Kitchen waste sorting device and sorting method thereof

Abstract

The present application belongs to the technical field of garbage sorting, and particularly relates to a kitchen garbage sorting device and a sorting method thereof. The sorting device comprises a storage pool, an internal imaging device, a transmission device, a visual recognition device, a moving device and a mechanical arm grabbing device. The storage pool is used for storing the drained garbage. The internal imaging device performs non-contact density scanning on the kitchen garbage in the storage pool, and realizes accurate positioning of impurities through object density analysis and voxel modeling. In combination with visual and image processing technologies, the auxiliary control system calculates the spatial centroid position and size of the impurities. Under the guidance of the moving device, the mechanical arm grabbing device quickly and accurately moves to the target position, constructs an envelope sphere with the garbage centroid coordinates as the center, and thus realizes accurate grabbing of the impurities. The whole sorting process has the characteristics of high automation and high accuracy, and effectively promotes the development of the environmental protection industry.

Description

Technical Field

[0001] This invention belongs to the field of waste sorting technology, specifically relating to a kitchen waste sorting device and its sorting method. Background Technology

[0002] Global urbanization is accelerating, urban populations are increasing, and living standards are rising, leading to a surge in food waste. Statistics show that my country's annual urban food waste production reaches tens of millions of tons and continues to rise, making proper disposal a significant challenge. Traditional food waste treatment methods have obvious drawbacks. The sorting process relies on manual labor, which is inefficient and inaccurate. Limited by physical strength and attention, manual labor is prone to fatigue, leading to missed or misjudged items. Furthermore, labor costs are constantly rising, increasing operating costs.

[0003] Currently, most existing sorting technologies use crushing methods, which can cause solid impurities such as stainless steel tableware to be crushed along with other waste, resulting in equipment wear and damage, increased maintenance costs, and difficulty in separating impurities, which is not conducive to resource utilization and can also pollute the environment. Impurities take a long time to degrade in the natural environment, and improper handling can disrupt the ecological balance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, a food waste sorting device and its sorting method are provided, which enables precise sorting of impurities in food waste and effectively solves the problem of difficulty in separating solid impurities from food waste in existing technologies.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A kitchen waste sorting device, comprising: A storage pool for storing drained waste; An internal imaging device is used to scan and determine the spatial location of waste inside the storage pool; A transmission device, wherein the storage pool is provided on the output end of the transmission device, and the transmission device is used to drive the storage pool to reciprocate along a first direction to pass through the internal imaging device; A visual recognition device is used to identify the location of waste on the surface of the storage pool; A mobile device is mounted on the storage pool; A robotic arm gripping device is located at the working end of the mobile device and moves with the mobile device above the storage pool. The robotic arm gripping device is used to grip the garbage in the storage pool layer by layer according to the recognition results of the visual recognition device and the internal imaging device.

[0006] Based on the above technical solution, the embodiments of this application can be further improved as follows: In one embodiment, the internal imaging device includes: A ring-shaped drive module, the rotation axis of which is arranged along a first direction, the ring-shaped drive module having a channel formed at its center for passing through the storage pool and the transmission device, the ring-shaped drive module comprising: The rotating support has transmission teeth sequentially formed on its outer circumference; The first drive motor has a worm gear connected to its output end, which meshes with the transmission gear, and is used to drive the rotary support to rotate around its axis. Several radiation sources are installed at intervals on the inner wall surface of the annular drive module; Several detectors are installed at intervals on the inner wall surface of the annular drive module.

[0007] In one embodiment, the moving device includes: a first transmission module and a second transmission module, wherein the second transmission module is disposed on the output end of the second transmission module, and the movement directions of the second transmission module and the first transmission module are perpendicular to each other; The first transmission module includes: Two first guide rails are respectively set on two opposite sides of the top of the storage pool; Two intermediate plates, each with a first slider at its bottom, are slidably connected to two first guide rails via the first slider. A first rack is disposed on the side of any of the first guide rails; The first support frame is disposed on the middle plate at one end near the first rack; A first motor is mounted on the first support frame, and the output end of the first motor is connected to a first gear that meshes with the first rack. The first baffle is disposed at the end of the first guide rail.

[0008] In one embodiment, the second transmission module includes: Two second guide rails are provided, which are spaced apart along a direction perpendicular to the first guide rail, and the two ends of the second guide rails are respectively provided on the two intermediate plates; A movable plate has second sliders at both ends of its bottom, and the second sliders are slidably connected to the second guide rail. The visual recognition device is installed at the bottom of the movable plate. The second rack is disposed on the side of any of the second guide rails; The second support frame is disposed on the movable plate at one end near the second rack; The second motor is mounted on the second support frame, and the output end of the second motor is connected to a second gear that meshes with the second rack. The second baffle is located at the end of the second guide rail.

[0009] In one embodiment, the robotic arm gripping device includes a rotating end, a large arm, a joint, a small arm, and a mechanical gripper connected in sequence, wherein the rotating end is mounted on the bottom of the movable plate; The visual recognition device is installed at the end of the rotating end away from the moving plate. The visual recognition device includes a visual imaging module, a FIFO frame buffer, and a light source installed on the side.

[0010] The present invention also discloses a sorting method using the above-mentioned food waste sorting device, which includes the following steps: S1. Pour the drained kitchen waste into the storage tank; S2. Use an internal imaging device to scan the internal debris in the storage tank to obtain voxel information and density information, construct a three-dimensional coordinate system, and analyze the voxel information and density information based on the three-dimensional coordinate system to obtain the spatial centroid coordinates of the target debris. S3. Use a visual recognition device to scan the surface debris in the storage pool, calculate the centroid coordinates of each target debris in the two-dimensional plane, and sort them. S4. Combine the centroid position coordinates to locate the corresponding spatial centroid coordinates, and use the robotic arm gripping device to grab the target debris in order and move it to the designated position; S5. After one grab, return to step S3, compare the centroid coordinates of the remaining target debris before and after, and correct its spatial centroid coordinates accordingly. Then, grab the target debris on the surface of the storage pool layer by layer. S6. If an empty grab occurs, return to step S2. After the target debris in the storage pool is grabbed, process the remaining ungrabbed debris.

[0011] In one embodiment, step S2 includes: The internal imaging device is used to scan the debris inside the storage pool. The voxel information is formed by dividing the scan data into several voxel units. A preset threshold variable is used to compare the size of the voxel unit with the set threshold variable, and the target voxel unit is selected. The point cloud data of the target clutter is determined based on the target voxel unit and density information. Establish a three-dimensional coordinate system and construct the spatial centroid coordinates of all target debris.

[0012] In one embodiment, the spatial centroid coordinates for constructing all target debris include: The spatial centroid coordinates of the target debris are determined using the centroid calculation formula, whereby: ; Where N represents the number of voxels contained in the target substance. The three-dimensional coordinates of each voxel.

[0013] In one embodiment, step S3 includes: performing image preprocessing, region of interest segmentation, feature extraction and classification, and object area calculation on the image captured by the visual recognition device; establishing a two-dimensional coordinate system for the captured image; calculating and recording the centroid coordinates of each object in the two-dimensional plane; and sorting the objects by area.

[0014] In one embodiment, the step S4 of sorting and grabbing the target debris specifically includes: S4.1 Find the debris with the largest surface area and determine if there are multiple debris with similar areas. If so, proceed to step S4.2; otherwise, directly grab the debris with the largest surface area. S4.2 Compare the volumes of miscellaneous objects with similar areas to determine if there are multiple miscellaneous objects with similar volumes. If so, proceed to step S4.3; otherwise, grab the miscellaneous object with the largest volume. S4.3 Compare the densities of miscellaneous objects with similar volumes to determine if there are multiple miscellaneous objects with similar densities. If so, proceed to step S4.4; otherwise, grab the miscellaneous object with the highest density. S4.4 Compare the positions of debris with similar densities and grab the debris closest to the robotic arm.

[0015] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This invention integrates internal imaging devices such as ring-shaped CT scanning technology, visual recognition devices, and automated robotic arm gripping devices to achieve accurate identification and sorting of debris in food waste. The high-precision scanning data provided by the ring-shaped CT scan details the internal structure and density information of the food waste pool. Combined with 3D modeling and debris recognition technology, it can accurately distinguish different types of debris and precisely calculate their spatial location, ensuring the accuracy and efficiency of sorting. In actual tests, the sorting accuracy rate is high for common food waste debris, such as bones and plastic bottles, greatly improving the sorting quality of food waste.

[0016] 2. This invention improves the recognition rate by combining CT scanning technology and visual recognition technology. To address the issue of displacement of waste within the pool caused by the previous grabbing, visual recognition is performed again before and after each grabbing operation. The centroid deviation value is calculated on a two-dimensional level to correct the centroid coordinates of the grabbing operation on a three-dimensional level. This eliminates the need for a second CT scan after each grabbing operation.

[0017] 3. This invention uses visual recognition to prioritize and lock onto large, high-area debris targets on the upper layers of the stack. These targets (such as large pieces of fruit and vegetable peels, sheet metal, and plastic bottles) have a larger contact surface area and a more stable physical shape, making it easier for the robotic gripper to achieve stable and reliable grasping and fixation. The high success rate of single-attempt grasping reduces time delays and energy consumption caused by repeatedly trying to grasp small, slippery targets, thus significantly improving the overall efficiency of the cleaning operation. Because of the high success rate and fewer attempts required to grasp large targets, and because it avoids frequent, high-precision fine-tuning operations for grasping small targets, this invention effectively reduces the number of empty grasps, slippage, and invalid collisions by the gripper, thereby reducing wear on mechanical components, extending the service life of key actuators such as grippers and suction cups, and lowering the long-term maintenance costs of the system.

[0018] 4. This invention uses visual recognition to identify the shortest point of an object's outline and adaptively adjusts the robotic gripper's posture for grasping. Compared to previous grasping strategies, where the robotic gripper typically approaches the target with its maximum or fixed opening to ensure coverage, this invention accurately identifies the shortest distance of the object's outline. This allows the robotic gripper to complete the grasp by opening only slightly wider than this minimum width, significantly reducing the stroke and work of the gripper's servo motor and avoiding unnecessary energy consumption. The energy efficiency improvement is particularly significant in high-frequency, high-volume continuous operations. The shortened opening and closing stroke of the robotic gripper directly translates to a reduction in the action time of a single grasp. Since the gripper does not need to perform a complete, maximum-stroke opening and closing motion each time, the entire grasping cycle is optimized, thereby increasing the number of grasps per unit time and significantly improving the overall production efficiency and throughput of the automated system. Furthermore, this strategy enhances the adaptability of the mechanical gripper in densely stacked environments, avoids interference and collisions with surrounding debris by reducing the opening amplitude, and finally, the reduction in mechanical movement stroke and load also means reduced mechanical wear, which helps to extend the service life of the gripping device and reduce maintenance costs.

[0019] 5. This invention can automatically complete a series of processing steps, including scanning, identifying, locating, and grabbing waste, greatly reducing manual operation. From the moment kitchen waste enters the storage tank to the successful grabbing and transfer of debris, the entire process is performed automatically by the equipment, requiring minimal human intervention. This fully automated operation mode not only reduces reliance on manual labor and lowers labor costs but also avoids sorting errors caused by human error, thus improving work efficiency.

[0020] 6. This invention integrates multiple processing stages into a single device through an integrated design, achieving centralized functionality. The various processing stages within the device are closely interconnected, forming a highly efficient processing system. From data acquisition via circular CT scans or cameras to data analysis and processing by the host computer and control of the robotic arm for grasping, the connections between each stage are seamless, allowing waste to be processed quickly and efficiently. Compared to traditional decentralized processing equipment, the equipment of this invention can process more food waste per unit time, effectively improving the processing efficiency of food waste.

[0021] 7. The equipment of this invention is suitable for processing needs ranging from small amounts of kitchen waste (1 kg) generated by households to large-scale waste treatment plants (5 tons). Whether it's a small amount of mixed waste from a household kitchen or a large amount of kitchen waste from large catering businesses or canteens, this equipment can leverage its advantages to perform efficient sorting and processing. By adjusting the equipment's operating parameters, such as the CT beam emission frequency and the gripping force of the mechanical claws, it can adapt to the processing requirements of kitchen waste of different scales and compositions, demonstrating strong versatility.

[0022] 8. This invention employs a non-crushing and non-contact scanning sorting technology, avoiding the excessive wear and damage to equipment caused by traditional crushing methods. It reduces the frequency of repairs and parts replacements due to frequent equipment failures, thereby lowering equipment maintenance costs. 9. This invention effectively separates solid impurities from kitchen waste through precise sorting, preventing these impurities from polluting the environment during subsequent processing. For example, the separated metal impurities can be recycled and reused, reducing the waste of metal resources; plastic impurities can be specially recycled after sorting, preventing plastic waste from polluting soil and water bodies. Furthermore, the noise and wastewater generated during equipment operation are effectively controlled, meeting environmental protection requirements and contributing to ecological conservation. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2 This is a schematic diagram of the base module of the present invention; Figure 3This is a schematic diagram of the transmission device on the base module of the present invention; Figure 4 This is a schematic diagram of the storage pool structure of the present invention; Figure 5 This is a schematic diagram of the internal imaging device of the present invention; Figure 6 This is a schematic diagram of the internal imaging device of the present invention; Figure 7 This is a schematic diagram showing the installation of the storage pool, mobile device, and robotic arm grasping device of the present invention. Figure 8 This is a schematic diagram of the structure of the mobile device of the present invention; Figure 9 This is a schematic diagram of the installation of the first transmission module of the mobile device of the present invention; Figure 10 This is a schematic diagram showing the installation of the second transmission module and the robotic arm gripping device of the mobile device of the present invention; Figure 11 This is a schematic diagram of the robotic arm gripping device of the present invention; Figure 12 This is a schematic diagram showing the installation of the camera and light source in the machine vision system of the present invention; Figure 13 This is a flowchart of the workflow of the present invention; Figure 14 This is a diagram illustrating the sorting and grabbing rules for debris in this invention. Detailed Implementation

[0025] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention. It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0026] In the description of this application, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0027] Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly defined.

[0028] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0029] To effectively overcome the shortcomings, a food waste sorting device and its sorting method are proposed. By integrating internal imaging devices such as industrial CT scanning technology, visual recognition devices such as machine vision, image processing technology, and robotic arm gripping devices, it can achieve accurate sorting of impurities in food waste and effectively solve the problems existing in the prior art.

[0030] In summary, the process begins with draining the water from the food waste and pouring it into a storage tank. A transmission device then moves the tank through an internal imaging system, such as a circular CT scanner. The circular CT scanner performs a non-contact density distribution scan of the food waste in the storage tank, generating detailed three-dimensional voxel data. Subsequently, the control system constructs a three-dimensional coordinate system based on this voxel data and identifies various debris by analyzing the density of the objects. Next, the control system calculates the position of the centroid of each debris within the three-dimensional coordinate system based on the three-dimensional voxel data. During the time the CT scanner is acquiring information, a camera in a vision recognition system located directly above the storage tank, under sufficient lighting, captures images of the debris at the top of the tank. The control system preprocesses and matches the captured images with templates, sorting the debris in descending order of area. The control system establishes a two-dimensional coordinate system for the captured images, calculates and records the centroid coordinates of each debris in the two-dimensional plane. Finally, the control system sends this precise coordinate information and the grasping order to the robotic arm control unit. Upon receiving location information from the control system, the system prioritizes grabbing the topmost layer of debris, proceeding in the designed grabbing sequence. Guided by the mobile device, the robotic arm quickly and accurately moves to the target location, constructing an envelope sphere centered on the centroid coordinates of the debris. Using a machine vision system as a guide, the robotic gripper adjusts its angle to grab the shortest point of the debris, thus precisely grabbing and extracting it. After grabbing, the camera re-images the top layer of debris in the pool and calculates the centroid coordinates of the remaining debris. This is compared with the previous centroid coordinates to correct the centroid coordinates for the next grab. If an empty grab occurs, the conveyor re-drives the storage pool for a CT scan, repeating the above process. Small debris and small amounts of residual water are centrally processed and are not considered for grabbing.

[0031] Specifically, such as Figure 1-12 As shown, the present invention provides a kitchen waste sorting device, which includes: a storage tank, an internal imaging device, a transmission device, a visual recognition device, a moving device and a robotic arm gripping device, a base and a control system.

[0032] The base 11 is a rigid frame structure with sufficient strength and stability to support the entire device. Six wheels 12 are mounted on the bottom of the base 11, evenly distributed around its perimeter to ensure balance and flexible movement. The wheels 12 may be equipped with brakes to allow the device to be fixed in place when needed. The wheels 12 enable easy movement of the entire device to the working position.

[0033] A support plate 13 is fixedly installed at the center of the upper part of the base 11. The support plate 13 is a flat plate structure, which is fastened to the base 11 by welding or bolting. It is used to support and install the transmission device 14. The four corners of the bottom of the support plate are provided with legs to raise the entire transmission device, so that the transmission device can pass through the middle of the internal imaging device. The transmission device 14 is mounted on the support plate 13 and is used to provide linear transmission power to drive the support plate and the storage pool 2 on it through the annular internal imaging device 3. In this embodiment, the transmission module 14 adopts a ball screw transmission structure, but it is understood that other transmission methods such as synchronous belt transmission can also be used instead. The transmission module 14 specifically includes a motor 141, a motor base 142, a coupling 143, a bearing 144, a bearing seat 145, a ball screw 146, a slide sleeve 147, and a slide table 148.

[0034] Specifically, the base module 1 is located at the bottom of the overall device. Through the above structure, it realizes multiple functions such as placing other devices, facilitating the movement of the overall device, and transmitting the storage tank. When the motor 141 is working, the ball screw 146 rotates, driving the slide sleeve 147 and the slider stage 148 to move axially along the ball screw 146, thereby driving the support plate and the storage tank 2 through the scanning area of ​​the annular CT device 3, realizing the automated transmission function.

[0035] The storage pool 2 mainly comprises two parts: a square tube support 21 and a waste pool 22. The square tube support 21 is welded or bolted together from standard square steel tubes, and its overall configuration is a three-dimensional frame structure. The internal dimensions of this frame are slightly larger than the external dimensions of the waste pool 22, allowing the waste pool 22 to be precisely embedded or fitted within the internal space of the square tube support 21. The waste pool 22 is an open-top container, preferably made of welded steel plate or injection molded engineering plastic, and can be treated for corrosion resistance or anti-adhesion. The bottom or sidewalls of the waste pool 22 can be fixedly connected to the square tube support 21 by bolts, clips, or welding, thereby ensuring that the two are combined into a whole to form a complete storage pool 2. The entire storage pool 2 is fixedly mounted on the transmission device 14 of the base 1 via the base of the square tube support 21 below it. The transmission device is used to drive the storage pool to reciprocate along a first direction to pass through the internal imaging device. The first direction is a horizontal direction and passes through the center of the internal imaging device.

[0036] Specifically, the square tube support 21 provides stable support and reinforcement for the garbage bin 22, preventing it from deforming when moving or carrying garbage. The garbage bin 22 is used to store collected kitchen waste. After the storage bin 2 is fixed on the slider table 148, it can move stably in a straight line along the path specified by the lead screw 146 in the transmission device 14 under the drive of the transmission motor 141.

[0037] The internal imaging device is used to scan and determine the spatial location of the waste inside the storage tank. It can be implemented by a ring CT device. The ring CT machine performs a non-contact density distribution scan on the kitchen waste in the storage tank to generate detailed three-dimensional voxel data. Subsequently, the control system constructs a three-dimensional coordinate system based on these voxel data and identifies various debris by analyzing the density of the objects. The control system can be implemented using a host computer, such as a PC. In this embodiment, the data collected by the internal imaging device and the visual recognition device are all sent to the control system for processing.

[0038] The visual recognition device is used to identify the location of the waste on the surface of the storage pool, and sends the collected image information to the control system. The mobile device is set on the storage pool. The robotic arm gripping device is located at the working end of the mobile device and moves above the storage pool with the mobile device. The robotic arm gripping device is used to grip the garbage in the storage pool layer by layer according to the recognition results of the visual recognition device and the internal imaging device.

[0039] The annular CT device 3 is the core functional module of the entire equipment, comprising: an annular drive module 31, several X-ray sources 32, several detectors 33, a CT scanner housing 34, and connecting plates 35. The annular drive module 31 is further composed of an annular rotary support 311, a first drive motor 312, bearings 313, bearing seats 314, and a worm gear 315. The CT scanner housing 34 serves as the structural foundation of the entire annular CT device 3, and is fixedly mounted to the center of the upper surface of the base 1 via four connecting plates 35. These four connecting plates 35 are preferably evenly distributed at the bottom of the CT scanner housing 34, and are reliably fixed using bolt connections or other methods, ensuring the stability of the CT device 3 on the base 11. The annular rotary support 311 is the core moving component that carries the imaging elements. Annular transmission gears are machined or fixedly mounted on its outer cylindrical surface; this gear structure is preferably a helical gear or similar structure that can mesh well with the worm gear 315. The worm gear 315 is horizontally positioned, and its two ends are supported by the bearings 313 and bearing seats 314. The bearing housing 314 is bolted to the upper surface inside the CT scanner housing 34, thereby suspending the worm gear 315 in a predetermined position. The first drive motor 312 is also fixedly mounted to the upper surface inside the CT scanner housing 34, and its output shaft is connected to one end of the worm gear 315 via a coupling or synchronous belt to provide power. The teeth of the worm gear 315 precisely mesh with the annular transmission teeth on the annular rotary support 311, forming a worm gear pair. This transmission method features smooth transmission, low noise, and reverse self-locking, preventing the annular rotary support 311 from rotating on its own. Multiple X-ray sources 32 and detectors 33 appear in pairs and are uniformly and fixedly installed on the inner ring surface of the annular rotary support 311 at certain intervals. The X-ray sources 32 and detectors 33 are radially opposite each other, ensuring that the X-rays can penetrate the object being detected and be received by the detector 33 opposite. The transmission device 14 in the base mold 1 extends through the central hole area of ​​the annular CT device 3 in its spatial layout. This design allows the storage pool 2, which is carried on the base module 1, to pass through the CT scan area in a straight line.

[0040] Specifically, when a CT scan is required, the first drive motor 312 starts, driving the worm gear 315 to rotate. The worm gear 315, through meshing with the annular transmission gear, transmits the rotational motion to the annular rotary support 311, causing the entire annular rotary support 311 to rotate smoothly 360 degrees around its central axis. All the X-ray sources 32 and detectors 33 fixed on it rotate synchronously, thereby enabling the acquisition of multi-angle projection data of the kitchen waste located within the scanning area.

[0041] The mobile device 4 is mounted above the storage pool 2, and its core function is to drive the mobile device 4 and its actuators to move precisely within a two-dimensional horizontal plane. The device mainly consists of two stages of motion mechanisms: a first transmission module 41 and a second transmission module 42. The second transmission module is located at its output end, and its movement direction is perpendicular to that of the first transmission module.

[0042] The first transmission module 41 is further composed of a first support frame 411, a first motor 412, a first gear 413, a first rack 414, a first guide rail 415, a slider 416, a first baffle 417, and an intermediate plate 418. Two first guide rails are respectively disposed on two opposite sides of the top of the storage pool; two intermediate plates are provided with first sliders at their bottoms, and the two intermediate plates are slidably connected to the two first guide rails through the first sliders; a first rack is disposed on the side of any of the first guide rails; a first support frame is disposed on one end of the intermediate plate near the first rack; a first motor is mounted on the first support frame, and the output end of the first motor is connected to a first gear that meshes with the first rack; a first baffle is disposed at the end of the first guide rail.

[0043] The second transmission module 42 is further composed of a square tube 421, a second support frame 422, a second motor 423, a second gear 424, a second rack 425, a second guide rail 426, a small slider 427, a second baffle 428, and a moving plate 429.

[0044] Two second guide rails are spaced apart along a direction perpendicular to the first guide rail, and the two ends of the second guide rails are respectively disposed on the two intermediate plates; the bottom ends of the movable plate are provided with second sliders, the second sliders are slidably connected to the second guide rails, and the visual recognition device is installed on the bottom of the movable plate; a second rack is disposed on the side of any of the second guide rails; A second support frame is disposed on one end of the movable plate near the second rack; a second motor is mounted on the second support frame, and the output end of the second motor is connected to a second gear that meshes with the second rack; a second baffle is disposed at the end of the second guide rail.

[0045] The first transmission module 41 is used to move the entire gantry device 4 along the left-right direction of the storage tank 2. The first guide rails 415 are fixedly and symmetrically installed on the upper surface of the square tube support 21 outside the storage tank 2. Each end of the first guide rail 415 is equipped with a first baffle 417 to limit the movement stroke. Each first guide rail 415 is equipped with a slider 416. An intermediate plate 418 is fixedly installed above each of the two sliders 416. To maintain structural rigidity and achieve synchronous movement, the two ends of the square tube 421 are fixed to the two intermediate plates 418 respectively, connecting them into an integral frame. A first motor 412 is fixedly installed on one of the intermediate plates 418 through a first support frame 411. The output shaft of the first motor 412 is connected to a first gear 413. A first rack 414 that meshes with the first gear 413 is fixedly installed on the side surface of the square tube support 21 of the storage tank 2, and the direction of the first rack 414 is parallel to the first guide rail 415. The second transmission module 42 is mounted on the first transmission module 41 and is used to realize the movement of the actuator in a forward-backward direction perpendicular to the left-right direction. The second guide rail 426 is fixedly mounted on the upper surface of the square tube 421, and its extension direction is perpendicular to the first guide rail 415 below. The second transmission module 42 is basically the same as the first transmission module 41 in structure and working principle, but smaller in scale. A movable plate 429 is fixed above the small slider 427, which is the final mounting platform for the actuator of the moving device 4.

[0046] Specifically, when the first motor 412 starts, it drives the first gear 413 to rotate. Since the first rack 414 is fixed, the rotational motion of the first gear 413 is converted into linear motion of the first motor 412 itself and the first support frame 411 and the intermediate plate 418 fixed thereto. This motion is transmitted to the intermediate plate 418 on the other side through the square tube 421, so that the entire frame composed of the two intermediate plates 418 and the square tube 421 moves stably left and right along the path determined by the two first guide rails 415 under the action of the four sliders 416. Similarly, the second transmission module 42 works on the same principle. In summary, through the superposition and combination of the first transmission module 41 and the second transmission module 42, the robotic arm gripping device 5 fixed on the moving plate 429 obtains the ability to move in a two-dimensional plane.

[0047] The robotic arm gripping device 5 serves as the end effector of the entire device. It is fixedly mounted on the moving plate 429 of the aforementioned moving device 4 and includes a rotating end 51, a large arm 52, a joint 53, a small arm 54, a robotic gripper 55, and a tension rod 56 connected in sequence. The rotating end 51 is the base of the robotic arm gripping device 5, with its base plate fixedly mounted on the moving plate 429 of the aforementioned second transmission module 42. The rotating end 51 contains a first drive source (a servo motor or stepper motor), and a reduction transmission mechanism is formed by a pair of meshing transmission gears, transmitting the motor's rotational motion to the connected large arm 52. This design allows the entire robotic arm gripping device 5 to perform a continuous 360° rotation around a vertical axis perpendicular to the moving plate 429. Through this movement, the robotic arm gripping device 5 can find the optimal gripping angle within a circular range without moving the moving device 4. The large arm 52 is a rigid arm, with its proximal end fixedly connected to the output shaft of the rotating end 51. The swing of the upper arm 52 is controlled by a second drive source. Driven by the second drive source, the upper arm 52 can pitch in the vertical plane, thereby changing the height and radial extension distance of the robotic gripper 5. The joint 53 is the hub connecting the distal end of the upper arm 52 and the proximal end of the forearm 54. The joint 53 contains a third drive source, which functions to drive the forearm 54 to bend or extend relative to the upper arm 52. The extension rod 56 integrates a telescopic mechanism. Through the action of a fifth drive source, the extension rod 56 can extend and retract along its axial direction. The extension length controls the degree to which the robotic gripper 55 opens and closes.

[0048] Specifically, the visual recognition device is installed at the end of the rotating end away from the moving plate, and the visual recognition device includes a visual imaging module, a FIFO frame buffer, and a light source installed on the side.

[0049] Specifically, during operation, the moving device 4 first moves the robotic arm gripping device 5 above the target area. Then, the rotating end 51 drives the robotic arm gripping device 5 to rotate to a suitable posture; the upper arm 52 and joint 53 work together to position the robotic claw 55 at the end of the forearm 54 directly above the target object; the extension rod 56 extends, causing the robotic claw 55 to open; according to the position of the center of mass of the debris space given by the control system, the robotic claw 55 forms an envelope sphere around the debris; finally, the extension rod 56 closes, the robotic claw 55 closes, and the object is grasped.

[0050] This invention also discloses a sorting method using the above-mentioned food waste sorting device, such as... Figure 13 As shown, it includes the following steps: Step S1: Pour the drained kitchen waste into the storage tank. Since too much liquid in the kitchen waste can easily affect the imaging effect of the visual recognition device, try to filter and drain the liquid before pouring it into the storage tank.

[0051] Since steps S2 and S3 can be performed simultaneously, the order of steps S2 and S3 is irrelevant.

[0052] Step S3: Use a visual recognition device to scan the surface debris in the storage pool, calculate the centroid coordinates of each target debris in the two-dimensional plane, and sort them. The visual recognition device mainly consists of an industrial camera 61 and a light source 62. The industrial camera 61 uses a CMOS camera and a FIFO frame buffer. The camera 61 needs to be fixedly mounted below the moving plate 429 of the moving device 4, so that when the moving plate 429 moves to a specific position, the camera 61 can capture an image of the surface layer of the storage tank 2. The light source 62 is installed next to the camera 61.

[0053] Specifically, under multiple light sources 62, the camera 61 captures images and transmits them to the control system in real time, converting them into frame-by-frame images. The transmitted images are then processed.

[0054] Step S3 includes: performing image preprocessing, region of interest segmentation, feature extraction and classification, and object area calculation on the image captured by the visual recognition device; establishing a two-dimensional coordinate system for the captured image; calculating and recording the centroid coordinates of each object in the two-dimensional plane; and sorting the objects by area.

[0055] Specifically, the image preprocessing includes: grayscale conversion, where a grayscale image has many color depth levels, namely 255 levels, and after image grayscale processing, the color image is converted into a grayscale image with different color depth levels; filtering and denoising, using algorithms such as median filtering and Gaussian filtering to eliminate random noise in the image; and contrast enhancement, using methods such as histogram equalization to enhance image contrast and make garbage features more obvious.

[0056] Next, the segmentation of the region of interest includes using edge detection methods, such as Canny and Sobel operators to detect the outline of the garbage pit, taking pictures of the garbage pit in a scene without garbage and extracting the edges of the garbage pit, learning and saving them as template images, and extracting the grayscale image of the garbage pit by template matching through the template image during normal operation, thus obtaining an image containing only the garbage pit and the kitchen waste inside. Then, feature extraction and classification include: establishing a template library, creating a series of templates for each type of waste at different scales and rotation angles, extracting features from the templates and matching them with the regions to be tested; shape features: extracting shape features such as Hu moments and Zernike moments of the segmented regions that are unaffected by translation, rotation, and scaling; texture features: extracting texture features such as LBP (Local Binary Pattern) and HOG (Histogram of Oriented Gradients).

[0057] Design matching algorithm: Use similarity measurement algorithms, such as cosine similarity and Euclidean distance, to calculate the matching degree between the features of the region to be tested and the features of the template library, and take the highest score as the classification result.

[0058] Finally, a two-dimensional coordinate system is established on the screen to calculate the area of ​​each piece of trash within its respective screen and the position of the centroid of the irregular shape within the two-dimensional coordinate system. Trash is categorized by area size, sorted in descending order of area, and the results are fed back to the control system.

[0059] Step S2: Use an internal imaging device to scan the internal debris in the storage pool to obtain voxel information and density information, construct a three-dimensional coordinate system, and analyze the voxel information and density information based on the three-dimensional coordinate system to obtain the spatial centroid coordinates of the target debris. This step specifically includes: This step includes using an internal imaging device to scan the internal debris of the storage pool, and the voxel information is a number of voxel units formed by dividing the scan data. Industrial CT, based on the principle of X-ray attenuation, uses the differences in X-ray absorption and attenuation by materials (related to density) to achieve accurate identification through detectors and computer processing. First, an industrial CT density model must be established. The densities of common solid impurities in kitchen waste are as follows: iron approximately 7.86 g / cm³, ceramics approximately 2.3-2.5 g / cm³, 304 stainless steel approximately 7.93 g / cm³, and bone approximately 1.7-2.0 g / cm³. During scanning, the X-rays emitted by the industrial CT machine penetrate the waste mixture. High-density materials like iron and stainless steel absorb X-rays strongly, appearing as bright areas in the CT image; lower-density materials such as ceramics and bone appear as areas of varying grayscale.

[0060] By employing image thresholding and edge detection algorithms, combined with impurity shape and texture feature analysis, even stacked impurities can be accurately located and identified. Based on point cloud voxel segmentation technology, CT scan data is finely divided, using 1mm³ as the basic voxel unit to construct a three-dimensional data grid. The volume of each impurity can be compared using a voxelization method, with the specific formula: Volume V = Total number of voxels for the impurity × Volume of a single voxel.

[0061] Different substances exhibit different CT values ​​in CT scan images. This provides a basis for distinguishing substances by setting thresholds. For example, metallic substances often show higher CT values ​​in CT images. If the CT value reaches or exceeds 2000 HU, there is a high probability of a metallic object present. The CT value threshold can be set using the following formula: ; In the CT program processing stage, the primary step is to standardize the reading of DICOM data. MATLAB's `dicomread` function can quickly load the raw 3D data from industrial CT scans, while the `dicominfo` function retrieves scan parameters and related metadata. A try-catch exception handling mechanism is employed to prevent errors and program termination due to file corruption or formatting errors. The raw pixel values ​​stored in the DICOM file need to be converted into clinically meaningful CT values ​​(Hounsfield Units). The conversion formula is: True CT value = Slope * Pixel value + Intercept.

[0062] The RescaleSlope and RescaleIntercept parameters are stored in the metadata, and pixel values ​​are converted to double type to avoid calculation overflow.

[0063] A binary mask is obtained by threshold segmentation. The threshold setting here needs to be adjusted according to the actual scanning protocol. The threshold should be lowered when there are obvious metal artifacts.

[0064] Then, 3D volumetric data visualization technology is used to select the middle layer for display, as this layer is representative of the whole and can reduce edge interference. The `imshow(I,[])` function can achieve automatic grayscale adaptation, optimizing the contrast and detail reproduction of the image. Here, `imshow` is a MATLAB command for displaying images, `I` is the image matrix to be displayed, and `[]` contains two numbers. A grayscale image can be divided into 255 levels according to its blackness, where 0 is black, 255 is white, and the values ​​in between represent different shades of gray. The two numbers entered are within this range, thus achieving automatic grayscale adaptation.

[0065] The original image and the mask are displayed side-by-side for intuitive comparison of the target area. When quantifying the detection results, the actual physical size should be considered. If the metadata includes a PixelSpacing field (unit: mm / pixel), the actual area of ​​the metallic region can be calculated.

[0066] A preset threshold variable is used to compare the size of the voxel unit with the set threshold variable, and the target voxel unit is selected. The point cloud data of the target clutter is determined based on the target voxel unit and density information. First, a threshold variable `metalThreshold` is defined and set to 2000 (the specific threshold setting needs to be adjusted according to the actual scanning protocol). Next, using the comparison operator "≥", all elements in `huVolume` (the 3D data matrix calibrated by CT values) with values ​​greater than or equal to 2000 are filtered out, creating a logical matrix `mask` of the same size as `huVolume`. In the `mask` matrix, elements that meet the condition (i.e., voxel positions corresponding to CT values ​​≥ 2000 HU) are assigned the logical value "1", while elements that do not meet the condition are assigned the value "0". This step is equivalent to drawing a "mask" in the entire CT data space, initially delineating areas where metallic substances may exist, allowing for targeted investigation. For each voxel, only the voxel corresponding to the target substance (value 1) is retained, while background and irrelevant substances (value 0) are removed, thus quickly and accurately extracting point cloud data of solid impurities such as iron, ceramics, stainless steel, and bone.

[0067] Establish a three-dimensional coordinate system and construct the spatial centroid coordinates of all target debris.

[0068] Specifically, the spatial centroid coordinates of all target debris include: The spatial centroid coordinates of the target debris are determined using the centroid calculation formula, whereby: ; Where N represents the number of voxels contained in the target substance. The three-dimensional coordinates of each voxel.

[0069] This formula uses a weighted average method to process the coordinates of all target voxels, enabling high-precision positioning of the impurity centroid. In the actual program, the `regionprops3` function can conveniently calculate various attributes of each marked region in the 3D image, including the centroid coordinates. A 100×100×100 maskLabeled 3D matrix (this matrix is ​​a labeling matrix used to label waste 1, 2, 3… in the storage pool, facilitating subsequent calculations of the volume of waste 1, 2, 3…) is created to simulate the actual working conditions of food waste sorting: cubes in the 20-30 voxel range are assigned the value 1, representing the initial object to be sorted; cubes in the 70-80 voxel range are assigned the value 2, identifying the second object to be sorted. This labeling method creates a structured data foundation for subsequent image segmentation, volume calculation, and feature extraction, facilitating the exploration of the spatial distribution patterns and morphological differences of different components. In food waste sorting scenarios, the `regionprops3` function can accurately analyze metallic impurities identified by CT scans: it obtains the centroid coordinates of the impurities through the `Centroid` attribute of the `regionprops3` function (the result of the `regionprops3` function includes the centroid coordinates), providing a precise gripping position for the robotic arm. Furthermore, the `cat` function integrates the features of multiple impurities into a numerical matrix, facilitating intelligent sorting and efficient sorting of metallic impurities of different densities and sizes, thus improving resource recovery efficiency.

[0070] Step S4: Combine the centroid position coordinates to locate the corresponding spatial centroid coordinates, and use the robotic arm gripping device to grab the target debris in order and move it to the designated position; Specifically, combining the spatial centroid coordinates obtained from CT scanning by the internal imaging device and the two-dimensional centroid coordinates of the visual recognition device, the device first locates the waste with the largest surface area based on visual perception. Then, based on the centroid coordinates (x1, y1) of this waste, it finds its corresponding three-dimensional centroid coordinates (x, y, z). The robotic gripper is then controlled to open and enclose the waste based on these centroid coordinates. The robotic arm gripping device operates continuously according to the gripping order from largest to smallest.

[0071] A visualization framework is established to intuitively display the distribution of impurities identified by industrial CT scans. A 3D scene is built using independent view windows (figures), employing layer overlays to present multiple types of impurities. Objects are processed according to their marked order: first, spatial coordinates are located, then surface contours are extracted, and a semi-transparent shell is drawn with random colors to distinguish materials. Marking the centroid provides a target point for the robotic arm to grasp, and a lighting system (camlight / lighting) enhances the visual difference between metals and organic matter, helping operators quickly identify impurity types and spatial locations, improving sorting efficiency. Based on the centroid coordinate values, subsequent sorting equipment can accurately plan the optimal grasping path, effectively sorting solid impurities from food waste, achieving higher accuracy and faster sorting speed.

[0072] Then, in step S5: after one grab, return to the point where the visual recognition device is used to scan the surface debris in the storage pool, calculate the centroid position coordinates of each target debris in the two-dimensional plane and sort them, compare the centroid position coordinates of the remaining target debris before and after and correct its spatial centroid coordinates accordingly, and then grab the target debris on the surface of the storage pool layer by layer. A second visual recognition process is performed using a camera to obtain the centroid coordinates (x2, y2) of various types of debris after the first grab. This is compared to the previously recorded centroid coordinates (x1, y1), and the resulting deviation (x2-x1, y2-y1) is added to the centroid coordinates (x, y, z) of each type of debris, forming new centroid coordinates for each type of debris. The robotic arm then performs the grabbing steps described above based on these new centroid coordinates.

[0073] The effect of this is as follows: By using machine vision technology, images are taken before and after the grabbing process. To prevent the garbage from shifting and changing its centroid coordinates due to the previous grabbing, the centroid coordinates of the garbage to be grabbed in the next grabbing are corrected by combining the centroid coordinate deviation value.

[0074] If an empty grab occurs, the process returns to using the internal imaging device to scan the debris inside the storage tank, obtain voxel information and density information, construct a three-dimensional coordinate system, analyze the voxel information and density information based on the three-dimensional coordinate system to obtain the spatial centroid coordinates of the target debris, perform a CT scan and camera image acquisition again, repeat the above work, and after the target debris in the storage tank is grabbed, the remaining ungrabbed debris is processed centrally.

[0075] Among them, such as Figure 14 As shown, step S4, which involves sorting and grabbing the target debris, specifically includes: S4.1 Find the debris with the largest surface area and determine if there are multiple debris with similar areas. If so, proceed to step S4.2; otherwise, directly grab the debris with the largest surface area. S4.2 Compare the volumes of miscellaneous objects with similar areas to determine if there are multiple miscellaneous objects with similar volumes. If so, proceed to step S4.3; otherwise, grab the miscellaneous object with the largest volume. S4.3 Compare the densities of miscellaneous objects with similar volumes to determine if there are multiple miscellaneous objects with similar densities. If so, proceed to step S4.4; otherwise, grab the miscellaneous object with the highest density. S4.4 Compare the positions of debris with similar densities and grab the debris closest to the robotic arm.

[0076] In this embodiment, objects with large areas and volumes are prioritized for gripping. This is because large objects are easier for the robotic gripper to grasp, avoiding situations where the gripper grabs empty or leaves debris in mid-air.

[0077] Furthermore, the workflow of this embodiment can be summarized as follows: 1. Scanning Data Acquisition Stage: After the kitchen waste to be processed is drained, it is poured into the storage tank 2. The transmission module 14 then guides the storage tank 2 through the annular radiation device 3. The annular CT device 3 is activated, and its radiation source 32 begins emitting CT rays. At this time, the rays penetrate the kitchen waste in real time, acquiring its internal structure and density information. Simultaneously, the annular drive module 31 is activated, driving the radiation source 32 and detector 33 to rotate under the drive of the first drive motor 312, performing a complete scan of the kitchen waste in the storage tank 2. During the scan, the detector 33 continuously receives the rays and converts the received signals into electrical signals, forming voxel information. This voxel information is sent to the control system in real time, providing raw data for subsequent debris identification and location.

[0078] 2. Deep Debris Identification and Location Stage: After receiving voxel data from the CT scan, the control system uses algorithms to construct a three-dimensional spatial coordinate system. This coordinate system provides a framework for precise debris location. Next, through in-depth analysis of object density, the control system accurately identifies various debris in the kitchen waste, such as bones, plastics, and metals. Then, based on the data analyzed using the constructed three-dimensional coordinate system, the control system calculates the spatial centroid position of the debris.

[0079] 3. Visual Processing of Top Layer Debris: During the CT scan operation described above, camera 61, located directly above storage tank 2, captures images of the debris at the top of the storage tank under sufficient lighting. The control system performs image preprocessing, region of interest segmentation, feature extraction and classification, and calculation and sorting of debris area sizes. The control system establishes a two-dimensional coordinate system for the captured images, calculates and records the centroid coordinates (x1, y1) of each debris in the two-dimensional plane, and sorts the debris by area from largest to smallest for grasping. This precise positional information and grasping order are sent to the control terminal of the robotic arm grasping device 5, providing accurate guidance for its grasping operation.

[0080] 4. Robotic Arm Positioning Stage: After receiving the position information sent by the control system, the robot prioritizes grabbing the topmost layer of debris, and then grabs it in descending order of size (the specific order is detailed in step S4). The robot identifies the debris with the largest surface area using the vision system, and finds its centroid coordinates (x, y, z) in the three-dimensional coordinate system based on its centroid coordinates (x1, y1). The motors 415 and 423, responsible for driving the gantry movement 4, are activated, causing the moving device 4 to move left, right, forward, and backward along the first guide rail 415 and the second guide rail 426. During the movement, the motors precisely control the speed and distance of the moving device 4 according to the preset program and position information, until the moving device 4 moves above the target object. The robot vision system extracts the outline of the debris and calculates its shortest point. The rotating end 51 begins to rotate, searching for the optimal grabbing position through a 360° rotation. The optimal grabbing position is when the rotating end 51 drives the robotic arm grabbing device 5 to rotate, and the two grippers of the robotic claw 55 are just able to grab the shortest part of the debris. After finding a suitable angle, based on the z-value, joint 53 drives the robotic arm gripping device 5 to bend and extend, causing the robotic claw 55 to slowly approach the target debris. During the approach, the extension rod 56 extends according to the size and position of the debris, expanding the opening range of the robotic claw 55 and constructing an envelope sphere centered on the debris's center coordinates. The robotic claw 55 automatically adjusts its opening range according to the shape and size of the debris, accurately locking onto the target debris. Then, the robotic claw 55 closes via the extension rod 56, automatically adjusting its gripping force according to the material of the debris, firmly grasping it. After grasping, the robotic arm gripping device 5 transfers the debris to a designated location for further processing according to a predetermined program.

[0081] 5. Continuous Grabbing Phase: After process 4 is completed, camera 61 performs process 3 again to obtain the centroid coordinates (x2, y2) of various types of debris after one grab. It then compares these coordinates with the centroid coordinates (x1, y1) recorded in process 3. The deviation (x2-x1, y2-y1) is added to the centroid coordinates (x, y, z) of each type of debris, forming new centroid coordinates for each type of debris. The robotic arm grabbing device then performs the grabbing step of process 4 based on these new centroid coordinates. The effect of this is that by using machine vision technology to take pictures before and after grabbing, the centroid coordinates of the second grabbing are corrected based on the centroid coordinate deviation value, preventing displacement of the debris due to the previous grabbing. If an empty grab occurs, the conveyor device restarts the storage tank for a CT scan, repeating the above process. Small debris and small amounts of residual water are centrally processed and are not within the grabbing range.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sorting method for a kitchen waste sorting device, characterized in that, The device includes: A storage pool for storing drained waste; An internal imaging device is used to scan and determine the spatial location of waste inside the storage pool; A transmission device, wherein the storage pool is provided on the output end of the transmission device, and the transmission device is used to drive the storage pool to reciprocate along a first direction to pass through the internal imaging device; A visual recognition device is used to identify the location of waste on the surface of the storage pool; A mobile device is mounted on the storage pool; A robotic arm gripping device is located at the working end of the mobile device and moves with the mobile device above the storage pool. The robotic arm gripping device is used to grip the garbage in the storage pool layer by layer according to the recognition results of the visual recognition device and the internal imaging device. The sorting method includes the following steps: S1. Pour the drained kitchen waste into the storage tank; S2. Use an internal imaging device to scan the internal debris in the storage tank to obtain voxel information and density information, construct a three-dimensional coordinate system, and analyze the voxel information and density information based on the three-dimensional coordinate system to obtain the spatial centroid coordinates of the target debris. S3. Use a visual recognition device to scan the surface debris in the storage pool, calculate the centroid coordinates of each target debris in the two-dimensional plane, and sort them. S4. Combine the centroid position coordinates to locate the corresponding spatial centroid coordinates, and use the robotic arm gripping device to grab the target debris in order and move it to the designated position; S5. After one grab, return to step S3, compare the centroid coordinates of the remaining target debris before and after, and correct its spatial centroid coordinates accordingly. Then, grab the target debris on the surface of the storage pool layer by layer. S6. If an empty grab occurs, return to step S2. After the target debris in the storage pool is grabbed, process the remaining ungrabbed debris in a centralized manner. Specifically, step S4, which involves sorting and grabbing the target debris, includes: S4.1 Find the debris with the largest surface area and determine if there are multiple debris with similar areas. If so, proceed to step S4.2; otherwise, directly grab the debris with the largest surface area. S4.2 Compare the volumes of miscellaneous objects with similar areas to determine if there are multiple miscellaneous objects with similar volumes. If so, proceed to step S4.3; otherwise, grab the miscellaneous object with the largest volume. S4.3 Compare the densities of miscellaneous objects with similar volumes to determine if there are multiple miscellaneous objects with similar densities. If so, proceed to step S4.4; otherwise, grab the miscellaneous object with the highest density. S4.4 Compare the positions of debris with similar densities and grab the debris closest to the robotic arm.

2. The sorting method of the kitchen waste sorting device according to claim 1, characterized in that, The internal imaging device includes: A ring-shaped drive module, the rotation axis of which is arranged along a first direction, the ring-shaped drive module having a channel formed at its center for passing through the storage pool and the transmission device, the ring-shaped drive module comprising: The rotating support has transmission teeth sequentially formed on its outer circumference; The first drive motor has a worm gear connected to its output end, which meshes with the transmission gear, and is used to drive the rotary support to rotate around its axis. Several radiation sources are installed at intervals on the inner wall surface of the annular drive module; Several detectors are installed at intervals on the inner wall surface of the annular drive module.

3. The sorting method of the kitchen waste sorting device according to claim 1, characterized in that, The mobile device includes: a first transmission module and a second transmission module, wherein the movement direction of the second transmission module is perpendicular to that of the first transmission module; The first transmission module includes: Two first guide rails are respectively set on two opposite sides of the top of the storage pool; Two intermediate plates, each with a first slider at its bottom, are slidably connected to two first guide rails via the first slider. A first rack is disposed on the side of any of the first guide rails; The first support frame is disposed on the middle plate at one end near the first rack; A first motor is mounted on the first support frame, and the output end of the first motor is connected to a first gear that meshes with the first rack. The first baffle is disposed at the end of the first guide rail.

4. The sorting method of the kitchen waste sorting device according to claim 3, characterized in that, The second transmission module includes: Two second guide rails are provided, which are spaced apart along a direction perpendicular to the first guide rail, and the two ends of the second guide rails are respectively provided on the two intermediate plates; A movable plate has second sliders at both ends of its bottom, and the second sliders are slidably connected to the second guide rail. The visual recognition device is installed at the bottom of the movable plate. The second rack is disposed on the side of any of the second guide rails; The second support frame is disposed on the movable plate at one end near the second rack; The second motor is mounted on the second support frame, and the output end of the second motor is connected to a second gear that meshes with the second rack. The second baffle is located at the end of the second guide rail.

5. The sorting method of the kitchen waste sorting device according to claim 4, characterized in that, The robotic arm gripping device includes a rotating end, a large arm, a joint, a forearm, and a robotic claw connected in sequence, with the rotating end installed at the bottom of the moving plate; The visual recognition device is installed at the end of the rotating end away from the moving plate. The visual recognition device includes a visual imaging module, a FIFO frame buffer, and a light source installed on the side.

6. The sorting method according to claim 1, characterized in that, Step S2 includes: The internal imaging device is used to scan the debris inside the storage pool. The voxel information is formed by dividing the scan data into several voxel units. A preset threshold variable is used to compare the size of the voxel unit with the set threshold variable, and the target voxel unit is selected. The point cloud data of the target clutter is determined based on the target voxel unit and density information. Establish a three-dimensional coordinate system and construct the spatial centroid coordinates of all target debris.

7. The sorting method according to claim 6, characterized in that, The spatial centroid coordinates for constructing all target debris include: The spatial centroid coordinates of the target debris are determined using the centroid calculation formula, whereby: ； Where N represents the number of voxels contained in the target substance. The three-dimensional coordinates of each voxel.

8. The sorting method according to claim 1, characterized in that, Step S3 includes: performing image preprocessing, region of interest segmentation, feature extraction and classification, and object area calculation on the image captured by the visual recognition device; establishing a two-dimensional coordinate system for the captured image; calculating and recording the centroid coordinates of each object in the two-dimensional plane; and sorting the objects by area.

Images