Mechanical arm-mounted laser cleaning system and method

Abstract

The application discloses a kind of mechanical arm carries laser cleaning system and cleaning method, comprising: mechanical arm, laser cleaning device and controller, wherein the mechanical arm is used to according to the instruction of the controller, to make the end of the mechanical arm reach the position of the dirt to be cleaned to carry out cleaning operation;The laser cleaning device is connected to the end of the mechanical arm, for according to the movement of the mechanical arm to the dirt to be cleaned laser cleaning;The controller is electrically connected with the mechanical arm and the laser cleaning device, the controller is used to identify the dirt to be cleaned under large-format application scene, and obtains the position of the dirt to be cleaned;Also used to control the most matched mechanical arm in the mechanical arm according to the cleaning dirt position, and laser cleaning device moves to the position of the dirt to be cleaned;Also used to control the laser cleaning device to the dirt to be cleaned cleaning.The application realizes the dirt under large-format application scene to carry out efficient and environmental cleaning purpose.

Description

Technical Field

[0001] This invention relates to the field of laser cleaning technology, and more specifically to a robotic arm equipped with a laser cleaning system and method. Background Technology

[0002] In industry, there is a widespread demand for large-scale cleaning. For example, during major ship repairs, it is necessary to clean the scale on the hull. This scale is composed of algae and shellfish deposits, which are dense and hard, and inevitably cause large-scale corrosion. Another example is the accumulation of dust, oil, carbon deposits, oxides and other pollutants on the outer surface and components of aircraft fuselages. These pollutants not only affect the appearance of the aircraft, but also reduce the surface smoothness, increase frictional resistance, and become a factor that induces corrosion of the aircraft.

[0003] Currently, the removal of coatings and scale from ship hulls still relies on workers manually hammering and scraping with hammers and shovels. Because the scale is hard, dense, and covers a large area, maintenance is time-consuming and labor-intensive. Ship cleaning has evolved from primitive manual cleaning and simple mechanical scraping to sandblasting, whitening, mechanical cleaning, and chemical cleaning, but it still suffers from time-consuming, inefficient, and poor cleaning precision. Furthermore, aircraft surface cleaning faces similar challenges. In terms of labor costs, a large number of ground staff are needed to apply cleaning agents and use cleaning tools to wipe the fuselage from top to bottom and front to back, resulting in heavy workloads and potential safety hazards. Regarding cleaning technology, aircraft surfaces need to be repainted after a certain period, but the old paint must be completely removed before repainting. Traditional mechanical paint removal methods can easily damage the aircraft's metal surfaces, posing a safety hazard.

[0004] Therefore, there is a need to provide a dirt cleaning system that can be applied to large-format applications, in order to solve the problems of low dirt cleaning efficiency and high pollution in existing technologies for large-format applications. Summary of the Invention

[0005] The purpose of this invention is to overcome the above-mentioned technical deficiencies and provide a robotic arm equipped with a laser cleaning system and method to solve the technical problems of low efficiency and high pollution when cleaning dirt in large-format application scenarios in the prior art.

[0006] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a robotic arm equipped with a laser cleaning system, comprising:

[0008] The robotic arm is used to perform cleaning operations by having its end arm reach the location of the dirt to be cleaned, according to the instructions of the controller.

[0009] The laser cleaning device corresponds one-to-one with the robotic arm and is connected to the end of the robotic arm, and is used to perform laser cleaning on the dirt to be cleaned according to the movement of the robotic arm.

[0010] The controller is electrically connected to the robotic arm and the laser cleaning device. The controller is used to identify the dirt to be cleaned in large-format application scenarios and to obtain the location of the dirt to be cleaned.

[0011] The controller is also used to control the most suitable robotic arm, equipped with a laser cleaning device, to move to the location of the dirt to be cleaned, based on the location of the dirt to be cleaned.

[0012] The controller is also used to control the laser cleaning device to clean the dirt to be cleaned.

[0013] In some embodiments, the robotic arm equipped with the laser cleaning system further includes a vehicle-mounted device, which includes a mobile platform, a lifting mechanism, and a retrieval device. The lifting mechanism is fixed to the platform of the mobile platform, and one end of the robotic arm is fixedly connected to the lifting mechanism so that the robotic arm can move up and down. The retrieval device is installed on the mobile platform.

[0014] In some embodiments, the controller includes an acquisition device for identifying dirt to be cleaned in large-format application scenarios and acquiring the location of the dirt to be cleaned. The acquisition device includes:

[0015] The model training module is used to train the YOLOX model that incorporates a coordinated attention mechanism;

[0016] An image acquisition module is used to acquire images of the dirt to be cleaned based on the improved YOLOX model.

[0017] The target positioning module is used to determine the position of the image of the dirt to be cleaned based on a preset coordinate transformation rule.

[0018] In some embodiments, the controller further includes a robotic arm trajectory planning module, which is used to select a matching robotic arm to clean the corresponding dirt based on the position of the robotic arm and the position of the dirt image to be cleaned, and to use an improved elite non-dominated sorting genetic algorithm to determine the path of the matching robotic arm from its initial position to the position of the dirt image to be cleaned.

[0019] In some embodiments, the robotic arm trajectory planning module includes an initial path determination module, a desired path determination module, and a target path determination module;

[0020] The initial path determination module is used to initialize the parameters of the improved elite non-dominated sorting genetic algorithm and determine the initial path of the matched robotic arm movement.

[0021] The desired path determination module is used to determine the desired path based on the Pareto front and the initial path.

[0022] The target path determination module is used to determine the parameters of the desired path according to the elite strategy, thereby obtaining the target path of the matched robotic arm movement.

[0023] In some embodiments, the target path determination module includes a sorting group selection module, a direction-based crossover module, and an adaptive, precision-controllable mutation module.

[0024] The sorting group selection module is used to determine the feasible nodes of the target path based on a preset sorting group algorithm.

[0025] The direction-based crossing module is used to determine the optimal range of the target path based on the feasible nodes of the target path according to a preset direction crossing algorithm.

[0026] The adaptive precision controllable mutation module is used to improve the speed of determining the target path within the optimal range of the target path based on the preset adaptive mechanism.

[0027] In some embodiments, the robotic arm trajectory planning module further includes an optimization module, which is used to determine the most matching target path from multiple trajectories to be selected.

[0028] In some embodiments, the initial path can be expressed by the following formula:

[0029]

[0030] in, is a binomial coefficient, Pi is a given control point for constructing the Bézier curve, T is the travel time, λ represents the normalized time, and satisfies t = λT. It is a Bernstein basis polynomial. Indicates joint velocity.

[0031] In some embodiments, the desired path can be expressed by the following formula:

[0032]

[0033] Where f1(t) is the total stroke time, f2(t) is the variance of the actuator torque, and τ i (t) and τ i (t-1) represents the torque of the actuator in the previous and current situations, respectively, and n represents the number of robot arm joints.

[0034] Secondly, the present invention also provides a cleaning method for a robotic arm equipped with a laser cleaning device, comprising:

[0035] Acquire images of dirt in large-format application scenarios, and determine the position of the dirt image to be cleaned based on preset coordinate transformation rules;

[0036] Based on the position of the image of dirt to be cleaned, an improved elite non-dominated sorting genetic algorithm is used to determine the target path for the best-matching robotic arm to move from its initial position to the position of the image of dirt to be cleaned.

[0037] Based on the position of the image of the dirt to be cleaned and the target path, the laser cleaning device is moved to the position of the image of the dirt to be cleaned by moving the most suitable robotic arm and the dirt is cleaned.

[0038] Compared with existing technologies, the cleaning system and method for robotic arms equipped with laser cleaning devices provided by this invention adopts an integrated system that combines a controller and a robotic arm, with each robotic arm equipped with a laser cleaning device. First, the controller acquires the dirt to be cleaned in a large-format application scenario and determines the location of the dirt. Then, the controller selects the robotic arm that best matches the location of the dirt based on the location of the dirt and the position of the robotic arm, and controls the robotic arm to drive the laser cleaning device to the location of the dirt. Finally, the laser cleaning device is controlled to perform laser cleaning on the dirt, thereby solving the problems of difficulty in locating dirt in large-format application scenarios, low efficiency of manual cleaning, and high pollution. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of an embodiment of the cleaning system with a robotic arm equipped with a laser cleaning device provided by the present invention;

[0040] Figure 2 This is a schematic diagram of an embodiment of a cleaning system with a robotic arm equipped with a laser cleaning device provided by the present invention, in which multiple robotic arms work together.

[0041] Figure 3 This is a schematic diagram of a controller embodiment in the cleaning system of the robotic arm equipped with a laser cleaning device provided by the present invention;

[0042] Figure 4 This is a schematic diagram of the structure of an embodiment of the acquisition device in the cleaning system of the robotic arm equipped with the laser cleaning device provided by the present invention;

[0043] Figure 5 This is a schematic diagram of the structure of a robotic arm trajectory planning module in a cleaning system with a robotic arm equipped with a laser cleaning device provided by the present invention.

[0044] Figure 6This is a flowchart of an embodiment of the cleaning method for a robotic arm equipped with a laser cleaning device provided by the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0046] The cleaning system and method involving a robotic arm equipped with a laser cleaning device disclosed in this invention are applicable to various large-format applications, such as airports, train stations, bus stations, solar photovoltaic panels, or the outer surfaces of ships. These applications are characterized by their large area and difficulty in cleaning. Ordinary cleaning methods are not only inefficient but also prone to environmental pollution and damage to the surface of the objects being cleaned due to the use of various chemical cleaning agents. Therefore, a cleaning system employing a mechanically coordinated laser cleaning device can not only meet the need for cleaning large-format applications by controlling the movement of the robotic arm, reducing manual labor, but also utilize laser cleaning to cause a series of complex physical changes in the dirt on the surface of the object to be cleaned, such as vibration, melting, evaporation, and combustion, causing the dirt to detach from the object's surface and allowing for the recovery of dirt powder, thus greatly reducing pollution.

[0047] This invention provides a robotic arm equipped with a laser cleaning system; please refer to [link / reference]. Figure 1 and Figure 2 It includes: multiple robotic arms 1, multiple laser cleaning devices 2, and a controller 3, wherein:

[0048] The robotic arm 1 is used to perform cleaning operations by having its end end reach the location of the dirt to be cleaned, according to the instructions of the controller 3.

[0049] The laser cleaning device 2 is connected to the end of the robotic arm 1 and is used to perform laser cleaning on the dirt to be cleaned according to the movement of the robotic arm 1.

[0050] The controller 3 is electrically connected to the robotic arm 1 and the laser cleaning device 2. The controller 3 is used to identify the dirt to be cleaned in large-format application scenarios and obtain the location of the dirt to be cleaned.

[0051] The controller 3 is also used to control the most compatible robotic arm in the robotic arm 1, equipped with a laser cleaning device, to move to the location of the dirt to be cleaned, according to the location of the dirt to be cleaned.

[0052] The controller 3 is also used to control the laser cleaning device 2 to clean the dirt to be cleaned.

[0053] In this embodiment, the cleaning system with a robotic arm equipped with a laser cleaning device adopts an integrated system that combines a controller and a robotic arm, with each robotic arm equipped with a laser cleaning device. First, the controller acquires the dirt to be cleaned in the large-format application scenario and determines the location of the dirt to be cleaned. Then, the controller selects the robotic arm that best matches the location of the dirt to be cleaned based on the location of the dirt to be cleaned and the position of the robotic arm, and controls the robotic arm to drive the laser cleaning device to the location of the dirt to be cleaned. Finally, the controller controls the laser cleaning device to perform laser cleaning on the dirt to be cleaned, thereby solving the problems of difficulty in locating the dirt in large-format application scenarios, low efficiency of manual cleaning, and high pollution.

[0054] It should be noted that the present invention employs multiple robotic arms working in concert. The controller selects the robotic arm closest to the location of the dirt to be cleaned or the robotic arm with the most convenient travel path, based on the location of the dirt to be cleaned and the positions of the multiple robotic arms. This optimal robotic arm cleans the dirt to be cleaned to a certain extent, thereby improving the efficiency of the cleaning work.

[0055] Furthermore, the robotic arm 1 involved in the embodiments of the present invention is a common robotic arm in the prior art. The number of joints of the robotic arm is not limited and can be set according to actual needs. At the same time, the number of joints of different robotic arms 1 can be the same or different. It is understood that when using the controller to plan the motion path of the robotic arm, the more joints the robotic arm has, the greater the difficulty of path planning.

[0056] Specifically, the laser cleaning device 2 can be a technologically mature laser cleaner from the existing technology, and the laser cleaner is electrically connected to the controller and mounted on the end of the robotic arm. The laser cleaning device consists of a laser cleaning module including a fiber laser, a fiber laser output arm, a laser cleaning head, fiber optic cables, displacement sensors, and a controller. The controller 3 is the "decision-making body" that issues commands, that is, it coordinates and directs the operation of the entire computer system. The controller can be a microcontroller, a PID controller, a programmable controller, etc.

[0057] In some embodiments, the robotic arm equipped with the laser cleaning system further includes a vehicle-mounted device 4, which includes a mobile platform 41, a lifting mechanism 42, and a retrieval device 43. The lifting mechanism 42 is fixed to the platform of the mobile platform 41, and one end of the corresponding robotic arm 1 is fixedly connected to the lifting mechanism 42 so that the robotic arm 1 can move up and down. The retrieval device 43 is installed on the mobile platform.

[0058] It should be noted that by mounting the robotic arm on a moving platform, the movement of the platform can move the robotic arm and the laser cleaning device horizontally, thereby changing the position of the laser cleaning device and bringing it infinitely closer to the location of the dirt to be cleaned. Furthermore, by setting up a lifting mechanism, the robotic arm can move vertically, changing the vertical position of the laser cleaning device and bringing it infinitely closer to the location of the dirt to be cleaned. At the same time, by setting up a recovery device, when the laser head is cleaning the target, the dust suction head is also activated, which generates negative pressure to suck up and recover the dust generated during the cleaning process. The sucked-up dust is then collected in the dust filtration and recovery device through the adsorption and recovery pipeline, further improving the environmental protection performance.

[0059] In some embodiments, please refer to Figure 3 The controller 3 includes an acquisition device 31, which is used to identify the dirt to be cleaned in large-format application scenarios and acquire the location of the dirt to be cleaned. Please refer to [link to relevant documentation]. Figure 4 The acquisition device 31 includes:

[0060] Model training module 311 is used to train the YOLOX model that incorporates the coordinated attention mechanism;

[0061] Image acquisition module 312 is used to acquire an image of the dirt to be cleaned based on the improved YOLOX model;

[0062] The target positioning module 313 is used to determine the position of the image of the dirt to be cleaned based on a preset coordinate transformation rule.

[0063] In this embodiment, the acquisition module 31 is used to identify the dirt to be cleaned in large-format application scenarios and to acquire the location of the dirt to be cleaned. First, the YOLOX model with a fusion coordination attention mechanism is trained to improve the recognition ability of the dirt to be cleaned. Then, the trained model is used to extract the dirt to be cleaned in large-format application scenarios, and the location of the image of the dirt to be cleaned is determined based on the preset coordinate transformation rules.

[0064] Specifically, a new network structure is formed by combining the existing YOLOX network structure with a coordinated attention mechanism. At the same time, a depth camera installed on the robotic arm acquires images of the dirt to be cleaned. The images are then preprocessed and labeled with bounding boxes. The network is then divided into training and testing sets. Finally, the model is trained in a configured environment to obtain a model for detecting target objects.

[0065] In one specific embodiment, the coordinated attention mechanism encodes channel relationships and long-range dependencies using precise location information. The specific steps can be divided into coordinate information embedding and coordinated attention generation, wherein:

[0066] In the process of embedding coordinate information, global pooling is often used to globally encode spatial information for channel attention. However, because it compresses global spatial information into the channel descriptor, it is difficult to preserve positional information. In order to enable the attention module to capture remote spatial interactions with precise positional information, the coordinated attention mechanism decomposes global pooling according to the following formula, transforming it into a one-to-one one-dimensional feature encoding operation: Specifically, given an input X with dimensions (H, W, C), where H represents the height of the image, W represents the width of the image, and C is the number of image channels, we first use pooling convolution kernels of size (H, 1) or (1, W) to encode each channel along the horizontal and vertical coordinates respectively. Therefore, the output of the Cth channel with height H can be represented as: Similarly, the output of the Cth channel with width W can be written as: The two transformations described above aggregate features along two spatial directions to obtain a pair of direction-aware feature maps. These two transformations also allow the attention module to capture long-range dependencies along one spatial direction and retain precise location information along the other spatial direction, which helps the network to more accurately locate the target of interest.

[0067] After generating the coordinated attention, a fully connected operation is performed on the above transformation after the coordinate information is embedded, and then the convolutional transformation function F1 is used to transform it. The resulting transformation can be expressed by the following formula: f = δ(F1([z h ,z w ])), where [z h ,z w [] represents a fully connected operation along the spatial dimension, δ is a non-linear activation function, f is an intermediate feature map encoding spatial information in the horizontal and vertical directions, and then the intermediate feature map f is decomposed into two separate tensors f along the spatial dimension. h and f w and f h and f w Through two convolution transformations F respectively h and F w Transforming it into a tensor with the same number of channels as the input X, we get: g h =σ(F h (f h )), g w =σ(F w (f w ), where σ is the sigmoid activation function, and then the output g is... h and g w After expanding the values ​​and using them as attention weights, the output of the coordinated attention module can finally be written as:

[0068] Furthermore, image information is acquired through the trained model, and the debris to be cleaned on the ship is identified by a depth camera mounted on the robotic arm. The improved YOLOX model will identify the debris to be cleaned on the ship.

[0069] Furthermore, the dirt to be cleaned is located, and the position of the dirt to be cleaned is determined. The camera coordinate information of all the dirt to be cleaned detected by the depth camera is converted into real-world coordinates and transmitted back to the vehicle controller to provide coordinate information for the robotic arm to perform trajectory planning.

[0070] In some embodiments, please refer to Figure 3 The controller also includes a robotic arm trajectory planning module 32, which is used to select a matching robotic arm to clean the corresponding dirt by using an improved elite non-dominated sorting genetic algorithm based on the position of the robotic arm and the position of the dirt image to be cleaned, and to determine the path of the matching robotic arm from the initial position to the position of the dirt image to be cleaned.

[0071] It should be noted that a thick layer of scale forms on the hull below the waterline of large ships. This scale is dense, hard, and covers a large area. In order to complete the cleaning task quickly and efficiently, this embodiment of the invention proposes to deploy a cluster of multiple robotic arms. Therefore, it is necessary to select an algorithm that can coordinate the regional directional operation of multiple robotic arms to distribute and control the operation of the cluster of multiple robotic arms. Hence, this embodiment of the invention proposes a multi-objective trajectory planning method based on an improved elite non-dominated sorting genetic algorithm (INSGA-II) to achieve the effects of maintaining the robotic arm formation, obstacle avoidance, and collaborative processing.

[0072] In this embodiment, the robotic arm trajectory planning module 32 in the controller uses an improved elite non-dominated sorting genetic algorithm to select the robotic arm closest to the dirt to be cleaned from multiple robotic arm groups, and plans the target trajectory of the robotic arm to move to the position of the dirt to be cleaned, so that the best matching robotic arm can reach the position of the dirt to be cleaned in the shortest time and drive the laser cleaning device to clean the dirt.

[0073] In some embodiments, please refer to Figure 5 The robotic arm trajectory planning module 32 includes an initial path determination module 321, a desired path determination module 322, and a target path determination module 323;

[0074] The initial path determination module 321 is used to initialize the parameters of the improved elite non-dominated sorting genetic algorithm and determine the initial path of the matched robotic arm movement.

[0075] The desired path determination module 322 is used to determine the desired path based on the Pareto front and the initial path.

[0076] The target path determination module 323 is used to determine the parameters of the desired path according to the elite strategy, thereby obtaining the target path of the matched robotic arm movement.

[0077] In this embodiment, the fifth-degree polynomial g(λ) = 10λ is first designed as follows. 3 -15λ 4 + 6λ 5 ,λ∈[0,1], and substituting the fifth-degree polynomial into the nth-degree Bézier curve, the Bézier curve is as shown in the formula. By setting x = g(λ), we obtain the composite polynomial. Thus, the initial path is obtained: And store the trajectory, where, It is a binomial coefficient, P i Given control points for constructing the Bézier curve, T is the travel time, λ represents the normalized time, and t = λT satisfies this condition. It is a Bernstein basis polynomial. Indicates joint velocity.

[0078] Furthermore, the expected path for each chromosome is calculated, and the mathematical definition of the expected path is as follows: As shown, the first generation population undergoes non-dominated sorting to find a set of Pareto fronts and is sorted according to crowding distance.

[0079] Where f1(t) is the total stroke time, f2(t) is the variance of the actuator torque, and τ i (t) and τ i (t-1) represents the torque of the actuator in the previous and current situations, respectively, and n represents the number of robot arm joints.

[0080] Finally, the parameters of the desired path are determined, thereby obtaining the target path. The movement of the robotic arm is controlled according to the target path, thereby driving the laser cleaning device to clean the dirt to be cleaned.

[0081] In some embodiments, the target path determination module includes a sorting group selection module, a direction-based crossover module, and an adaptive, precision-controllable mutation module.

[0082] The sorting group selection module is used to determine the feasible nodes of the target path based on a preset sorting group algorithm.

[0083] The direction-based crossing module is used to determine the optimal range of the target path based on the feasible nodes of the target path according to a preset direction crossing algorithm.

[0084] The adaptive precision controllable mutation module is used to improve the speed of determining the target path within the optimal range of the target path based on the preset adaptive mechanism.

[0085] In this embodiment, a new parent population is generated by a sorting group selection module, a direction-based crossover module, and an adaptive, precision-controllable mutation module, and the parent and offspring are combined into a population of N individuals according to an elite strategy.

[0086] It should be noted that during the ranking group selection, firstly, based on the constraints of the design variables, a parent population P0 of size N is randomly initialized, where N is set to a multiple of 4. Then, based on non-dominated ranking, the initialized population is ranked into multiple levels, and the fitness of the solution is equal to the corresponding non-dominated level. Individuals on the first front are assigned a fitness value of 1, individuals on the second front are assigned a fitness value of 2, and so on. Subsequently, the ranked population is evenly divided into 4 elements in sequence, namely X1, X2, X3, and X4. Finally, the paired individual groups formed by the ranking group selection (RGS) module are IA = (X1, X1, X1, X2, X2, X3) and IB = (X2, X3, X4, X3, X4). During the iteration process, IA is responsible for guiding the population to the optimal region, and IB is responsible for increasing the population diversity.

[0087] Furthermore, when performing direction-based cross-training, a direction-based cross-training (DBX) module was designed based on the principle that the better the objective function, the closer the individual is to the optimal region. in, Assume a center, where i represents the i-th individual, j represents the variable dimension, and the parameter r... ij A uniformly distributed random number within the interval [-1, 1]. This generates new individuals along random steps in the intersection direction; subsequently, if the population generated by the direction-based intersection module crosses the boundary, then... It will be confined within the boundaries to ensure the rationality of the population's genes, where p it p represents the value of the i-th individual in the i-th iteration. imin and p imax These represent the minimum and maximum values ​​of the design variables.

[0088] Furthermore, by incorporating an adaptive mechanism into the precision-controlled mutation, the convergence speed of the algorithm is improved. Firstly, the exploration and development of precision-controlled mutation can represent... Here, p is a parameter controlling the precision of the decision space. The function Random(p) can generate pseudo-random numbers in the range of 0 to p-1. Subsequently, an adaptive mechanism is added to form adaptive precision controllable mutation, as shown below:

[0089]

[0090]

[0091] in, and It represents the maximum and minimum values ​​of an individual in the contemporary population, ximin imax It represents the size of the decision space.

[0092] In some embodiments, the robotic arm trajectory planning module 3 further includes an optimization module 33, which is used to determine the most matching target path from multiple trajectories to be selected.

[0093] In this embodiment, the optimization module 33 determines whether the number of iterations has been reached and compares the objective functions of each trajectory. Finally, the Pareto solution of the objective function is obtained through iteration, and the corresponding design parameters are output.

[0094] Based on the aforementioned robotic arm equipped with a laser cleaning system, this invention also provides a cleaning method using a robotic arm equipped with a laser cleaning device. Please refer to [link to relevant documentation]. Figure 6 ,include:

[0095] S601. Obtain a dirt image in a large-format application scenario, and determine the position of the dirt image to be cleaned based on a preset coordinate transformation rule.

[0096] S602. Based on the position of the image of dirt to be cleaned, an improved elite non-dominated sorting genetic algorithm is used to determine the target path for the best-matching robotic arm to move from its initial position to the position of the image of dirt to be cleaned.

[0097] S603. Based on the position of the image of the dirt to be cleaned and the target path, the laser cleaning device is moved to the position of the image of the dirt to be cleaned by moving the most matching robotic arm and the dirt is cleaned.

[0098] In this embodiment, the location of the image of the dirt to be cleaned is first obtained. Then, an improved elite non-dominated sorting genetic algorithm is used to select the robotic arm in the robotic arm population that best matches the location of the image of the dirt to be cleaned. The target path for the robotic arm to move from the initial position to the location of the dirt image is determined. Finally, based on the location of the dirt image and the target path, the robotic arm carrying the laser cleaning device cleans the dirt to be cleaned, thus achieving the goal of efficient and pollution-free cleaning of dirt in large-format application scenarios.

[0099] The specific embodiments of the present invention described above do not constitute a limitation on the scope of protection of the present invention. Any other corresponding changes and modifications made in accordance with the technical concept of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A robotic arm equipped with a laser cleaning system, characterized in that, include: The components include a robotic arm, a laser cleaning device, and a controller, among which: The robotic arm is used to perform cleaning operations by having its end arm reach the location of the dirt to be cleaned, according to the instructions of the controller. The laser cleaning device is connected to the end of the robotic arm and is used to perform laser cleaning on the dirt to be cleaned according to the movement of the robotic arm. The controller is electrically connected to the robotic arm and the laser cleaning device. The controller is used to identify the dirt to be cleaned in large-format application scenarios and to obtain the location of the dirt to be cleaned. The controller is also used to control the most suitable robotic arm, equipped with a laser cleaning device, to move to the location of the dirt to be cleaned, based on the location of the dirt to be cleaned. The controller is also used to control the laser cleaning device to clean the dirt to be cleaned; The robotic arm equipped with the laser cleaning system also includes a vehicle-mounted device. The vehicle-mounted device includes a mobile platform, a lifting mechanism, and a retrieval device. The lifting mechanism is fixed to the platform of the mobile platform, and one end of the robotic arm is fixedly connected to the lifting mechanism so that the robotic arm can move up and down. The retrieval device is installed on the mobile platform. The controller includes an acquisition device for identifying dirt to be cleaned in large-format application scenarios and acquiring the location of the dirt. The acquisition device includes: The model training module is used to train the YOLOX model that incorporates a coordinated attention mechanism; An image acquisition module is used to acquire images of the dirt to be cleaned based on the improved YOLOX model. The target positioning module is used to determine the position of the image of the dirt to be cleaned based on a preset coordinate transformation rule; The controller also includes a robotic arm trajectory planning module, which, based on the position of the robotic arm and the position of the image of the dirt to be cleaned, uses an improved elite non-dominated sorting genetic algorithm to select a matching robotic arm to clean the corresponding dirt, and determines the path of the matching robotic arm from its initial position to the position of the image of the dirt to be cleaned. The robotic arm trajectory planning module includes an initial path determination module, a desired path determination module, and a target path determination module; The initial path determination module is used to initialize the parameters of the improved elite non-dominated sorting genetic algorithm and determine the initial path of the matched robotic arm movement. The desired path determination module is used to determine the desired path based on the Pareto front and the initial path. The target path determination module is used to determine the parameters of the desired path according to the elite strategy, thereby obtaining the target path of the matched robotic arm movement.

2. The robotic arm equipped with a laser cleaning system according to claim 1, characterized in that, The target path determination module includes a sorting group selection module, a direction-based crossover module, and an adaptive, precision-controllable mutation module. The sorting group selection module is used to determine the feasible nodes of the target path based on a preset sorting group algorithm. The direction-based crossing module is used to determine the optimal range of the target path based on the feasible nodes of the target path according to a preset direction crossing algorithm. The adaptive precision controllable mutation module is used to improve the speed of determining the target path within the optimal range of the target path based on the preset adaptive mechanism.

3. The robotic arm equipped with a laser cleaning system according to claim 1, characterized in that, The robotic arm trajectory planning module also includes an optimization module, which is used to determine the most matching target path from multiple trajectories to be selected.

4. The robotic arm equipped with a laser cleaning system according to claim 1, characterized in that, The initial path can be expressed by the following formula: , in, It is a binomial coefficient. Given control points for constructing the Bézier curve, T is the travel time, λ represents the normalized time, and the following conditions must be met: , It is a Bernstein basis polynomial. Indicates joint velocity.

5. The robotic arm equipped with a laser cleaning system according to claim 1, characterized in that, The desired path can be expressed by the following formula: , in, Total travel time The variance of the actuator torque, and These represent the torque of the actuator in the previous and current situations, respectively. This indicates the number of joints in the robotic arm.

6. A cleaning method based on a robotic arm equipped with a laser cleaning system as described in any one of claims 1-5, characterized in that, include: Acquire images of dirt in large-format application scenarios, and determine the position of the dirt image to be cleaned based on preset coordinate transformation rules; Based on the position of the image of dirt to be cleaned, an improved elite non-dominated sorting genetic algorithm is used to determine the target path for the best-matching robotic arm to move from its initial position to the position of the image of dirt to be cleaned. Based on the position of the image of the dirt to be cleaned and the target path, the laser cleaning device is moved to the position of the image of the dirt to be cleaned by moving the most matching robotic arm and the dirt is cleaned. The robotic arm equipped with the laser cleaning system also includes a vehicle-mounted device. The vehicle-mounted device includes a mobile platform, a lifting mechanism, and a retrieval device. The lifting mechanism is fixed to the platform of the mobile platform, and one end of the robotic arm is fixedly connected to the lifting mechanism so that the robotic arm can move up and down. The retrieval device is installed on the mobile platform. The controller includes an acquisition device for identifying dirt to be cleaned in large-format application scenarios and acquiring the location of the dirt. The acquisition device includes: The model training module is used to train the YOLOX model that incorporates a coordinated attention mechanism; An image acquisition module is used to acquire images of the dirt to be cleaned based on the improved YOLOX model. The target positioning module is used to determine the position of the image of the dirt to be cleaned based on a preset coordinate transformation rule; The controller also includes a robotic arm trajectory planning module, which, based on the position of the robotic arm and the position of the image of the dirt to be cleaned, uses an improved elite non-dominated sorting genetic algorithm to select a matching robotic arm to clean the corresponding dirt, and determines the path of the matching robotic arm from its initial position to the position of the image of the dirt to be cleaned. The robotic arm trajectory planning module includes an initial path determination module, a desired path determination module, and a target path determination module; The initial path determination module is used to initialize the parameters of the improved elite non-dominated sorting genetic algorithm and determine the initial path of the matched robotic arm movement. The desired path determination module is used to determine the desired path based on the Pareto front and the initial path. The target path determination module is used to determine the parameters of the desired path according to the elite strategy, thereby obtaining the target path of the matched robotic arm movement.

Images