. Through a unified monitoring platform, operation and maintenance personnel can promptly discover and handle abnormal situations in the system, reduce failure rates, and improve the overall efficiency of the system.

[0099] It is understandable that the four-wheel drive car is abstracted as Figure 8 The model in which W represents the distance between the center points of the left and right motors; L represents the distance between the center points of the front and rear motors; driving motors M1, M2, M3, and M4 represent the four driving motors 102 of the car; V represents the linear velocity (forward and backward speed) of the car 1, and ω represents the angular velocity (rotation speed) of the car. Through PID control of the linear velocity V and the angular velocity ω, it is converted into controlling the four motors, thereby controlling the movement of the four-wheel drive car 1. The car 1 of this structure is more flexible and convenient to move. A storage bin 104 can also be set on the car 1 to facilitate the remote transportation of the items to be grasped.

[0100] See also Figure 3 In some embodiments, the grabbing robot arm component 2 includes a servo 209, a linkage arm, a robot arm platform 201, a robot arm base 202, a large arm 208, a large arm connecting rod 203, a small arm 204, a posture retaining frame 205, a small arm connecting rod 206 and a grabbing piece. The robot arm platform 201 is assembled on the trolley 1, and the servo 209 and the robot arm base 202 are both assembled on the robot arm platform 201. The servo 209 is linked with the posture retaining frame 205 through the linkage arm. The posture retaining frame 205 is hingedly assembled on the large arm 208. The two ends of the small arm connecting rod 206 are respectively hingedly connected to the posture retaining frame 205 and the large arm connecting rod 203. The large arm 208 and the large arm connecting rod 203 are both assembled on the robot arm base 202.

[0101] It can be understood that the servo 209 drives the posture holding frame 205 to move through the linkage arm, the posture holding frame 205 is hinged to the upper arm 208, and the upper arm connecting rod 203 is hinged to the posture holding frame 205 through the small arm connecting rod 206, thereby achieving stable movement of the posture holding frame 205, and the robot arm gimbal 201 can drive the robot arm base 202 to rotate, thereby achieving adjustment of the grasping position of the grasping piece, and the overall grasping range of the grasping robot arm component 2 of this mechanism is wider.

[0102] See also Figure 3 and Figure 4The gripping member may include two types, and the two types may include a suction cup type and a manipulator. The suction cup 207 may be used to grasp objects with triangular and rectangular structures, while the manipulator may be used to grasp objects such as spheres. The manipulator may include a wrist body 210, a gripper base 211, a gripper servo 212 and a gripper 214. The number of grippers 214 may be two, and the two grippers 214 are symmetrically hinged and assembled on the gripper base 211. The gripper servo 212 is also assembled on the gripper base 211, and is used to drive the gripper 214 to move so as to achieve the approach of the opposite ends of the two grippers 214 to each other, thereby achieving the grasping of the object to be grasped. The wrist body 210 is assembled on the posture holding frame 205 and connected to the gripper base 211. The image acquisition component 3 may include a camera bracket 301 and a camera 302. The camera 302 is assembled on the gripper base 211 through the camera bracket 301.

[0103] From the above description, it can be known that the grabbing method and sorting robot for warehousing logistics provided by the present invention can identify the objects to be grabbed, distinguish the types of the objects to be grabbed, and select the corresponding manipulator for grabbing according to the types. The grabbing stability is better, the grabbing error rate is lower, and the efficiency is also greatly improved.

[0104] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A method for grabbing in warehousing logistics, characterized in that: The steps include: Determine the starting point according to the location information of the object to be grasped, and enter the area where the starting point is located; Acquire photos of the area where the starting point is located; Acquire characteristic information of the object to be identified in the photo, wherein the characteristic information at least includes shape, color and damage, and output a judgment result according to the characteristic information; When the judgment result meets the requirements, a corresponding grasping robot arm is selected for grasping according to the feature information.

2. The method for grabbing warehousing logistics according to claim 1, characterized in that: The step of determining the starting point according to the position information of the object to be grasped further includes: Obtaining the location information of the object to be grasped; Constructing a radar map around the object to be grasped; The starting point is determined by calculation based on the position of the object to be grasped and the position information of the object to be grasped.

3. The method for grabbing storage logistics according to claim 2, characterized in that: The step of calculating and determining the starting point based on the position of the self and the position information of the object to be grasped further includes: Determine the grasping range of the grasping robot arm; The starting point is calculated according to the position information of the object to be grasped and the own position in combination with the grasping range.

4. The method for grabbing warehousing logistics according to claim 2, characterized in that: The step of determining the starting point according to the position information of the object to be grasped also includes: When at the position of said starting point; Calculating the base angle according to the position information of the object to be grasped and the position information of the starting point; Whether it is within the grasping range of the grasping robot arm is judged in combination with the base angle, the position information of the object to be grasped and the position information of the starting point.

5. The method for grabbing warehousing logistics according to claim 1, characterized in that: Acquiring characteristic information of the object to be identified in the photo, wherein the characteristic information at least includes shape, color and damage, and outputting a judgment result according to the characteristic information, further comprising: Divide the input photo into an S×S grid and perform adaptive anchor box calculation to select a bounding box from the grid; Extract and fuse features for each bounding box to form a feature map; Predictions are made based on the feature map, and the world coordinates, size, and confidence of the bounding box are output to obtain an output result.

6. The method for grabbing warehousing logistics according to claim 5, characterized in that: The step of judging the object to be identified according to the feature information, and selecting a corresponding grasping robot arm to grasp the object according to the feature information when the feature information meets the requirements, further includes: Determining the type of the object to be identified according to the color information; In the case where the color information matches, A corresponding grasping robot arm is selected according to the shape information for grasping.

7. A sorting robot for warehousing logistics, characterized in that: The grabbing method for warehousing logistics adopts any one of claims 1 to 6, comprising a trolley, a grabbing robot arm component, an image acquisition component, a map component module and a controller component, wherein the grabbing robot arm component, the map component module and the controller component are all mounted on the trolley, the image acquisition component is mounted on the grabbing robot arm component, and the image acquisition component, the grabbing robot arm component and the map component module are all communicatively connected with the controller component.

8. The sorting robot for warehousing logistics according to claim 7, characterized in that: The trolley includes a bracket, at least four wheels and a corresponding number of drive motors. The drive motors and the wheels are both mounted on the bracket. The drive motors are used to drive the corresponding wheels to move. The drive motors are communicatively connected to the controller component. The grabbing robot arm component, the map component module and the controller component are all mounted on the bracket.

9. The sorting robot for warehousing logistics according to claim 7, characterized in that: The grabbing robot arm components include a servo, a linkage arm, a robot arm platform, a robot arm base, a large arm, a large arm connecting rod, a small arm, a posture retaining frame, a small arm connecting rod and a grabbing piece. The robot arm platform is mounted on the trolley. The servo and the robot arm base are both mounted on the robot arm platform. The servo is linked to the posture retaining frame through the linkage arm. The posture retaining frame is hingedly mounted on the large arm. Both ends of the small arm connecting rod are hingedly connected to the posture retaining frame and the large arm connecting rod respectively. The large arm and the large arm connecting rod are both mounted on the robot arm base.

Images