Spraying track control method and device for cleaning equipment

Abstract

The invention relates to the technical field of cleaning equipment, and discloses a spraying track control method and device for cleaning equipment, and the method comprises the following steps: 1, collecting the motion data of a plurality of spraying arms, and collecting the curved surface data of a target cleaning object; 2, planning an optimal motion track according to the collected curved surface data; step 3, generating a target motion angle set of a plurality of spraying arms based on the optimal motion track; and step 4, generating a control instruction for controlling a motor based on the target motion angle set. Through fine control over the cleaning track, the surface of tableware can be effectively covered, the cleaning efficiency is improved, and water resources are saved; motion planning is carried out according to tableware of different shapes to adapt to various cleaning scenes; the cleaning effect of the dish-washing machine and the user experience are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cleaning equipment, and in particular to a spray trajectory control method and device for cleaning equipment. Background Art

[0002] In cleaning equipment, the spray device is the primary cleaning component. The rotating axis of the rotating spray arm is driven by a motion system or power unit, spraying cleaning water to achieve cleaning. In existing technologies, the spray arm uses water pressure to cause the rotating spray arm to rotate randomly. The nozzle uses a linear water jet to spray water onto the surface of the object to be cleaned. Over time, the water is fully covered and the cleaning operation is completed.

[0003] However, the existing technology has the following problems:

[0004] Most of the spray devices provided in the prior art are hydraulically driven. The hydraulically driven spray devices may have the problem of low motion control accuracy, which results in a relatively fixed motion trajectory of the water flow. As a result, the water flow has poor cleaning effect on local corners or shielded areas during cleaning, affecting the cleaning efficiency. At the same time, there is also the problem of large water consumption. Summary of the Invention

[0005] The purpose of the present invention is to provide a spray trajectory control method and device for cleaning equipment in order to solve the above problems and overcome the defects of the prior art. Please refer to the following for details.

[0006] To achieve the above objectives, the present invention provides the following technical solutions:

[0007] The present invention provides a method for controlling the spray trajectory of a cleaning device, comprising the following steps:

[0008] Step 1: Collect motion data of multiple spray arms and surface data of the target object to be cleaned;

[0009] Step 2: Plan the optimal motion trajectory based on the collected surface data;

[0010] Step 3: generating a target motion angle set of multiple spray arms based on the optimal motion trajectory;

[0011] Step 4: generating a control instruction for controlling the motor based on the target motion angle set;

[0012] Multiple spray arms are respectively controlled by multiple motion motors, and motion detection sensors are respectively provided on the multiple motion motors. The multiple motion motors are all controlled by a main controller. The control instructions are transmitted to the main controller, and the main controller drives the multiple spray arms to move at corresponding target motion angles through the multiple motion motors.

[0013] Preferably, the step 1 comprises the following steps:

[0014] Step 1-1: Get the current position, current speed and current acceleration of each spray arm;

[0015] Step 1-2: When the target cleaning object is a regular curved surface, a surface model pre-stored in a surface model database is retrieved and the surface model is used as the surface data of the target cleaning surface;

[0016] Step 1-3: When the target cleaning object is an irregular curved surface, a surface model of the target cleaning object is obtained.

[0017] Preferably, steps 1-3 include the following steps:

[0018] Step 1-3-1: Acquire a two-dimensional image of the target cleaning object;

[0019] Step 1-3-2: Input the two-dimensional image into a pre-trained depth estimation model to obtain a depth map or depth class output by the depth estimation model;

[0020] Step 1-3-3: Based on the depth map or depth class, obtain a point cloud distribution map of the target cleaning object, and use the point cloud distribution map as the surface data of the target cleaning surface.

[0021] Preferably, the depth estimation model is obtained by performing deep training on a pre-built deep learning network through training samples, and the deep training comprises the following steps:

[0022] Step S1: collecting a multi-view image sequence of tableware (including samples of various tableware);

[0023] Step S2: Using a structured light scanner to synchronously acquire true depth data;

[0024] Step S3: Create a depth map or depth class pairing database

[0025] Step S4: Implement illumination simulation, random occlusion and geometric transformation amplification strategies to perform data enhancement and sorting;

[0026] Step S5: Using a convolutional neural network architecture, inputting a two-dimensional image and outputting a depth probability distribution, a network architecture is constructed;

[0027] Step S6: Using the mean square error between the predicted depth and the true depth, design a loss function and optimize the training process;

[0028] Step S7: The training phase is divided into a pre-training phase and a fine-tuning phase. The pre-training phase uses a synthetic dataset to initialize network parameters, and the fine-tuning phase migrates to the real tableware dataset for optimization.

[0029] Preferably, the training samples include tableware samples, a perspective transformation module, a background simulation module, a lighting simulation module and an occlusion simulation module. The tableware samples include tableware of various materials and shapes. The perspective transformation module is used to collect information of tableware samples from multiple perspectives. The background simulation module is used to render and simulate multiple usage scenarios when collecting tableware sample information. The lighting simulation module is used to render and simulate different lighting scenes when collecting tableware sample information. The occlusion simulation module is used to simulate the situation of partial overlapping and occlusion between tableware when collecting tableware sample information.

[0030] Preferably, the step 2 comprises the following steps:

[0031] Step 2-1: extracting multiple sampling points from the surface data;

[0032] Step 2-2: Calculate the normal vectors of the plurality of sampling points to generate a one-to-one corresponding set of sampling points and a set of normal vectors;

[0033] Step 2-3: Obtain the optimal motion trajectory based on the sampling point set, the normal vector set, and the working mapping space of each spray arm.

[0034] Preferably, step 3 comprises the following steps:

[0035] Step 3-1: According to the optimal motion trajectory, that is, the jet motion trajectory P in three-dimensional space t =(x t ,y t ,z t ), t is the discrete motion time, through P t The physical size of the spray arm can be calculated to obtain the corresponding motion angle set θ of each spray arm motor j,k .

[0036] Preferably, step 4 comprises the following steps:

[0037] Step 4-1: According to the θ j,k The target angle set is used to control the motion motor. Each motion motor calculates and processes the target angle θ at that moment to obtain a control signal and gives it to the motion motor. At the same time, the motion detection sensor collects the motion parameter error at that moment and feeds it back to the motion control unit for the next control signal calculation.

[0038] A spray trajectory control device for cleaning equipment, comprising:

[0039] A data acquisition unit is used to obtain the motion state data of each spray arm, the position data of the target cleaning object, and the surface data of the target cleaning surface;

[0040] A trajectory generating unit, configured to plan an optimal motion trajectory according to the surface data;

[0041] An angle calculation unit, configured to generate a target motion angle set for a plurality of spray arms based on the optimal motion trajectory;

[0042] an instruction generating unit, configured to generate a control instruction for each of the control motors based on the target motion angle set, wherein the control instruction is used to control each of the control motors to drive each of the spray arms to move at a corresponding target motion angle;

[0043] The spray arm control unit is used to coordinate the multi-axis control of the spray arm to ensure that the spray arm performs cleaning according to the precise path.

[0044] Preferably, the spray arm control unit includes a spray arm control module, a trajectory generation module and a multi-axis collaborative control module. The spray arm control module is responsible for receiving cleaning target instructions from the interactive end and decomposing them into specific plans for multi-axis collaborative tasks. The trajectory generation module converts task requirements into trajectories of multi-axis collaborative motion modules based on the specific cleaning tasks issued by the spray arm control module to support the spray arm to perform cleaning according to a precise path. The multi-axis collaborative control module is used to coordinate and synchronize the movements between different spray arms to ensure the uniformity of the overall execution of the system.

[0045] Preferably, the spray arm control unit also includes a first-axis control system and a second-axis control system, the first-axis control system including a first-axis controller, a first motor driver, a first motor, a spray arm load and a first position sensor, the first-axis controller receives a control signal from the multi-axis collaborative control and sends it to the first motor driver, the first motor driver is used to amplify the signal of the first-axis controller and drive the first motor, the first motor drives the spray arm to perform rotational motion, the spray arm load is used to carry the spray arm and follow the command to complete the required precise positioning, the first position sensor is installed at a position next to the center of the spray arm; the second-axis control system includes a second-axis controller, a second motor driver, a second motor, a spray head load and a second position sensor, the second-axis controller receives a control signal from the multi-axis collaborative control and sends it to the second motor driver, the second motor driver is used to amplify the signal of the second-axis controller and drive the second motor, the second drive spray arm to perform rotational motion, the spray head load is used to carry the spray arm and follow the command to complete the required precise positioning, and the second position sensor is installed at a position next to the center of the spray arm.

[0046] Preferably, motion data feedback is respectively provided between the first axis controller and the first position sensor, and between the second axis controller and the second position sensor, and the motion data feedback is used to feed back the detected information to the first axis controller to ensure closed-loop regulation, and synchronization data is respectively provided between the first position sensor, the second position sensor and the multi-axis collaborative control, and the synchronization data is used to provide motion state information for global synchronization of the multi-axis collaborative control.

[0047] The beneficial effects are:

[0048] 1. The spray trajectory control method for cleaning equipment obtains motion state data of each spray arm, position data of the target cleaning object, and surface data of the target cleaning surface; plans an optimal motion trajectory based on the surface data; generates a target motion angle set for multiple spray arms based on the optimal motion trajectory; and generates control instructions for each control motor based on the target motion angle set. The control instructions are used to control each control motor to drive each spray arm to move at a corresponding target motion angle, thereby realizing automated control of the spray device, improving the controllability of the water flow trajectory, and thereby improving cleaning efficiency while reducing water consumption.

[0049] 2. This spray trajectory control method for cleaning equipment can effectively cover the surface of tableware, improve cleaning efficiency, and save water resources through fine control of the cleaning trajectory; it can also perform motion planning based on the different shapes of tableware to adapt to various cleaning scenarios; and improve the cleaning effect of the dishwasher and the user experience.

[0050] 3. This spray trajectory control method for cleaning equipment, through the settings of training samples and depth training, enables the depth estimation model to output and process depth information, and then generate a point cloud distribution map as surface data. With the help of surface data, it is possible to plan cleaning trajectories for tableware of various materials and shapes under various working conditions, accurately fit the surface of the tableware, greatly enhance the cleaning effect, ensure that there are no dead angles in cleaning, and comprehensively improve the quality of tableware cleaning. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 This is a flow chart of the spray trajectory control method for cleaning equipment provided by the present invention;

[0053] Figure 2It is a structural schematic diagram of the spray trajectory control device for cleaning equipment provided by the present invention;

[0054] Figure 3 It is a logical schematic diagram of the spray arm control unit provided by the present invention. DETAILED DESCRIPTION

[0055] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other implementations obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention.

[0056] Example 1

[0057] See also Figure 1 - Figure 3 , a spray trajectory control method for cleaning equipment, comprising the following steps:

[0058] Step 1: Collect motion data of multiple spray arms and surface data of target cleaning objects; in actual use scenarios, each degree of freedom of the spray arm includes a motion state detection sensor, and the required motion state includes the position, velocity, and acceleration of each spray degree of freedom. The motion state can be obtained by detecting one of the position, velocity, and acceleration states of each degree of freedom, and then obtaining the other states through differentiation and integration; each degree of freedom of each spray arm includes an independent motion state detection sensor, so that each degree of freedom can be finely controlled separately, and for complex cleaning targets, the coordinated motion of multiple degrees of freedom can be achieved. Among them, the position data of the target cleaning object (such as tableware) is obtained in order to determine its placement and direction; the surface data of the target cleaning surface is obtained, for example, the three-dimensional shape of the tableware surface is obtained through 3D modeling or visual recognition technology;

[0059] Step 2: Plan the optimal motion trajectory based on the collected surface data; split the planned optimal motion trajectory and the target cleaning surface to generate the front surface and the back surface, and then generate the coordinated trajectory of the front spray and the back spray respectively. In other words, based on the surface data of the target cleaning surface, through surface analysis, determine the shape and characteristics of the tableware surface, and then use normal vector calculation to calculate the normal vector of each sampling point on the target cleaning surface to determine the spray direction of the spray arm; finally, based on the sampling points, normal vectors and the working space of the spray arm, plan an optimal motion trajectory to ensure that the spray arm can cover the entire target cleaning surface and the spray direction is perpendicular to the target surface, thereby improving cleaning efficiency;

[0060] Step 3: Generate a set of target motion angles for multiple spray arms based on the optimal motion trajectory. Decompose the optimal motion trajectory into a series of discrete time points, each of which corresponds to a target position for the spray arm. Calculate the required rotation angles for each time point based on the physical dimensions and target position of the spray arm, and generate a set of target motion angles for each spray arm.

[0061] Step 4: Generate control instructions for the control motors based on the target motion angle set; the control instructions are used to control each control motor to drive each spray arm to move at the corresponding target motion angle; based on the target motion angle set, control instructions for the control motors are generated, and the control motors drive the spray arms to move according to the target motion angles, thereby achieving precise cleaning trajectory control;

[0062] The multiple spray arms are controlled by multiple motion motors, each of which is provided with a motion detection sensor. The multiple motion motors are all controlled by a main controller, and control instructions are transmitted to the main controller. The main controller drives the multiple spray arms to move at corresponding target motion angles through the multiple motion motors.

[0063] Through fine control of the cleaning trajectory, it can effectively cover the surface of tableware, improve cleaning efficiency and save water resources; it can perform motion planning according to tableware of different shapes to adapt to various cleaning scenarios; and improve the cleaning effect of the dishwasher and the user experience.

[0064] Furthermore, step 1 includes the following steps:

[0065] Step 1-1: Get the current position, current speed and current acceleration of each spray arm;

[0066] Step 1-2: When the target cleaning surface is a regular surface, the surface model pre-stored in the surface model database is retrieved and used as the surface data of the target cleaning surface. When the target cleaning surface is a regular surface, offline motion planning and motion control are performed. That is, the optimal motion trajectory is planned by using the pre-set bowl surface model database, achieving a relatively universal cleaning effect.

[0067] Step 1-3: When the target cleaning object is a special-shaped surface, obtain the surface model of the target cleaning object.

[0068] Furthermore, steps 1-3 include the following steps:

[0069] Step 1-3-1: Acquire a two-dimensional image of the target cleaning object;

[0070] Step 1-3-2: Input the two-dimensional image into the pre-trained depth estimation model to obtain the depth map or depth class output by the depth estimation model;

[0071] Step 1-3-3: Based on the depth map or depth class, obtain the point cloud distribution map of the target cleaning object, and use the point cloud distribution map as the surface data of the target cleaning surface;

[0072] When cleaning special-shaped tableware, motion planning is required to obtain better cleaning results. Through visual methods, the surface of the target tableware is modeled, and a two-dimensional image of the target tableware is obtained through a camera placed in the cleaning machine cavity. The two-dimensional image is input into a pre-trained depth estimation model, and the corresponding depth information can be output. The directly output depth information needs to be processed, including denoising and smoothing, to ensure the quality and accuracy of the depth map. The depth information in the depth map and the camera parameters can be used to obtain the point cloud distribution of the target tableware in the coordinates of the cleaning machine cavity. Through this three-dimensional point cloud distribution, the three-dimensional point cloud distribution result is used as surface data for cleaning trajectory motion planning.

[0073] In addition, the depth estimation model is obtained by deep training a pre-built deep learning network through training samples. The deep training includes the following steps:

[0074] Step S1: collecting a multi-view image sequence of tableware (including samples of various tableware);

[0075] Step S2: Using a structured light scanner or manual measurement to synchronously obtain true depth data;

[0076] Step S3: Create a depth map or depth class pairing database

[0077] Step S4: Implement illumination simulation, random occlusion and geometric transformation amplification strategies to perform data enhancement and sorting;

[0078] Step S5: Using a convolutional neural network architecture, inputting a two-dimensional image and outputting a depth probability distribution, a network architecture is constructed;

[0079] Step S6: Using the mean square error between the predicted depth and the true depth, design a loss function and optimize the training process;

[0080] Step S7: The training phase is divided into a pre-training phase and a fine-tuning phase. In the pre-training phase, the network parameters are initialized using a synthetic dataset, and in the fine-tuning phase, the optimization is migrated to a real tableware dataset.

[0081] In addition, the training samples include tableware samples, perspective transformation module, background simulation module, lighting simulation module and occlusion simulation module. The tableware samples include tableware of various materials and shapes. The perspective transformation module is used to collect information of tableware samples from multiple perspectives. The background simulation module is used to render and simulate various usage scenarios when collecting tableware sample information. The lighting simulation module is used to render and simulate different lighting scenes when collecting tableware sample information. The occlusion simulation module is used to simulate the partial overlapping occlusion between tableware when collecting tableware sample information. Through the settings of training samples and depth training, the depth estimation model can output and process depth information, and then generate a point cloud distribution map as surface data. With the help of surface data, it is possible to plan cleaning trajectories for tableware of various materials and shapes under various working conditions, accurately fit the surface of the tableware, greatly enhance the cleaning effect, ensure cleaning without dead angles, and comprehensively improve the quality of tableware cleaning.

[0082] In addition, step 1-3-2: inputting the two-dimensional image into a pre-trained depth estimation model to obtain the depth map or depth class output by the depth estimation model also includes a depth category processing module. The depth category processing can identify different tableware categories and divide different depths into multiple depth intervals. Each interval corresponds to a depth class. The depth of tableware in each depth interval is close. For different depth classes, the depth category processing module plans a variety of optimal trajectories offline in advance. The multiple optimal trajectories correspond to multiple depth intervals. In actual use, the optimal trajectory can be obtained by identifying the depth class of the tableware. This consumes less computing power and has a faster recognition speed. While ensuring recognition efficiency, it can also optimize computing power consumption.

[0083] It is worth noting that step 2 includes the following steps:

[0084] Step 2-1: Extract multiple sampling points from the surface data;

[0085] Step 2-2: Calculate the normal vectors of multiple sampling points and generate a one-to-one corresponding set of sampling points and a set of normal vectors;

[0086] Step 2-3: Obtain the optimal motion trajectory based on the sampling point set, normal vector set, and the working mapping space of each spray arm;

[0087] When the surface and the motion parameters of the spray arm are known, motion trajectory optimization becomes an optimization problem: optimize the motion trajectory of the spray arm so that it produces a uniform and large impact force on the target 3D surface.

[0088] Define the objective function and use the optimization algorithm to obtain the optimal motion trajectory and calculate the comprehensive cleaning force, the comprehensive cleaning force W m The calculation process is as follows:

[0089] Comprehensive cleaning power W m It is used to measure the effect of the mth trajectory in cleaning the target n surface scattered points, which is defined as the impact force F m,n The weighted sum of the mean and variance of , where the weight of the mean is 1 and the weight of the variance is -λ;

[0090] The formula is: Among them, W m represents the comprehensive cleaning force of the mth trajectory, μ m represents the average impact force of all surface scattered points under the mth trajectory, represents the variance of the impact force of all surface scattered points under the mth trajectory; λ represents the weight parameter, which is used to adjust the influence of the variance on the comprehensive cleaning force and needs to be adjusted according to the cleaning force amplitude and surface point position;

[0091] Expand formula ① and substitute the specific expressions of mean and variance into formula ① to obtain formula ②:

[0092]

[0093] Among them, F m,i represents the impact force of the i-th surface scattered point under the m-th trajectory, n represents the total number of surface scattered points; i = 1, 2, 3, ..., n;

[0094] For formula ②, further expand the variance term to obtain formula ③:

[0095]

[0096] The final expansion formula ⑤ is obtained:

[0097]

[0098] In order to maximize the comprehensive cleaning power, it is necessary to iteratively optimize the trajectory so that W m To reach the maximum value, the formula is expressed as:

[0099] in, Indicates the maximum value of comprehensive cleaning power;

[0100] The F m,i It can be calculated from the motion parameters of the spray arm and the depth distribution predicted by the previous model, that is, the 3D surface distribution of the bowl surface.

[0101] In actual usage scenarios, the 3D surface model of the target cleaning surface is modeled based on the actual placement of the tableware and the position of the rotating spray arm, and then multiple sampling points are generated on the surface. According to the surface and the distribution of the sampling points, the normal vector of each sampling point can be calculated to generate a one-to-one corresponding sampling point set and normal vector set; based on the sampling point set and the normal vector set, as well as the working mapping space of each spray arm, the optimal path can be optimized to ensure full coverage of the target cleaning space and the perpendicularity of the spray angle to the target surface in the shortest time.

[0102] It is worth noting that step 3 includes the following steps:

[0103] Step 3-1: According to the optimal motion trajectory, that is, the jet motion trajectory P in three-dimensional space t =(x t ,y t ,z t ), t is the discrete motion time, through P t The physical size of the spray arm can be calculated to obtain the corresponding motion angle set θ of each spray arm motor j,k , where j represents the spray arm modules at different positions in the cleaning equipment, such as the front spray arm and the rear spray arm; k represents the different degrees of freedom of each spray arm.

[0104] It is worth mentioning that step 4 includes the following steps:

[0105] Step 4-1: According to θ j,k The target angle set is used to control the motion motors. Each motion motor calculates and processes the target angle θ at that moment to obtain a control signal and gives it to the motion motor. At the same time, the motion detection sensor collects the motion parameter error at that moment and feeds it back to the motion control unit for the next control signal calculation.

[0106] By acquiring the motion state data of each spray arm, the position data of the target cleaning object and the surface data of the target cleaning surface; planning the optimal motion trajectory according to the surface data; generating a target motion angle set for each spray arm based on the optimal motion trajectory; generating control instructions for each control motor based on the target motion angle set, and the control instructions are used to control each control motor to drive each spray arm to move at the corresponding target motion angle, thereby realizing automatic control of the spray device, improving the controllability of the water flow trajectory, thereby improving the cleaning efficiency, and reducing water consumption.

[0107] Example 2

[0108] See also Figure 2 - Figure 3 , a spray trajectory control device for cleaning equipment, applied to the spray trajectory control method for cleaning equipment in Example 1, comprising:

[0109] A data acquisition unit is used to obtain the motion state data of each spray arm, the position data of the target cleaning object, and the surface data of the target cleaning surface;

[0110] A trajectory generation unit, used to plan the optimal motion trajectory based on the surface data;

[0111] An angle calculation unit, used to generate a target motion angle set for each spray arm based on the optimal motion trajectory;

[0112] An instruction generation unit is used to generate control instructions for each control motor based on the target motion angle set, and the control instructions are used to control each control motor to drive each spray arm to move at the corresponding target motion angle;

[0113] The spray arm control unit is used to coordinate the multi-axis control of the spray arm to ensure that the spray arm performs cleaning according to the precise path.

[0114] Furthermore, the spray arm control unit includes a spray arm control module, a trajectory generation module and a multi-axis collaborative control module. The spray arm control module is responsible for receiving cleaning target instructions from the interactive end and decomposing them into specific plans for multi-axis collaborative tasks. The trajectory generation module converts the task requirements into the trajectory of the multi-axis collaborative motion module according to the specific cleaning tasks issued by the spray arm control module to support the spray arm to perform cleaning according to a precise path. The multi-axis collaborative control module is used to coordinate and synchronize the movements between different spray arms to ensure the uniformity of the overall execution of the system.

[0115] Furthermore, the spray arm control unit also includes a first-axis control system and a second-axis control system. The first-axis control system includes a first-axis controller, a first motor driver, a first motor, a spray arm load and a first position sensor. The first-axis controller receives the control signal from the multi-axis collaborative control and sends it to the first motor driver. The first motor driver is used to amplify the signal of the first-axis controller and drive the first motor. The first motor drives the spray arm to perform rotational motion. The spray arm load is used to carry the spray arm and follow the command to complete the required precise positioning. The first position sensor is installed at a position next to the center of the spray arm; the second-axis control system includes a second-axis controller, a second motor driver, a second motor, a spray head load and a second position sensor. The second-axis controller receives the control signal from the multi-axis collaborative control and sends it to the second motor driver. The second motor driver is used to amplify the signal of the second-axis controller and drive the second motor. The second drive spray arm performs rotational motion. The spray head load is used to carry the spray arm and follow the command to complete the required precise positioning. The second position sensor is installed at a position next to the center of the spray arm.

[0116] Furthermore, motion data feedback is respectively provided between the first axis controller and the first position sensor, and between the second axis controller and the second position sensor. The motion data feedback is used to feed back the detected information to the first axis controller to ensure closed-loop regulation. Synchronization data is respectively provided between the first position sensor, the second position sensor and the multi-axis collaborative control. The synchronization data is used to provide motion status information for global synchronization of the multi-axis collaborative control.

[0117] In addition, the spray trajectory control device for cleaning equipment is operated by a computer device, which includes a processor, a memory and a network interface connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The model prediction of the computer device is used to store static information and dynamic information data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it can implement the spray trajectory control method for cleaning equipment in Example 1.

[0118] In addition, the computer device is also provided with a computer storage medium, which contains a plurality of program instructions, including instructions for executing the spray trajectory control method for cleaning equipment in Example 1.

[0119] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A spray trajectory control method for cleaning equipment, characterized in that: The following steps are involved: Step 1: Collect motion data of multiple spray arms and surface data of the target object to be cleaned; Step 2: Plan the optimal motion trajectory based on the collected surface data; Step 3: generating a target motion angle set of multiple spray arms based on the optimal motion trajectory; Step 4: generating a control instruction for controlling the motor based on the target motion angle set; Multiple spray arms are respectively controlled by multiple motion motors, and motion detection sensors are respectively provided on the multiple motion motors. The multiple motion motors are all controlled by a main controller. The control instructions are transmitted to the main controller, and the main controller drives the multiple spray arms to move at corresponding target motion angles through the multiple motion motors.

2. The spray trajectory control method for cleaning equipment according to claim 1, characterized in that: The step 1 comprises the following steps: Step 1-1: Get the current position, current speed and current acceleration of each spray arm; Step 1-2: When the target cleaning object is a regular curved surface, a surface model pre-stored in a surface model database is retrieved and the surface model is used as the surface data of the target cleaning surface; Step 1-3: When the target cleaning object is an irregular curved surface, a surface model of the target cleaning object is obtained.

3. The spray trajectory control method for cleaning equipment according to claim 2, characterized in that: The steps 1-3 include the following steps: Step 1-3-1: Acquire a two-dimensional image of the target cleaning object; Step 1-3-2: Input the two-dimensional image into a pre-trained depth estimation model to obtain a depth map or depth class output by the depth estimation model; Step 1-3-3: Based on the depth map or depth class, obtain a point cloud distribution map of the target cleaning object, and use the point cloud distribution map as the surface data of the target cleaning surface.

4. The spray trajectory control method for cleaning equipment according to claim 3, characterized in that: The depth estimation model is obtained by performing deep training on a pre-built deep learning network through training samples, and the deep training includes the following steps: Step S1: collecting a multi-view image sequence of tableware (including samples of various tableware); Step S2: Using a structured light scanner or manual measurement modeling to synchronously obtain true depth data; Step S3: establishing a depth map or depth class pairing database; Step S4: Implement illumination simulation, random occlusion and geometric transformation amplification strategies to perform data enhancement and sorting; Step S5: Using a convolutional neural network architecture, inputting a two-dimensional image and outputting a depth probability distribution, a network architecture is constructed; Step S6: Using the mean square error between the predicted depth and the true depth, design a loss function and optimize the training process; Step S7: The training phase is divided into a pre-training phase and a fine-tuning phase. The pre-training phase uses a synthetic dataset to initialize network parameters, and the fine-tuning phase migrates to the real tableware dataset for optimization.

5. The spray trajectory control method for cleaning equipment according to claim 4, characterized in that: The training samples include tableware samples, a perspective transformation module, a background simulation module, a lighting simulation module and an occlusion simulation module. The tableware samples include tableware of various materials and shapes. The perspective transformation module is used to collect information of the tableware samples from multiple perspectives. The background simulation module is used to render and simulate multiple usage scenarios when collecting tableware sample information. The lighting simulation module is used to render and simulate different lighting scenes when collecting tableware sample information. The occlusion simulation module is used to simulate the situation of partial overlap and occlusion between tableware when collecting tableware sample information.

6. The spray trajectory control method for cleaning equipment according to claim 5, characterized in that: The step 2 comprises the following steps: Step 2-1: extracting multiple sampling points from the surface data; Step 2-2: Calculate the normal vectors of the plurality of sampling points to generate a one-to-one corresponding set of sampling points and a set of normal vectors; Step 2-3: Obtain the optimal motion trajectory based on the sampling point set, the normal vector set, and the working mapping space of each spray arm.

7. The spray trajectory control method for cleaning equipment according to claim 6, characterized in that: The step 3 comprises the following steps: Step 3-1: According to the optimal motion trajectory, that is, the jet motion trajectory P in three-dimensional space t =(x t ,y t ,z t ), t is the discrete motion time, through P t The physical size of the spray arm can be calculated to obtain the corresponding motion angle set θ of each spray arm motor j,k , where j represents the rotary arm modules at different positions in the cleaning equipment; k represents the different degrees of freedom of each rotary arm.

8. The spray trajectory control method for cleaning equipment according to claim 7, characterized in that: The step 4 comprises the following steps: Step 4-1: According to the θ j,k The target angle set is used to control the motion motor. Each motion motor calculates and processes the target angle θ at that moment to obtain a control signal and gives it to the motion motor. At the same time, the motion detection sensor collects the motion parameter error at that moment and feeds it back to the motion control unit for the next control signal calculation.

9. A spray trajectory control device for cleaning equipment, applied to the spray trajectory control method for cleaning equipment according to any one of claims 1 to 8, characterized in that: include: A data acquisition unit is used to obtain the motion state data of each spray arm, the position data of the target cleaning object, and the surface data of the target cleaning surface; A trajectory generating unit, configured to plan an optimal motion trajectory according to the surface data; An angle calculation unit, configured to generate a target motion angle set for a plurality of spray arms based on the optimal motion trajectory; an instruction generating unit, configured to generate a control instruction for each of the control motors based on the target motion angle set, wherein the control instruction is used to control each of the control motors to drive each of the spray arms to move at a corresponding target motion angle; The spray arm control unit is used to coordinate the multi-axis control of the spray arm to ensure that the spray arm performs cleaning according to the precise path.

10. The spray trajectory control device for cleaning equipment according to claim 9, characterized in that: The spray arm control unit includes a spray arm control module, a trajectory generation module and a multi-axis collaborative control module. The spray arm control module is responsible for receiving cleaning target instructions from the interactive end and decomposing them into specific plans for multi-axis collaborative tasks. The trajectory generation module converts task requirements into trajectories of multi-axis collaborative motion modules based on the specific cleaning tasks issued by the spray arm control module to support the spray arm to perform cleaning according to a precise path. The multi-axis collaborative control module is used to coordinate and synchronize the movements between different spray arms to ensure the uniformity of the overall execution of the system.

11. The spray trajectory control device for cleaning equipment according to claim 10, characterized in that: The spray arm control unit also includes a first axis control system and a second axis control system. The first axis control system includes a first axis controller, a first motor driver, a first motor, a spray arm load, and a first position sensor. The first axis controller receives a control signal from the multi-axis collaborative control and sends it to the first motor driver. The first motor driver is used to amplify the signal of the first axis controller and drive the first motor. The first motor drives the spray arm to perform rotational motion. The spray arm load is used to carry the spray arm and follow the command to complete the required precise positioning. The first position sensor is installed next to the center of the spray arm driven by the first motor. The second-axis control system includes a second-axis controller, a second motor driver, a second motor, a nozzle load and a second position sensor. The second-axis controller receives a control signal from the multi-axis collaborative control and sends it to the second motor driver. The second motor driver is used to amplify the signal of the second-axis controller and drive the second motor. The second motor drives the spray arm to perform rotational motion. The nozzle load is used to carry the spray arm and follow the command to complete the required precise positioning. The second position sensor is installed next to the center of the spray arm driven by the second motor.

12. The spray trajectory control device for cleaning equipment according to claim 11, characterized in that: Motion data feedback is respectively provided between the first axis controller and the first position sensor, and between the second axis controller and the second position sensor. The motion data feedback is used to feed back the detected information to the first axis controller to ensure closed-loop regulation. Synchronization data is respectively provided between the first position sensor, the second position sensor and the multi-axis collaborative control. The synchronization data is used to provide motion state information for global synchronization of the multi-axis collaborative control.

Images