Sorting robot capable of automatically dumping garbage and garbage classification treatment method

Abstract

The invention relates to the technical field of garbage classification, and discloses a sorting robot capable of automatically dumping garbage and a garbage classification processing method.The sorting robot capable of automatically dumping garbage comprises an automatic garbage dumping mechanism, a moving mechanism, a sorting manipulator, a tracking module, a camera and a controller; the garbage can is installed on the dumping structure, the tracking module and the camera are both installed at the front end of the moving mechanism, the sorting mechanical arm is installed behind the tracking module, the can base is installed on the periphery of the extension mechanical arm, and the sorting mechanical arm is installed on the can base. The dumping structure, the garbage can, the moving mechanism, the sorting manipulator, the tracking module and the camera are all connected with a controller, and the sorting manipulator is matched with the garbage can. The garbage can can automatically walk along the track line, garbage is recognized and classified, then the garbage is put into the corresponding garbage cans, the garbage can be automatically dumped, and the labor cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of garbage classification, and in particular to a sorting robot for automatically dumping garbage and a garbage classification processing method. Background Art

[0002] iBin, an unmanned interactive media storage robot developed by Guangzhou Aike Robotics Co., Ltd., is a robot that integrates laser sensors, real-time positioning and map construction, allowing the robot to navigate unimpeded through obstacles. It integrates four major systems: intelligent navigation, safety protection, digital media, and the Internet of Things. It provides a more environmentally friendly garbage collection service, instant mapping, real-time positioning, autonomous navigation, and multiple obstacle avoidance. It has the following advantages: 1. Autonomous driving; 2. The garbage bin has a capacity of 80L, and the double-lift bucket design is convenient for clearing garbage; 3. It supports the perception of various gases such as domestic smoke, domestic odors, and volatile gases of organic compounds; 4. It has smoke, infrared, and human body sensors; 5. 27-inch dual-screen advertising streaming media, which can carry out garbage classification knowledge propaganda while cleaning.

[0003] BARYL is a mobile smart trash can developed in France. The BARYL mobile smart trash can is equipped with sensors. When an object approaches, the trash can will automatically open, making it convenient for users to put out trash, and will automatically close after a certain period of time. BARYL can recognize the user's movements through cameras and artificial intelligence technology. After the user waves to it, it will move towards the user and accept the trash discarded by the passenger. BARYL can also sense the garbage filling status in the trash can through sensors, and remind users when to empty the trash can through mobile phone software. BARYL can also be remotely controlled. Users can remotely control BARYL through mobile phone software, such as opening and closing the trash can, checking the filling status of the trash can, etc.

[0004] The above two robots have the following disadvantages: (1) Both are mobile smart trash cans and do not have the function of automatically collecting trash. (2) Both do not have the ability to autonomously sort trash. (3) Both require manual labor to carry out the subsequent trash dumping work. Summary of the invention

[0005] The purpose of the present invention is to overcome the deficiencies of the above prior art and to provide a sorting robot for automatically dumping garbage and a garbage sorting method.

[0006] The object of the present invention is achieved through the following technical scheme: a sorting robot for automatic garbage dumping comprises an automatic garbage dumping mechanism, a moving mechanism, a sorting manipulator, a tracking module, a camera and a controller, the automatic garbage dumping mechanism comprises a dumping structure and a garbage can, the dumping structure comprises a bucket base, a rotating shaft, a roll plate, a first connecting rod, a second connecting rod, a dumping servo and a servo base, the garbage can is installed on the bucket base through the rotating shaft, the lower end of the roll plate is hinged to the bucket base, the upper end of the roll plate is sequentially connected to the first connecting rod, the second connecting rod and the dumping servo, the dumping servo is installed on the bucket base through the servo base, the tracking module and the camera are both installed at the front end of the moving mechanism, the sorting manipulator is installed behind the tracking module, the bucket base is installed around the extension manipulator, the dumping servo, the garbage can, the moving mechanism, the sorting manipulator, the tracking module, and the camera are all connected to the controller, and the sorting manipulator matches the garbage can.

[0007] A better choice, the trash can includes a lid, a transmission shaft, a barrel body, a flip servo and a tipping shaft seat, the lid is fixedly connected to the transmission shaft, the transmission shaft is rotatably mounted on the top of the barrel body, the flip servo is mounted on the barrel body, the flip servo is connected to the transmission shaft, the bottom of the barrel body is connected to the tipping shaft seat, the tipping shaft seat is rotatably connected to the shaft, a garbage dumping port is provided on one side of the barrel body, the garbage dumping port corresponds to the side tilt plate, and the flip servo is connected to the controller.

[0008] A more preferred option is that the barrel body is provided with an infrared sensor, and the infrared sensor is connected to the controller.

[0009] A better option is that the sorting robot includes a clamping structure, a swing arm and a rotating platform connected in sequence, the rotating platform is installed on the moving mechanism, and the clamping structure, the swing arm and the rotating platform are all connected to a controller.

[0010] A better choice is that the clamping structure includes two clamping jaws, two third connecting rods, a clamping base, two gear rockers, a clamping servo, a connecting piece, a rotating servo and a rotating mounting seat, the two clamping jaws are respectively hinged to the two third connecting rods and the gear rockers, the two third connecting rods are hinged to the clamping base, one gear rocker is rotatably mounted on the clamping base, the other gear rocker is connected to the clamping servo, the clamping servo is mounted on the clamping base, the two gear rockers are meshed, the clamping base is connected to the rotating servo through a connecting piece, the rotating servo is mounted on the rotating mounting seat, the rotating mounting seat is connected to the swing arm, and the clamping servo and the rotating servo are both connected to the controller.

[0011] A garbage sorting and processing method for a sorting robot for automatically dumping garbage comprises the following steps:

[0012] S1. Load the optimized MobileNetV2 training model into the sorting robot;

[0013] S2, the sorting robot drives according to the tracking logic judgment method, and the camera takes pictures of the garbage to obtain image information;

[0014] S3, preprocessing the image information of step S2 to obtain image data;

[0015] S4. According to the image data of step S3, the optimized MobileNetV2 training model identifies the image data to determine whether there is garbage. If yes, the garbage type is output. If not, return to step S2;

[0016] S5. According to the type of garbage in step S4, the sorting robot stops moving, the sorting manipulator grabs the garbage, and the corresponding garbage can automatically opens the lid. At the same time, the sorting manipulator transfers the garbage to the corresponding garbage can, and the sorting manipulator resets and closes the lid;

[0017] S6. Repeat steps S1-S5 until the trash can is detected to be full, the sorting robot moves to the designated position, and the automatic trash dumping mechanism dumps the trash.

[0018] A better choice, the learning process of the optimized MobileNetV2 training model in step S1 includes the following steps:

[0019] S11, collecting information about junk images and related labels, preprocessing the junk images, and constructing a data set, which is divided into a training set, a validation set, and a test set;

[0020] S12, load the trained MobileNetV2 network model, and then establish a trainable layer on the trained MobileNetV2 network model;

[0021] S13, using the training set of step S11 to train the trainable layer of the MobileNetV2 network model of step S12 to obtain a MobileNetV2 training model;

[0022] S14, using the validation set of step S11 to evaluate the MobileNetV2 training model of step S13, and adjusting the parameters of the MobileNetV2 training model to obtain an optimized MobileNetV2 training model;

[0023] S15. Use the test set to evaluate the accuracy, precision, and recall of the optimized MobileNetV2 training model.

[0024] A better option is to pre-process the junk images in step S11, including image flipping, brightness adjustment, image denoising, scale normalization and data enhancement.

[0025] Preferably, the trainable layer of step S12 includes a fully connected layer or a convolutional layer.

[0026] A better option is that the tracking logic judgment method of step S2 includes the following steps:

[0027] S21, the tracking module includes a first tracking module, a second tracking module, a third tracking module and a fourth tracking module arranged in sequence, and the first tracking module, the second tracking module, the third tracking module and the fourth tracking module detect the track line simultaneously;

[0028] S22, determining whether only the second tracking module and the third tracking module detect the track line, if yes, the sorting robot drives in a straight line, if not, executing step S23;

[0029] S23, if the first tracking module and / or the second tracking module detects a track line and the fourth tracking module does not detect a track line, the sorting robot turns toward the installation position of the first tracking module, otherwise, execute step S24;

[0030] S24: If the third tracking module and / or the fourth tracking module detects a track line and the first tracking module does not detect a track line, the sorting robot turns toward the installation position of the fourth tracking module, otherwise the sorting robot stops moving.

[0031] The present invention has the following advantages and beneficial effects compared with the prior art:

[0032] The present invention provides a sorting robot for automatically dumping garbage and a garbage classification and processing method, which can automatically walk along a trajectory, identify and classify garbage, and then put the garbage into corresponding garbage bins. It can also automatically dump the garbage, thereby reducing labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0034] Figure 2 It is a schematic diagram of an automatic garbage dumping mechanism of a sorting robot for automatic garbage dumping and a garbage classification and processing method of the present invention;

[0035] Figure 3It is a schematic diagram of the dumping structure of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0036] Figure 4 It is a schematic diagram of a garbage bin of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0037] Figure 5 It is a schematic diagram of a garbage bin of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0038] Figure 6 It is a schematic diagram of a dumping rotating shaft seat of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0039] Figure 7 It is a schematic diagram of a dumping rotating shaft seat of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0040] Figure 8 It is a schematic diagram of a mobile mechanism of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0041] Fig. 9 It is a schematic diagram of a sorting robot for automatically dumping garbage and a sorting manipulator of a garbage classification and processing method of the present invention;

[0042] Fig.10 It is a schematic diagram of a rotating platform of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0043] Fig.11 It is a cross-sectional view of a rotating platform of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0044] Fig.12 It is a schematic diagram of a swing arm of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0045] Fig.13 It is a schematic diagram of the clamping structure of a sorting robot for automatically dumping garbage and a garbage classification processing method of the present invention;

[0046] Fig.14 It is a schematic diagram of the clamping structure of a sorting robot for automatically dumping garbage and a garbage classification processing method of the present invention;

[0047] Fig.15 It is a flow chart of a sorting robot for automatically dumping garbage and a garbage classification and processing method of the present invention;

[0048] Fig.16It is a learning process diagram of an optimized MobileNetV2 training model of an automatic garbage dumping sorting robot and a garbage classification processing method of the present invention;

[0049] Fig.17 It is a schematic diagram of a tracking logic judgment method of a sorting robot for automatically dumping garbage and a garbage classification processing method of the present invention;

[0050] The markings of the components in the attached drawings are as follows: 1-automatic dumping mechanism; 11-dumping structure; 111-barrel base; 112-rotating shaft; 113-rolling plate; 114-first connecting rod; 115-second connecting rod; 116-dumping servo; 117-servo base; 12-trash can; 121-barrel cover; 122-transmission shaft; 123-barrel body; 123a-dumping port; 124-flip servo; 125-dumping shaft seat; 125a-bottom plate; 125b-arc dumping block; 125c-limiting frame; 125d-shaft hole; 125e-barrel slot; 2-moving mechanism; 201-frame; 202-driving motor; 203-wheel; 204-first tracking module; 205-second tracking module; 206-third tracking module; 207 - the fourth tracking module; 3- sorting robot; 31- clamping structure; 311- clamping claw; 312- third connecting rod; 313- clamping base; 314- gear rocker; 315- clamping servo; 316- connecting piece; 317- rotating servo; 318- rotating mounting seat; 32- swing arm; 321- first servo mounting seat; 322- first swing servo; 323- first swing frame; 324- second servo mounting seat; 325- second swing servo; 326- second swing frame; 327- third swing frame; 328- third swing servo; 329- third servo mounting frame; 33- rotating platform; 331- supporting column; 332- servo fixing plate; 333- rotating base; 334- rotating seat; 335- rotating servo; 4- camera. DETAILED DESCRIPTION

[0051] The purpose of the present invention is further described in detail below with reference to the accompanying drawings and specific examples. The examples cannot be described one by one here, but the implementation methods of the present invention are not therefore limited to the following examples.

[0052] like Figure 1As shown, an automatic garbage dumping sorting robot comprises four automatic garbage dumping mechanisms 1, a mobile mechanism 2, a sorting manipulator 3, four tracking modules, a camera 4 and a controller. The four tracking modules are sequentially installed at the front end of the mobile mechanism 2 from left to right, the sorting manipulator 3 is installed at the front part of the mobile mechanism 2, and the camera 4 is installed on the right side of the sorting manipulator 3. The four automatic garbage dumping mechanisms 1 are respectively installed at the rear end of the mobile mechanism 2 and on both sides of the mobile mechanism 2. The controller is respectively connected to the automatic garbage dumping mechanism 1, the mobile mechanism 2, the sorting manipulator 3, the four tracking modules, the camera 4 and the controller.

[0053] The automatic garbage dumping mechanism 1 is used to store different types of garbage and automatically dump the garbage. The moving mechanism 2 is used to enable the sorting robot to move and turn. The sorting robot 3 is used to carry the garbage to each automatic garbage dumping mechanism 1. The tracking module is used to detect the trajectory line and feedback the information to the controller for judgment. The camera 4 is used to take pictures and feedback to the controller to determine whether there is garbage. The controller is a single-chip microcomputer, which is used to control various mechanisms and sensors to achieve the coordination of garbage sorting and dumping.

[0054] like Figure 2-5 As shown, the automatic garbage dumping mechanism 1 includes a dumping structure 11 and a garbage can 12, wherein the dumping structure 11 includes a base 111, a rotating shaft 112, a tilting plate 113, two first connecting rods 114, two second connecting rods 115, two dumping steering gears 116 and a steering gear base 117, and the garbage can 12 includes a cover 121, a transmission shaft 122, a barrel 123, a flip cover steering gear 124 and a dumping rotating shaft seat 125. The base 111 is installed on the side of the moving mechanism 2, the bottom of the barrel 123 is connected to the dumping rotating shaft seat 125, the rotating shaft 112 is symmetrically installed at the bottom of the base 111, and the two rotating shafts 112 are rotatably inserted into the dumping rotating shaft seat 125. The transmission shaft 122 is rotatably installed on the top of the barrel 123, and the cover 121 is fixedly connected to the transmission shaft 122. The flip servo 124 is installed on the top side of the barrel body 123, and the flip servo 124 is connected to the transmission shaft 122. An infrared sensor (not shown in the figure) is provided on the barrel body 123. A garbage dumping port is provided on the side of the barrel body 123, and the garbage dumping port corresponds to the roll plate 113, and the bottom end of the roll plate 113 is hinged to the barrel base 111. The upper end of the first connecting rod 114 is hinged on both sides of the middle position of the roll plate 113, and the lower end of the first connecting rod 114 is hinged to one end of the second connecting rod 115, and the other end of the second connecting rod 115 is fixedly connected to the dumping servo 116, and the dumping servo 116 is installed on both sides of the barrel base 111 through the servo base 117. The infrared sensor, the flip servo 124 and the two dumping servos 116 are all connected to the controller.

[0055] like Figure 6-7As shown, the tipping shaft seat 125 includes a bottom plate 125a, an arc tipping block 125b and a limit frame 125c. The arc tipping block 125b is connected to the bottom of the bottom plate 125a. The arc tipping block 125b is provided with an axis hole 125d, and the two rotating shafts 112 are rotatably embedded in the axis hole 125d. The limit frame 125c is connected to the edge of the top of the bottom plate 125a to form a bucket groove 125e, and the bucket groove 125e is used to position and fix the bucket body 123. The bottom plate 125a is used to support the bucket body 123, and the arc tipping block 125b can fix the tipping direction of the garbage can 12; the limit frame 125c is used to prevent the bucket body 123 from detaching from the bottom plate.

[0056] like Figure 8 As shown, the mobile mechanism 2 includes four wheels 203, four drive motors 202, a frame 201, and four tracking modules, and the four tracking modules are respectively a first tracking module 204, a second tracking module 205, a third tracking module 206, and a fourth tracking module 207. The four wheels 203 are respectively installed on both sides of the frame 201 through four drive motors 202, and the first tracking module 204, the second tracking module 205, the third tracking module 206, and the fourth tracking module 207 are all installed at the front end of the frame 201. The first tracking module 204, the second tracking module 205, the third tracking module 206, and the fourth tracking module 207 are arranged linearly from left to right. Four automatic garbage dumping mechanisms 1 are installed at the rear end and both sides of the frame 201, the sorting manipulator 3 is installed at the front end of the frame 201, and the camera 4 is installed on the side of the sorting manipulator 3. The first tracking module 204 , the second tracking module 205 , the third tracking module 206 , the fourth tracking module 207 and the four driving motors 202 are all controlled by the controller.

[0057] The wheel 203 is a Mecanum wheel, which generates friction with the ground to realize the steering and movement of the sorting robot; the drive motor 202 converts electrical energy into mechanical energy to provide power for the movement of the sorting robot; the frame 201 acts as a bracket; the first tracking module 204, the second tracking module 205, the third tracking module 206 and the fourth tracking module 207 are all infrared transmitters and infrared receivers, which are used to detect the trajectory line and feed back to the controller. When working in a normal state, the transmitter emits infrared rays and the receiver receives reflected light. If there is an object within its detection range and the object can reflect infrared rays of sufficient intensity, the infrared receiver will receive the reflected light and turn on. At this time, TTL is a low-level output and the indicator diode is on. If there is no object within its detection range or the infrared rays are absorbed by black, the infrared receiver cannot receive the reflected light, so it remains in the off state. At this time, TTL is a high-level output and the indicator diode is off.

[0058] According to the truth table of the motor driver chip TB6612, by setting the level of the controller IO port, that is, controlling the level of the four pins of the single-chip microcomputer, and thus controlling the level of the four pins of the motor driver chip (which can be purchased in the existing market), the drive motor 202 can be made to complete the forward, reverse, brake and stop operations. Since these operations can only allow the sorting robot to travel in a straight line, it is decided to use a timer to generate a PWM signal to control the speed of the drive motor 202, and to enable the sorting robot to have the ability to turn through the differential speed. At this time, the sorting robot can travel on a preset suitable route according to the tracking logic judgment method. According to the tracking logic judgment method, the controller needs to judge whether the four tracking modules have detected the track line. According to the four tracking modules at the front end of the sorting robot detect the track line, after the judgment of the controller, four situations may occur. The controller needs to control the rotation of the drive motor 202 according to these four situations, thereby controlling the driving direction of the sorting robot. The tracking logic judgment method is as follows Fig.16 .exist Fig.16 In the above equation, 1 means the track line is detected, and 0 means no track line is detected.

[0059] like Fig. 9 As shown, the sorting robot 3 includes a clamping structure 31, a swing arm 32 and a rotating platform 33 connected in sequence. The rotating platform 33 is installed on the frame 201 of the mobile mechanism 2. The clamping structure 31, the swing arm 32 and the rotating platform 33 are all connected to the controller. The clamping structure 31 is used to grab garbage, and the swing arm 32 is used to adjust the position of the clamping structure 31 so that the clamping structure 31 can reach a suitable position. The rotating platform 33 is used to adjust the angle of the swing arm 32.

[0060] like Fig.10 and 11 As shown, the rotating platform 33 includes four supporting columns 331, a steering gear fixing plate 332, a rotating base 333, a rotating seat 334 and a rotating steering gear 335. The bottom of the steering gear fixing plate 332 is mounted on the frame 201 through the four supporting columns 331. The rotating steering gear 335 is mounted on the bottom of the steering gear fixing plate 332, and the rotating steering gear 335 passes through the steering gear fixing plate 332 and is connected to the rotating seat 334, and the rotating seat 334 is rotatably mounted on the rotating base 333. The rotating base 333 is mounted on the top of the steering gear fixing plate 332. The rotating seat 334 is connected to the lower end of the swing arm 32. The rotating steering gear 335 is controlled by the controller.

[0061] The support column 331 is used to support the steering gear fixing plate 332 to form a certain height to facilitate the installation of the rotating steering gear 335. The steering gear fixing plate 332 is used to fix and install the rotating steering gear 335. The rotating base 333 is used to install the rotating seat 334 to reduce the friction with the rotating seat 334. The rotating seat 334 plays a transmission role and drives the swing arm 32 to rotate. The rotating steering gear 335 provides power for the rotation of the swing arm 32.

[0062] like Fig.12 As shown, the swing arm 32 includes a first steering gear mounting seat 321, a first swing steering gear 322, a first swing frame 323, a second steering gear mounting seat 324, a second swing steering gear 325, a second swing frame 326, a third swing frame 327, a third swing steering gear 328 and a third steering gear mounting frame 329. The first steering gear mounting seat 321 is connected to the rotating mounting seat 318 of the clamping structure 31. The first steering gear mounting seat 321 is hinged to the first swing frame 323, the first swing steering gear 322 is mounted on the first steering gear mounting seat 321, and the rudder of the first swing steering gear 322 is connected to the first swing frame 323. The first swing frame 323 is fixedly connected to the second steering gear mounting seat 324. The second swing steering gear 325 is mounted on the second steering gear mounting seat 324. The second steering gear mounting seat 324 is hinged to the second swing frame 326. The rudder of the second swing steering gear 325 is connected to the second swing frame 326. The second swing frame 326 is fixedly connected to the third swing frame 327 , the rudder of the third swing servo 328 is connected to the third swing frame 327 , and the third swing servo 328 is installed on the rotating seat 334 of the rotating platform 33 through the third servo mounting frame 329 .

[0063] The first servo mounting seat 321 is used to mount the first swing servo 322. The first swing servo 322 is used to drive the first swing frame 323 to rotate relative to the first servo mounting seat 321. The first swing frame 323 plays a transmission role. The second servo mounting seat 324 is used to mount the second swing servo 325. The second swing servo 325 is used to drive the second swing frame 326 to rotate relative to the second servo mounting seat 324. The second swing frame 326 and the third swing frame 327 play a transmission role. The third swing servo 328 is used to drive the third swing frame 327 to rotate relative to the third servo mounting frame 329. The third servo mounting frame 329 is used to mount the third swing servo 328.

[0064] like Fig.13 and 14As shown, the clamping structure 31 includes two clamping jaws 311, two third connecting rods 312, a clamping base 313, two gear swing rods 314, a clamping servo 315, a connecting piece 316, a rotating servo 317 and a rotating mounting seat 318. The middle parts of the two clamping jaws 311 are respectively hinged to one end of the two third connecting rods 312, and the rear ends of the two clamping jaws 311 are respectively hinged to one end of the two gear swing rods 314. The other ends of the two third connecting rods 312 are hinged to the front end of the clamping base 313, and the other end of one gear swing rod 314 is hinged to the left side of the clamping base 313. The other end of the other gear swing rod 314 is fixedly connected to the rudder of the rotating servo 317, and the two gear swing rods 314 are meshed. The clamping servo 315 is installed on the clamping base 313. The rear end of the clamping base 313 is connected to the connecting piece 316 by bolts. The rudder bar of the rotating steering gear 317 passes through the clamping base 313 and is then connected to the connecting member 316, and the rotating steering gear 317 is installed on the rotating mounting seat 318. The clamping steering gear 315 and the rotating steering gear 317 are both controlled by the controller.

[0065] The clamping jaw 311 is used to clamp garbage; the third connecting rod 312 is used to connect the clamping base 313 and the clamping jaw 311; the clamping base 313 is used to install the gear rocker 314 and the clamping servo 315; the gear rocker 314 plays a transmission role, used to drive the clamping jaw 311 to clamp or open; the clamping servo 315 provides power for the clamping and opening of the clamping jaw 311; the connecting piece 316 is used to install the clamping base 313 on the rotating servo 317; the rotating servo 317 provides power for adjusting the rotation angle of the clamping jaw 311; the rotating mounting seat 318 is used to install and fix the rotating servo 317.

[0066] In this embodiment, a transfer learning method based on the MobileNetV2 model is adopted. By sharing the core parameters of the model such as weights and biases, the learning migration from the existing data set to the garbage data set is realized. The organic integration of deep learning and transfer learning significantly reduces the training time of the model, accelerates the convergence of the model, and thus greatly improves the training efficiency. The key to transfer learning is to accurately apply the knowledge of the source field to the target field to enhance the learning effect of the target field. The specific operation is: first load the trained MobileNetV2 network model, and then add appropriate trainable layers, such as fully connected layers or convolutional layers, according to the characteristics of the new problem. Next, these layers are specially trained using the new data set to meet the needs of the new problem. In this process, fine-tuning technology is used, that is, the parameters of the original layer are kept unchanged, and only the newly added trainable layer is fine-tuned. Since MobileNetV2 has accumulated rich experience in image features, this makes the feature extraction in the new task more accurate, thereby improving the accuracy and generalization performance of the model. In addition, with the optimized MobileNetV2 training model, not only does it save the time and computing resources required to design and train the model from scratch, but it also ensures that the existing parameters can be fully utilized without losing the existing learning characteristics. Fig.15 As shown, a garbage sorting method for a sorting robot for automatically dumping garbage includes the following steps:

[0067] S1. Load the optimized MobileNetV2 training model into the sorting robot; Fig.16 As shown, the learning process of the optimized MobileNetV2 training model in step S1 includes the following steps:

[0068] S11, collecting information about a large number of junk images and related labels, preprocessing the junk images, and constructing a data set, where the data set is divided into a training set, a validation set, and a test set; the junk image preprocessing in step S11 includes image flipping, brightness adjustment, image denoising, scale normalization, and data enhancement.

[0069] S12. Load the trained MobileNetV2 network model, and then establish a trainable layer (including a fully connected layer or a convolutional layer) on the trained MobileNetV2 network model;

[0070] S13, using the training set of step S11 to train the trainable layer of the MobileNetV2 network model of step S12 to obtain a MobileNetV2 training model;

[0071] S14, using the validation set of step S11 to evaluate the MobileNetV2 training model of step S13, and adjusting the parameters (weight, bias) of the MobileNetV2 training model to obtain an optimized MobileNetV2 training model;

[0072] S15. Use the test set to evaluate the accuracy, precision, and recall of the optimized MobileNetV2 training model.

[0073] S2, the sorting robot drives according to the tracking logic judgment method, and the camera 4 takes pictures of the garbage to obtain image information; the tracking logic judgment method of step S2 includes the following steps:

[0074] S21, the tracking module includes a first tracking module 204, a second tracking module 205, a third tracking module 206 and a fourth tracking module 207 arranged in sequence, and the first tracking module 204, the second tracking module 205, the third tracking module 206 and the fourth tracking module 207 detect the track line simultaneously;

[0075] S22, determining whether only the second tracking module 205 and the third tracking module 206 detect the track line, if yes, the sorting robot drives in a straight line, if not, executing step S23;

[0076] S23, if the first tracking module 204 and / or the second tracking module 205 detects a track line and the fourth tracking module 207 does not detect a track line, the sorting robot turns toward the installation position of the first tracking module 204, otherwise, execute step S24;

[0077] S24: If the third tracking module 206 and / or the fourth tracking module 207 detects a track line and the first tracking module does not detect a track line, the sorting robot turns toward the installation position of the fourth tracking module 207, otherwise the sorting robot stops moving.

[0078] S3, preprocessing the image information of step S2, the preprocessing including image flipping, brightness adjustment, image denoising, scale normalization and data enhancement to obtain image data;

[0079] S4. According to the image data of step S3, the optimized MobileNetV2 training model identifies the image data to determine whether there is garbage. If yes, the garbage type is output. If not, return to step S2;

[0080] S5. According to the type of garbage in step S4, the sorting robot 3 grabs the garbage, and the corresponding garbage can 12 automatically opens the lid 121. The sorting robot 3 transfers the garbage to the corresponding garbage can 12, and the sorting robot 3 resets and closes the lid 121.

[0081] S6. Repeat steps S1-S5 until it is detected that the trash can 12 is full, and the sorting robot moves to a designated position and the garbage dumping mechanism 1 automatically dumps the garbage.

[0082] The above specific implementation modes are preferred embodiments of the present invention and cannot be used to limit the present invention. Any other changes or other equivalent replacement methods that do not deviate from the technical solution of the present invention are included in the protection scope of the present invention.

Claims

1. A sorting robot for automatically dumping garbage, characterized in that: It includes an automatic garbage dumping mechanism, a moving mechanism, a sorting manipulator, a tracking module, a camera and a controller. The automatic garbage dumping mechanism includes a dumping structure and a garbage can. The dumping structure includes a bucket base, a rotating shaft, a roll plate, a first connecting rod, a second connecting rod, a dumping servo and a servo base. The garbage can is installed on the bucket base through the rotating shaft. The lower end of the roll plate is hinged to the bucket base. The upper end of the roll plate is connected to the first connecting rod, the second connecting rod and the dumping servo in sequence. The dumping servo is installed on the bucket base through the servo base. The tracking module and the camera are both installed at the front end of the moving mechanism. The sorting manipulator is installed behind the tracking module. The bucket base is installed around the extension manipulator. The dumping servo, the garbage can, the moving mechanism, the sorting manipulator, the tracking module and the camera are all connected to the controller. The sorting manipulator matches the garbage can.

2. The automatic garbage dumping sorting robot according to claim 1, characterized in that: The trash can includes a bucket cover, a transmission shaft, a bucket body, a flip-lid servo and a tipping shaft seat, the bucket cover is fixedly connected to the transmission shaft, the transmission shaft is rotatably installed on the top of the bucket body, the flip-lid servo is installed on the bucket body, the flip-lid servo is connected to the transmission shaft, the bottom of the bucket body is connected to the tipping shaft seat, the tipping shaft seat is rotatably connected to the shaft, a garbage dumping port is provided on one side of the bucket body, the garbage dumping port corresponds to the side tilting plate, and the flip-lid servo is connected to the controller.

3. The automatic garbage dumping sorting robot according to claim 2, characterized in that: The barrel body is provided with an infrared sensor, and the infrared sensor is connected to the controller.

4. The automatic garbage dumping sorting robot according to claim 1, characterized in that: The sorting robot comprises a clamping structure, a swing arm and a rotating platform which are connected in sequence. The rotating platform is installed on the moving mechanism. The clamping structure, the swing arm and the rotating platform are all connected to a controller.

5. The automatic garbage dumping sorting robot according to claim 1, characterized in that: The clamping structure includes two clamping jaws, two third connecting rods, a clamping base, two gear rockers, a clamping servo, a connecting piece, a rotating servo and a rotating mounting seat. The two clamping jaws are respectively hinged to the two third connecting rods and the gear rockers, the two third connecting rods are hinged to the clamping base, one gear rocker is rotatably mounted on the clamping base, the other gear rocker is connected to the clamping servo, the clamping servo is mounted on the clamping base, the two gear rockers are meshed, the clamping base is connected to the rotating servo through a connecting piece, the rotating servo is mounted on the rotating mounting seat, the rotating mounting seat is connected to the swing arm, and the clamping servo and the rotating servo are both connected to the controller.

6. A garbage sorting method for a sorting robot that automatically dumps garbage, characterized in that: The following steps are involved: S1. Load the optimized MobileNetV2 training model into the sorting robot; S2, the sorting robot drives according to the tracking logic judgment method, and the camera takes pictures of the garbage to obtain image information; S3, preprocessing the image information of step S2 to obtain image data; S4. According to the image data of step S3, the optimized MobileNetV2 training model identifies the image data to determine whether there is garbage. If yes, the garbage type is output. If not, return to step S2; S5. According to the type of garbage in step S4, the sorting robot stops moving, the sorting manipulator grabs the garbage, and the corresponding garbage can automatically opens the lid. At the same time, the sorting manipulator transfers the garbage to the corresponding garbage can, and the sorting manipulator resets and closes the lid; S6. Repeat steps S1-S5 until the trash can is detected to be full, the sorting robot moves to the designated position, and the automatic trash dumping mechanism dumps the trash.

7. The method for sorting garbage using a sorting robot for automatically dumping garbage according to claim 6, characterized in that: The learning process of the optimized MobileNetV2 training model in step S1 includes the following steps: S11, collecting information about junk images and related labels, preprocessing the junk images, and constructing a data set, which is divided into a training set, a validation set, and a test set; S12, load the trained MobileNetV2 network model, and then establish a trainable layer on the trained MobileNetV2 network model; S13, using the training set of step S11 to train the trainable layer of the MobileNetV2 network model of step S12 to obtain a MobileNetV2 training model; S14, using the validation set of step S11 to evaluate the MobileNetV2 training model of step S13, and adjusting the parameters of the MobileNetV2 training model to obtain an optimized MobileNetV2 training model; S15. Use the test set to evaluate the accuracy, precision, and recall of the optimized MobileNetV2 training model.

8. The method for sorting garbage using a sorting robot for automatically dumping garbage according to claim 7, characterized in that: The junk image preprocessing in step S11 includes image flipping, brightness adjustment, image denoising, scale normalization and data enhancement.

9. The method for sorting garbage using a sorting robot for automatically dumping garbage according to claim 7, characterized in that: The trainable layer of step S12 includes a fully connected layer or a convolutional layer.

10. The method for sorting garbage using a sorting robot for automatically dumping garbage according to claim 6, characterized in that: The tracking logic determination method of step S2 includes the following steps: S21, the tracking module includes a first tracking module, a second tracking module, a third tracking module and a fourth tracking module arranged in sequence, and the first tracking module, the second tracking module, the third tracking module and the fourth tracking module detect the track line simultaneously; S22, determining whether only the second tracking module and the third tracking module detect the track line, if yes, the sorting robot drives in a straight line, if not, executing step S23; S23, if the first tracking module and / or the second tracking module detects a track line and the fourth tracking module does not detect a track line, the sorting robot turns toward the installation position of the first tracking module, otherwise, execute step S24; S24: If the third tracking module and / or the fourth tracking module detects a track line and the first tracking module does not detect a track line, the sorting robot turns toward the installation position of the fourth tracking module, otherwise the sorting robot stops moving.

Images