Skylight production and assembly control method and system based on robot collaborative network

Abstract

The invention provides a skylight production and assembly control method and system based on a robot collaboration network, and relates to the technical field of intelligent control, and the method comprises the steps: obtaining a customized produced skylight part, carrying out the alignment preassembly, and obtaining a preassembled skylight part; starting the distributed collaborative robot; constructing a force cooperation control model, and obtaining cooperation pressure parameters and cooperation suction parameters; and the upper pressure robot conducts pressure control on the pre-assembled skylight part according to the cooperative pressure parameters, the lower suction robot conducts suction control on the pre-assembled skylight part according to the cooperative suction parameters, and the assembled skylight part is obtained. The technical problems that due to the fact that a global coordination mechanism is lacked when a plurality of robots work together, movement of the robots is not matched, and the assembly efficiency of the skylight is affected are solved, the dual-pressure assembly robots are arranged for cooperative work, high-precision and high-efficiency assembly of the skylight assembly is achieved, and the assembly efficiency is improved. And the product quality and the production efficiency of skylight assembly are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, and particularly to a skylight production and assembly control method and system based on a robot cooperation network. Background Art

[0002] The skylight structure is complex and involves various materials (such as glass, metal frames, rubber seals, etc.). Its assembly accuracy requirements are relatively high, and it is necessary to ensure the precise docking of the glass and the frame, the uniform fitting of the seals, and the stable connection of the power mechanism. With the development of industrial automation and intelligent manufacturing, robot technology has been introduced into the skylight assembly process. Usually, an independent control mode is adopted, that is, each robot performs operations according to its own tasks, lacking a global scheduling and information sharing mechanism, resulting in conflicting motion trajectories or asynchronous operations, and reducing the overall production efficiency.

[0003] In summary, in the prior art, there is a technical problem that due to the lack of a global coordination mechanism when multiple robots work together, it is easy to cause the robots to move mismatched, affecting the skylight assembly efficiency. Summary of the Invention

[0004] The purpose of this application is to provide a skylight production and assembly control method and system based on a robot cooperation network, so as to solve the technical problem in the prior art that due to the lack of a global coordination mechanism when multiple robots work together, it is easy to cause the robots to move mismatched, affecting the skylight assembly efficiency.

[0005] In view of the above problems, this application provides a skylight production and assembly control method and system based on a robot cooperation network.

[0006] In the first aspect, this application provides a skylight production and assembly control method based on a robot cooperation network. The skylight production and assembly control method based on a robot cooperation network is implemented through a skylight production and assembly control system based on a robot cooperation network. Among them, the skylight production and assembly control method based on a robot cooperation network includes: obtaining a customized skylight part, performing alignment pre-assembly according to the assembly position of the skylight part to obtain a pre-assembled skylight part; starting distributed cooperative robots, the distributed cooperative robots include an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; constructing a force cooperation control model, and obtaining a cooperation pressure parameter and a cooperation suction parameter according to the force cooperation control model; the upper pressure robot performs pressure control on the pre-assembled skylight part according to the cooperation pressure parameter, and the lower suction robot performs suction control on the pre-assembled skylight part according to the cooperation suction parameter to obtain a skylight part with assembly completed.

[0007] Optionally, the force cooperation control model is connected to the assembly path planning module. The visual extraction unit collects the assembly position image of the pre-assembled skylight part to obtain an assembly image. Pixel recognition is performed on the assembly image to obtain a first pixel data set and a second pixel data set. The first unit coverage pixel data corresponding to the upper pressure robot and the second unit coverage pixel data corresponding to the lower suction robot are respectively obtained. The assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data to obtain an assembly planning path. The upper pressure robot and the lower suction robot are controlled according to the assembly planning path.

[0008] Optionally, the assembly path planning module separates the first pixel data set by using the first unit coverage pixel data as a pixel step to obtain a first set of path nodes; the assembly path planning module separates the second pixel data set by using the second unit coverage pixel data as a pixel step to obtain a second set of path nodes; trajectory connection is performed on the first set of path nodes to obtain a first planning path, trajectory connection is performed on the second set of path nodes to obtain a second planning path, and the first planning path and the second planning path are combined to generate an assembly planning path, where the first planning path and the second planning path are not synchronized.

[0009] Optionally, according to the first unit coverage pixel data and the second unit coverage pixel data, average pixel data is determined; the first pixel data set and the second pixel data set are separated by using the average pixel data as a pixel step to obtain a first set of path nodes and a second set of path nodes; trajectory connection is performed on any one set of path nodes in the first set of path nodes or the second set of path nodes according to a preset trajectory to obtain a synchronous planning path, and an assembly planning path is generated according to the synchronous planning path.

[0010] Optionally, the upper pressure robot includes a six-degree-of-freedom pressurizing robotic arm, a pressurizing platform, a pressure control module, and a force sensor; the cooperative pressure parameter is sent to the six-degree-of-freedom pressurizing robotic arm, and the pressure control module drives the pressurizing platform to perform pressure control on the pre-assembled skylight part by controlling the six-degree-of-freedom pressurizing robotic arm; pressure monitoring is performed by the force sensor, and the obtained pressure sensing data is uploaded to the pressure control module, and force feedback is performed on the six-degree-of-freedom pressurizing robotic arm based on the pressure control module.

[0011] Optionally, the lower suction robot includes a negative pressure generator, a vacuum adsorption platform, and a suction control module; the cooperative suction parameter is sent to the suction control module, and the suction control module controls the suction of the suction cups on the vacuum adsorption platform according to the negative pressure generator.

[0012] Optionally, the suction cups on the vacuum adsorption platform are arranged in a honeycomb array.

[0013] Optionally, obtain the force sample data set of the distributed cooperative robot, where the force sample data set includes the pressure sample data set of the upper pressure robot and the suction sample data set of the lower suction robot, the difference data set corresponding to the pressure sample data set and the suction sample data set, and the label information characterizing the assembly error; define the parameters of the impedance control model, including the initial stiffness, initial damping, and initial mass parameters of the upper pressure robot and the lower suction robot; perform model training on the defined impedance control model according to the force sample data set, and when the model converges, output the force cooperative control model.

[0014] Optionally, collect the material information and geometric information of the skylight part; perform force protection analysis according to the material information and geometric information, output the force protection threshold, use the force protection threshold as a constraint condition for training the impedance control model, and obtain the force cooperative control model.

[0015] In a second aspect, the present application also provides a skylight production and assembly control system based on a robot cooperation network, which is used to execute the skylight production and assembly control method based on a robot cooperation network as described in the first aspect. The skylight production and assembly control system based on a robot cooperation network includes: a skylight part acquisition module, which is used to acquire a customized skylight part, align and pre-assemble it according to the assembly position of the skylight part, and obtain a pre-assembled skylight part; a robot startup module, which is used to start a distributed cooperative robot, and the distributed cooperative robot includes an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; a control model construction module, which is used to construct a force cooperative control model and obtain cooperative pressure parameters and cooperative suction parameters according to the force cooperative control model; a skylight part assembly module, which is used for the upper pressure robot to perform pressure control on the pre-assembled skylight part according to the cooperative pressure parameters, and the lower suction robot to perform suction control on the pre-assembled skylight part according to the cooperative suction parameters, and obtain a skylight part with assembly completed.

[0016] One or more technical solutions provided in the present application have at least the following beneficial effects: By obtaining customized skylight parts, aligning and pre-assembling them according to the assembly positions of the skylight parts to obtain pre-assembled skylight parts; starting distributed collaborative robots, the distributed collaborative robots include an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; constructing a force collaborative control model, and obtaining collaborative pressure parameters and collaborative suction parameters according to the force collaborative control model; the upper pressure robot performs pressure control on the pre-assembled skylight parts according to the collaborative pressure parameters, and the lower suction robot performs suction control on the pre-assembled skylight parts according to the collaborative suction parameters to obtain the assembled skylight parts. That is to say, by setting up dual-pressure assembly robots, with a pressure robot distributed above the skylight and a suction robot distributed below the skylight, after the skylight is aligned, the two robots flexibly cooperate for collaborative assembly, achieving precise coordination and improving the product quality and production efficiency of skylight assembly.

[0017] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specifically gives the specific implementation manners of this application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of this application, nor is it used to limit the scope of this application. Other features of this application will become easily understandable through the following description. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0019] Figure 1 It is a schematic flow chart of the skylight production and assembly control method based on the robot collaboration network of this application; Figure 2 It is a schematic structural diagram of the skylight production and assembly control system based on the robot collaboration network of this application.

[0020] Explanation of the reference numerals: skylight part acquisition module 11, robot startup module 12, control model construction module 13, skylight part assembly module 14. Detailed Description of the Invention

[0021] By providing a skylight production and assembly control method and system based on a robot collaboration network, the present application solves the technical problem in the prior art that due to the lack of a global coordination mechanism when multiple robots work together, the robot movements are prone to mismatch, affecting the skylight assembly efficiency. By setting a double-pressure assembly robot, with a pressure robot distributed above the skylight and a suction robot distributed below the skylight, after the skylight is aligned, the two robots flexibly cooperate for collaborative assembly, achieving precise coordination and improving the product quality and production efficiency of skylight assembly.

[0022] Next, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application. Additionally, it should be noted that for the sake of description, only the parts related to the present application are shown in the accompanying drawings rather than all.

[0023] Embodiment 1, please refer to the attached Figure 1 , the present application provides a skylight production and assembly control method based on a robot collaboration network. Among them, the skylight production and assembly control method based on a robot collaboration network is executed by a skylight production and assembly control system based on a robot collaboration network. The skylight production and assembly control method based on a robot collaboration network specifically includes the following steps: S100: Obtain the customized skylight parts, align and pre-assemble them according to the assembly positions of the skylight parts to obtain pre-assembled skylight parts.

[0024] Specifically, the skylight components are customized according to the specific vehicle models and customer requirements in the order. Different vehicle models require different specifications of skylights. Therefore, it is necessary to determine the current skylight parts to be produced. Some of these skylight parts will carry data on the assembly positions during the manufacturing process. The skylight parts are the main components of the skylight, including glass, metal frames, sealing strips, motors, etc. During the assembly process, these components need to be precisely matched to ensure the final airtightness and stability. Transport the skylight parts to the assembly line, move the skylight parts to the assembly positions for alignment and preliminary fixation to obtain pre-assembled skylight parts. That is to say, place the skylight parts at the assembly positions and use temporary fixing devices for preliminary fixation. Through pre-assembly, ensure the precise alignment of the skylight parts, improve the assembly accuracy and efficiency, and lay a foundation for the subsequent formal assembly.

[0025] S200: Start the distributed collaborative robots. The distributed collaborative robots include an upper pressure robot and a lower suction robot. The upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture.

[0026] Specifically, a double-pressure assembly robot is set up, including an upper-pressure robot and a lower-suction robot, one distributed above the sunroof and the other distributed below the sunroof. After the sunroof is aligned, the two robots cooperate flexibly to perform force collaboration assembly. Start the distributed collaboration robot, that is, make the two robots cooperate with each other through a distributed control system to jointly execute complex tasks. The robot control center reads the specification data of the sunroof part and assigns tasks to the two collaborative robots, including the upper-pressure robot and the lower-suction robot. The upper-pressure robot is responsible for applying pressure from above to ensure that the sunroof part fits tightly with the vehicle body or other components, and the lower-suction robot provides adsorption and fixation from below to ensure that the sunroof part does not slide or shift during the assembly process.

[0027] Through a distributed control architecture, synchronously control the distributed collaboration robot to ensure that the actions of the two robots are coordinated when performing tasks. The central control center sends synchronous control instructions to the two robots. The upper-pressure robot starts to apply pressure downward according to the instructions, and the lower-suction robot simultaneously applies suction upward. Real-time monitor the actions and applied forces of the two robots through force sensors and other sensors to ensure that they operate according to the preset parameters. The upper-pressure robot and the lower-suction robot apply pressure and suction simultaneously according to the instructions to ensure that the sunroof part is assembled onto the vehicle body smoothly and precisely.

[0028] Exemplarily, different robots are used for the sunroof test assembly, and the test data is as follows: the alignment error of traditional single-robot assembly is ±1.2 mm, the assembly success rate is 87%, the installation time is 10 s, and the maximum force is 550 N; the alignment error of the distributed collaboration robot without synchronous control is ±0.8 mm, the assembly success rate is 92%, the installation time is 7 s, and the maximum force is 520 N; the alignment error of the distributed collaboration robot with synchronous control is ±0.2 mm, the assembly success rate is 99.5%, the installation time is 5 s, and the maximum force is 490 N. The test results show that using the distributed collaboration robot improves the assembly accuracy by 83% compared with single-robot assembly, shortens the installation time by 50%, reduces the risk of glass breakage, and improves the assembly stability.

[0029] Furthermore, the present application S200 includes: The force cooperation control model is connected to the assembly path planning module. The visual extraction unit collects the assembly position image of the pre-assembled skylight part to obtain the assembly image. Pixel recognition is performed on the assembly image to obtain a first pixel data set and a second pixel data set. The first unit coverage pixel data corresponding to the upper pressure robot and the second unit coverage pixel data corresponding to the lower suction robot are respectively obtained. The assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data to obtain the assembly planning path. The upper pressure robot and the lower suction robot are controlled according to the assembly planning path.

[0030] Specifically, the force cooperation control model is mainly responsible for calculating and adjusting the forces applied by the robots in real time during the assembly process (such as the pressure applied by the upper pressure robot and the adsorption force generated by the lower suction robot), ensuring that the skylight part is evenly stressed during assembly and will not cause displacement or damage due to excessive or insufficient local forces. The assembly path planning module uses the data collected by vision to analyze the edge pixel data of the pre-assembled skylight part, and according to the unit coverage pixel data of the robots, plans an optimal assembly path, which includes both the geometric trajectory of the robot movement and the assembly accuracy requirements that should be achieved at each key node.

[0031] The visual extraction unit first collects the pre-assembly position image of the skylight part, and through edge detection and pixel recognition, generates the upper surface edge (the first pixel data set) and the lower surface edge (the second pixel data set) respectively. At the same time, the unit coverage pixel data of the robot end (the upper pressure robot and the lower suction robot) is obtained. The assembly path planning module uses the unit coverage pixel data to divide the first pixel data set and the second pixel data set according to the preset pixel step size, and generates two sets of discrete path nodes respectively. The trajectories of these two sets of nodes are connected respectively to generate the first planning path and the second planning path.

[0032] The visual extraction unit collects the image of the assembly position of the pre-assembled skylight part to obtain the assembly image of the pre-assembled skylight part. The visual extraction unit is a visual sensing unit for obtaining image information, including devices such as industrial cameras, 3D laser scanners, and depth cameras. The image of the skylight part is collected through an industrial camera or a 3D laser scanner. The assembly image refers to the visual data of the skylight assembly area captured by the visual extraction unit, including key information such as the skylight glass, the installation base, and the sealing strip.

[0033] Use an image processing algorithm to perform pixel recognition on the assembly image, and extract the first pixel dataset (the edge pixels of the upper surface of the skylight assembly) and the second pixel dataset (the edge pixels of the lower surface of the skylight assembly) respectively. First, preprocess the image to eliminate noise and irrelevant information in the image and enhance the edge features to facilitate subsequent pixel recognition. Use an edge detection algorithm (such as Canny edge detection) to identify the edges of the upper and lower surfaces of the skylight part. Classify the identified edge pixels into the first pixel dataset (upper surface edge) and the second pixel dataset (lower surface edge).

[0034] According to the working range of the robot and the assembly requirements, determine the respective operating areas of the upper and lower robots. Calculate the number of pixels corresponding to each unit area based on the area of the operating area and the resolution of the image. Obtain the first unit coverage pixel data corresponding to the upper pressure robot, that is, the pixel data corresponding to each unit area within the operating area of the upper pressure robot. Obtain the second unit coverage pixel data corresponding to the lower suction robot, that is, the pixel data corresponding to each unit area within the operating area of the lower suction robot.

[0035] The assembly path planning module separates the first pixel dataset and the second pixel dataset according to the unit coverage pixel data, and then generates an assembly planning path based on the generated path nodes. Here, a synchronous or asynchronous strategy can be selected according to requirements. In the synchronous strategy, the upper and lower robots execute the assembly actions simultaneously along the same path. In the asynchronous strategy, the lower suction robot can start first to complete the adsorption positioning, and the upper pressure robot starts with a delay to apply pressure, avoiding mutual interference.

[0036] Taking the asynchronous strategy as an example, use the first unit coverage pixel data as the step size to segment the first pixel dataset to obtain the first set of path nodes, which represent the key positions where the upper pressure robot needs to stop or operate during the assembly process. Similarly, the assembly path planning module uses the first unit coverage pixel data as the pixel step size to separate the second pixel dataset to obtain the second set of path nodes, which represent the key positions where the lower suction robot needs to stop or operate during the assembly process. Adopt a trajectory connection technique, such as Bezier curve fitting, to connect the first set of path nodes into a continuous and smooth trajectory, that is, the first planning path; similarly, connect the second set of path nodes into the second planning path. Finally, combine the two paths to generate a complete assembly planning path. The first planning path and the second planning path are asynchronous, which means that the two robots will operate according to different schedules and paths during the assembly process.

[0037] The obtained assembly planning path is transmitted to the robot control center and sent to the corresponding upper pressure robot and lower suction robot. The robots execute the assembly tasks according to their respective motion instructions to ensure that the upper and lower surfaces of the skylight part are accurately aligned and the force is balanced during the assembly process. For example, after generating the assembly planning path through the preset trajectory planning algorithm, the test results show that: when executed synchronously, the average assembly error of the upper and lower robots is controlled within ±0.25 mm; when using the asynchronous strategy (the lower one goes first and the upper one delays for 0.5 seconds), the overall assembly success rate reaches 99.5%, and the glass breakage rate is reduced to 0.1%. Through the vision extraction unit and the assembly path planning module, the high-precision and automated assembly of the skylight part can be realized, significantly improving the assembly accuracy and speed, and the efficiency and quality of the skylight production line can be significantly improved.

[0038] Furthermore, the present application further includes the following steps: The assembly path planning module separates the first pixel data set by using the first unit coverage pixel data as the pixel step to obtain a first set of path nodes; the assembly path planning module separates the second pixel data set by using the second unit coverage pixel data as the pixel step to obtain a second set of path nodes; performs trajectory connection on the first set of path nodes to obtain a first planning path, performs trajectory connection on the second set of path nodes to obtain a second planning path, and generates an assembly planning path from the first planning path and the second planning path, wherein the first planning path and the second planning path are asynchronous.

[0039] Specifically, the assembly path planning module first receives the image data of the upper surface (the first pixel data set) and the lower surface (the second pixel data set) collected from the vision extraction unit. Using the first unit coverage pixel data (such as 50 pixels) as the pixel step, the first pixel data set is separated to obtain a first set of path nodes. For example, if the upper surface edge is 1000 pixels long in the image, about 20 discrete path nodes will be obtained after segmentation at a 50-pixel step. The second unit coverage pixel data (such as 40 pixels) is used as the pixel step to separate the second pixel data set to obtain a second set of path nodes. The path nodes refer to the discrete key position points obtained by segmenting the image edge data according to the preset step, and these nodes can reflect the geometric features of the skylight edge and provide a basis for subsequent trajectory fitting. The first set of path nodes includes the key position points extracted from the first pixel data set, and the second set of path nodes includes the key position points extracted from the second pixel data set.

[0040] Connect the trajectory of the first set of path nodes to form the first planned path. Usually, a trajectory connection algorithm is used to smoothly connect discrete path nodes into a continuous motion trajectory. For example, Bezier curve fitting and spline curve interpolation are used to convert the discontinuity between nodes into a smooth curve, reducing sharp turns and vibrations during motion. Bezier curve fitting has the characteristics of smoothness and easy control and is often used in robot motion trajectory planning, making the motion path have better continuity and flexibility. Spline curve interpolation uses spline functions (such as cubic spline interpolation) to interpolate discrete data points to generate a smooth curve.

[0041] According to the specific assembly requirements and the distribution of nodes, select an appropriate smoothing algorithm. Taking the Bezier curve as an example, select the starting point, ending point, and several intermediate control points according to the first set of path nodes, and then use the Bezier curve formula to calculate the coordinates of each point on the entire curve. For example, assume the selected nodes are P 0 , P 1 , ……, P 19 . By adjusting the control points, a curve can be generated so that the curve passes through or is close to the original nodes at each node as much as possible. The obtained smooth curve is used as the first planned path to guide the movement of the upper pressure robot. For example, in an automotive sunroof assembly experiment, an industrial camera is used to collect images of the upper surface of the sunroof. After image processing, the pixel data of the upper surface edge is obtained, and 20 discrete nodes are separated at a step of 50 pixels. Using the Bezier curve fitting algorithm, the first planned path generated shows that the average path error is ±0.2 mm (the deviation from the actual edge measurement value), the motion curve is smooth, and the curve fitting degree reaches more than 95%.

[0042] Similarly, in a similar manner to the above steps, connect the trajectory of the second set of path nodes to generate a continuous second planned path. The first planned path is used to guide the upper pressure robot to perform the force application action, and the second planned path is used to guide the lower suction robot to perform the adsorption and positioning action. Since the two sets of data are from different surfaces (the upper surface and the lower surface) of the sunroof, there may be slight differences in the shape and node distribution of the two generated planned paths. Combine the first planned path and the second planned path to form a complete assembly planned path. The first planned path and the second planned path are not synchronized. During the assembly execution process, different time sequences or starting moments are adopted, and they do not move completely simultaneously, thus avoiding mutual interference when the two robots operate.

[0043] The method of asynchronous planning mainly separates path nodes independently from the pixel data of the upper surface and the lower surface respectively. After generating the planned paths respectively, they are combined into an assembly planning path. Since the two paths are independent when formed, an asynchronous strategy is adopted. This method is convenient for optimizing the paths for the upper and lower different surfaces respectively, but it is necessary to handle the time coordination problem between the two paths to avoid interference and assembly errors.

[0044] Exemplarily, in the sunroof assembly test, the visual extraction unit collects that the length of the edge data of the upper surface is 1000 pixels, and the length of the edge data of the lower surface is 800 pixels. For the upper surface data, the first unit covers the pixel data with a 50-pixel step length, and about 20 path nodes are obtained. For the lower surface data, the second unit covers the pixel data with a 40-pixel step length, and about 20 path nodes are obtained. After using the Bezier curve fitting algorithm, the average assembly error of the first planned path is measured to be ±0.25 mm, while the average error of the second planned path is about ±0.3 mm. The lower suction robot starts first to complete adsorption, and after a delay of 0.5 seconds, the upper pressure robot starts to apply pressure. After actual testing, the overall assembly success rate reaches 99.2%, and due to the asynchronous strategy, the local stress concentration caused by synchronous force application is effectively reduced, and the glass breakage rate is reduced from the traditional 1.5% to 0.1%. By adopting an asynchronous execution strategy, the upper pressure robot and the lower suction robot perform asynchronously during the assembly process, thus effectively avoiding mechanical interference and operation conflicts and ensuring assembly stability.

[0045] Furthermore, the present application further includes the following steps: Determine the mean pixel data according to the first unit covering pixel data and the second unit covering pixel data; use the mean pixel data as the pixel step length to separate the first pixel data set and the second pixel data set to obtain a first set of path nodes and a second set of path nodes; connect the trajectories of any one set of path nodes in the first set of path nodes or the second set of path nodes according to a preset trajectory to obtain a synchronous planning path, and generate an assembly planning path according to the synchronous planning path.

[0046] Specifically, calculate the average value of the first unit covering pixel data and the second unit covering pixel data to obtain the mean pixel data. Use the mean pixel data as the pixel step length to separate the first pixel data set and the second pixel data set to obtain a first set of path nodes and a second set of path nodes. Different from the foregoing, the separation step lengths of the two pixel sets here are the same. The mean pixel data as a unified pixel step length can not only take into account the working characteristics of the upper and lower parts, but also ensure the scale consistency when the two sets of data are segmented.

[0047] In robot assembly path planning, a preset trajectory refers to the robot's motion path preset during the design phase, usually based on the shape, size, and assembly requirements of the skylight part, as well as the robot's motion capabilities and accuracy. According to the shape, size, and assembly requirements of the skylight part, analyze the trajectory that the robot needs to follow during assembly. Based on the analysis results, design one or more preset trajectories, which can be straight lines, curves, arcs, etc., depending on the specific assembly requirements. Before actual application, verify the preset trajectory to ensure its feasibility and effectiveness in the actual assembly process.

[0048] Select either the first set of path nodes or the second set of path nodes and connect them according to the preset trajectory to obtain a synchronized planning path. Due to the use of a unified step size segmentation, the generated path has good continuity and consistency, and the robot can move strictly synchronously along this path, ensuring that the upper and lower assembly actions are consistent. According to the synchronized planning path, generate the final assembly planning path and transmit it to the robot control center, so that the upper pressure robot and the lower suction robot operate synchronously according to this path, ensuring the synchrony of the assembly process. The final motion trajectory generated according to the synchronized planning path is used to guide the robot to move precisely during assembly, ensuring that the upper and lower edges of the skylight part are aligned consistently.

[0049] The synchronized planning method calculates the mean of the first and second unit coverage pixel data, unifies the pixel step size, divides the two pixel data sets, and only selects one set of nodes for trajectory connection to generate a single synchronized planning path. This method simplifies the path planning process, enables the robot to move strictly synchronously, thereby improving the assembly consistency and the stability of the overall system, and is suitable for occasions requiring a high degree of consistency in the assembly process. Using the mean pixel data to unify the step size and separating the upper and lower surface data sets can ensure that the distribution of the two sets of path nodes is consistent, so that the generated synchronized planning path can keep the robot at a high precision during motion, and the two robots can work more closely together, thereby improving the performance of the overall assembly line.

[0050] Furthermore, this application also includes the following steps: The upper pressure robot includes a six-degree-of-freedom pressurizing robotic arm, a pressurizing platform, a pressure control module, and a force sensor; send the collaborative pressure parameter to the six-degree-of-freedom pressurizing robotic arm, and the pressure control module drives the pressurizing platform to control the pressure on the pre-assembled skylight part by controlling the six-degree-of-freedom pressurizing robotic arm; the force sensor monitors the pressure, uploads the obtained pressure sensing data to the pressure control module, and performs force feedback on the six-degree-of-freedom pressurizing robotic arm based on the pressure control module.

[0051] Specifically, the upper pressure robot is a robot used to apply pressure above the skylight during the skylight assembly process. It mainly consists of a six-degree-of-freedom pressurizing robotic arm, a pressurizing platform, a pressure control module, and a force sensor. The six-degree-of-freedom pressurizing robotic arm is a robotic arm with six degrees of freedom (i.e., translation in three spatial directions and three rotation axes), that is, translation around the X, Y, and Z axes and rotation around the X, Y, and Z axes, which can achieve flexible posture adjustment and positioning. The pressurizing platform is usually installed at the end of the robotic arm and is used to convert the motion output by the robotic arm into uniform pressure on the skylight part. The pressure control module is responsible for receiving and processing pressure parameters from the host computer or the collaborative system, controlling the motion of the robotic arm, and enabling the pressurizing platform to apply a predetermined pressure. The force sensor is installed on the pressurizing platform or the robotic arm and is used to monitor the actual pressure applied to the skylight part in real time and feedback the data to the pressure control module.

[0052] The robot control center sends the collaborative pressure parameters obtained according to the requirements of the assembly task and the force collaboration control model to the six-degree-of-freedom pressurizing robotic arm and the pressure control module. The pressure control module sends control instructions to the six-degree-of-freedom pressurizing robotic arm according to the collaborative pressure parameters. The six-degree-of-freedom pressurizing robotic arm drives the pressurizing platform to perform pressure control on the pre-assembled skylight part. That is to say, driven by the control instructions, the six-degree-of-freedom pressurizing robotic arm drives the pressurizing platform to apply pressure to the pre-assembled skylight part to ensure that the skylight part fits tightly with the assembly base or other components. At the same time, the force sensor starts to monitor the actual pressure value applied to the pre-assembled skylight part in real time to obtain pressure sensing data.

[0053] The monitored pressure sensing data is uploaded to the pressure control module for pressure feedback. The pressure control module compares the actual pressure value with the target pressure parameters. If there is a deviation between the actual pressure and the target, the motion of the robotic arm is adjusted through closed-loop control, thereby realizing force feedback control to ensure that the finally applied pressure is consistent with the collaborative pressure parameters. Through closed-loop control, the upper pressure robot can ensure that precise and stable pressure is applied to the skylight part during the assembly process, improving the assembly accuracy and reliability. It not only improves production efficiency but also reduces human errors and enhances the overall quality of the product.

[0054] Furthermore, the present application further includes the following steps: The lower suction robot includes a negative pressure generator, a vacuum adsorption platform, and a suction control module; the collaborative suction parameters are sent to the suction control module, and the suction control module controls the suction of the suction cups on the vacuum adsorption platform according to the negative pressure generator.

[0055] The suction cups on the vacuum adsorption platform are arranged in a honeycomb array.

[0056] Specifically, the lower suction robot is a robot that fixes the pre-assembled skylight parts by adsorption force from below during the skylight assembly process. It mainly consists of a negative pressure generator, a vacuum adsorption platform, and a suction control module. The negative pressure generator is a device that can generate negative pressure (vacuum) and is used to provide suction force for the vacuum adsorption platform. The negative pressure generator reduces the atmospheric pressure to form a negative pressure environment required for adsorption. The vacuum adsorption platform is a working platform installed on the lower suction robot, and suction cups are installed on it for directly contacting and fixing the skylight parts. The suction control module is responsible for receiving the collaborative suction parameters and precisely controlling the suction cups on the vacuum adsorption platform to generate appropriate suction force according to the negative pressure provided by the negative pressure generator. It usually receives external instructions and adjusts the suction force generated by the negative pressure generator. On the vacuum adsorption platform, the suction cups are arranged in a honeycomb pattern, so that the suction cups are evenly distributed within the assembly area, increasing the contact area, improving the adsorption uniformity and stability, while dispersing the local negative pressure, making the overall adsorption effect better, and also facilitating flexibility when adsorbing skylight parts of different shapes and sizes.

[0057] The collaborative suction parameters are obtained according to the assembly task and the force collaboration control model, and the collaborative suction parameters are sent to the suction control module of the lower suction robot. After receiving the collaborative suction parameters, the suction control module issues an instruction to the negative pressure generator to adjust the negative pressure output. After receiving the instruction, the negative pressure generator starts to work, generates negative pressure and transmits the negative pressure to the suction cups on the vacuum adsorption platform through a pipeline. The suction cups on the vacuum adsorption platform are arranged in a honeycomb array, and this structure ensures that the suction cups are evenly distributed throughout the assembly area and can provide balanced suction force. When the suction force reaches the set value, the suction cups start to adsorb the skylight parts. The suction control module ensures that the suction cups adsorb the skylight parts with uniform suction force, avoiding deformation or damage of the skylight parts caused by uneven suction force. Once the skylight parts are firmly adsorbed, the lower suction robot can work in cooperation with the upper pressure robot and operate according to the assembly planning path to complete the assembly of the skylight parts.

[0058] The suction control module monitors the adsorption state of the suction cups, adjusts the output of the negative pressure generator according to the feedback information, ensures that the adsorption force of each suction cup reaches the set target, and then stably fixes the pre-assembled skylight part. The actual negative pressure data collected is fed back to the suction control module to achieve closed-loop control and ensure that the actual suction force is consistent with the cooperative suction force parameters. For example, on a certain automotive skylight assembly line, the lower suction robot generates negative pressure through a negative pressure generator and distributes it to the suction cups arranged in a honeycomb pattern on the vacuum adsorption platform. The cooperative suction force parameter is set to keep the negative pressure at the suction cup at least 60 kPa. After receiving the target suction force parameter, the suction control module adjusts the output of the negative pressure generator through the PID algorithm. In the experiment, the negative pressure was 55 kPa at the initial start-up and stabilized within the range of 60 ± 1 kPa after closed-loop adjustment. After adopting the honeycomb suction cup arrangement, the adsorption uniformity of the entire adsorption area is improved, ensuring that the skylight part will not be displaced due to poor local adsorption during the assembly process. The results show that the data transmission delay is less than 10 ms and the control response time is 0.3 s. Through precise suction control, the lower suction robot can closely cooperate with the upper pressure robot to achieve high-precision assembly of the skylight part, which not only improves production efficiency but also reduces the uncertainty of manual operation and ensures the consistency of product quality.

[0059] Furthermore, step S300 of the present application includes: Obtain the force sample data set of the distributed collaborative robots, where the force sample data set includes the pressure sample data set of the upper pressure robot and the suction sample data set of the lower suction robot, the difference data set corresponding to the pressure sample data set and the suction sample data set, and the label information characterizing the assembly error; define the parameters of the impedance control model, including the initial stiffness, initial damping, and initial mass parameters of the upper pressure robot and the lower suction robot; perform model training on the defined impedance control model according to the force sample data set, and when the model converges, output the force collaborative control model.

[0060] Specifically, obtaining the force sample data set of the distributed robots includes the pressure sample data set of the upper pressure robot and the suction sample data set of the lower suction robot, that is, the data set of the pressure values (in Newtons) applied by the upper pressure robot during the assembly process, and the data set of the suction force generated by the lower suction robot during adsorption during the assembly process. In addition, it also includes the difference data set between the pressure sample data set and the suction sample data set, which reflects the collaborative difference in force between the two robots during the assembly process. The label information characterizing the assembly error refers to quantifying and annotating the errors generated during the assembly process (such as alignment errors, deformations caused by uneven force, etc.) to identify the assembly error situation corresponding to the force sample data set, such as assembly deviation, assembly speed, etc.

[0061] Utilize the force sensors built into each robot (such as the force sensor of the upper pressure robot and the negative pressure sensor of the lower suction robot) to record the applied pressure and adsorption suction force in real time during the assembly process. Preprocess the collected raw data, including filtering, normalization, and abnormal data removal. Calculate the difference between the corresponding data of the upper and lower robots (such as subtracting the suction force value from the pressure value or calculating the ratio of the two) to generate difference data reflecting the coordination degree between the two. Label each data sample according to the assembly error detection (such as using vision to detect assembly deviation) for supervised learning.

[0062] Define initial stiffness, damping, and mass parameters for the upper pressure robot and the lower suction robot respectively. The initial stiffness affects the degree of response of the robot to force, and a higher stiffness will cause the robot to decelerate rapidly when encountering resistance. The initial damping is used to reduce the oscillation of the system and improve stability. The initial mass affects the dynamic response of the robot, and a larger mass will cause the robot to accelerate and decelerate more slowly. The impedance control model is a control strategy aimed at enabling the robot to have good compliance and force control performance when contacting the workpiece. Its core idea is to regard the robot system as a mass-damping-stiffness system and achieve an active response to external forces by adjusting model parameters (such as stiffness, damping, and mass), so as to achieve stable assembly and precise force application. The initial stiffness, initial damping, and initial mass parameters are the initial parameters of the impedance control model, used to define the initial dynamic characteristics of the robot.

[0063] Establish an impedance control model and train the impedance control model using the force sample dataset (including pressure samples, suction samples, difference data, and label information). Use a deep learning framework to construct a regression model, take the model parameters (stiffness, damping, mass) as variables to be optimized, and continuously iterate through gradient descent (or Adam optimizer) to minimize the mean square error (MSE) between the predicted value and the actual assembly error. Use the mean square error (MSE) as the loss function to calculate the difference between the predicted assembly error of the model and the actual assembly error. The goal is to minimize the MSE, that is, the optimized model can accurately predict the assembly error. In each training step, calculate the gradient of the loss function with respect to the model parameters (including stiffness, damping, mass) through the backpropagation algorithm, and then update the parameters to reduce the loss. The training process is repeated continuously until the loss function drops to a preset convergence threshold, indicating that the model parameters have stabilized and the model has converged.

[0064] When the training process meets the convergence conditions (such as the MSE is lower than a certain threshold or the training loss changes very little for several consecutive epochs), output the trained force collaboration control model. This model can adjust the collaborative force between the upper pressure robot and the lower suction robot in real time during the actual assembly process to achieve precise and dynamic force feedback control.

[0065] For example, in a car sunroof assembly test, the following data were collected by force sensors: upper pressure robot pressure sample data: [490, 495, 500, 505, 510] N, lower suction robot suction sample data: [58, 59, 60, 61, 62] kPa, difference data set (after unit conversion): [432, 435, 440, 445, 448] N, assembly error label (obtained by visual inspection): [0.3, 0.25, 0.2, 0.22, 0.2] mm. The initial impedance parameters were set as follows: upper: stiffness 1000 N / m, damping 50 N·s / m, mass 200 kg, lower: stiffness 800 N / m, damping 40 N·s / m, mass 150 kg. After 2000 trainings, the mean square error of the model on the validation set dropped to ±0.05 mm, indicating that the model has converged. Through the training and adjustment of the impedance control model, the stability and accuracy of the robot assembly process can be achieved, and the assembly accuracy and efficiency can be improved.

[0066] Furthermore, the present application also includes the following steps: Collect material information and geometric information of the skylight component; perform force protection analysis based on the material information and geometric information, output a force protection threshold, use the force protection threshold as a constraint condition for training the impedance control model, and obtain a force cooperation control model.

[0067] Specifically, the material properties (such as elastic modulus, yield strength, fracture toughness, etc.) and geometric parameters (such as thickness, size, curvature, etc.) of the skylight components are obtained using sensors, industrial cameras, laser scanners and other equipment as well as product design documents. Material information refers to the physical and chemical property data of the skylight component, such as material type (such as steel, aluminum, glass, plastic, etc.), strength, elastic modulus, fracture toughness, etc., which determine the bearing capacity and safety range of the skylight component when subjected to stress. Geometric information refers to the geometric parameters of the skylight component, such as size, shape, thickness, area, etc., which directly affect the stress distribution and deformation characteristics of the skylight component when subjected to stress.

[0068] Use finite element software (such as ANSYS, ABAQUS) to simulate the skylight parts, and analyze the stress state based on material and geometric information. In other words, input the material information and geometric information of the skylight parts into the existing finite element software for simulation, and simulate the assembly process of the skylight parts. Determine the maximum force value allowed during the assembly or stress process of the skylight parts through the simulation results, and evaluate the maximum and minimum forces that the skylight parts can safely withstand in different assembly steps to avoid damage. For example, calculate that the safe working force of the skylight parts is 500N (or other units) under the maximum allowable stress. Use this safe working force as the force protection threshold and as a constraint in the subsequent training process to ensure that the model prediction and control output do not exceed this threshold during the optimization process.

[0069] During the model training process, the force protection threshold is used as a constraint condition to design a loss function or a regularization term, so that when the model adjusts the stiffness, damping, and mass parameters, no prediction output exceeding the safety threshold will be generated. Then, according to the training process introduced in detail above, the model parameters are continuously adjusted to match the prediction output with the actual assembly error by minimizing the mean square error, while satisfying the force protection threshold constraint. When the training loss reaches the preset convergence criterion and all prediction outputs meet the safety constraints, the final force collaboration control model is output. Through training, the model will learn how to coordinate the force outputs of the two robots to complete the assembly task without exceeding the force protection threshold. The model obtained after training convergence is the force collaboration control model, and the collaborative pressure parameter and the collaborative suction parameter can be obtained. In this way, it can be ensured that the trained force collaboration control model can not only efficiently complete the assembly task, but also protect the skylight parts from damage during the assembly process.

[0070] S300: Construct a force collaboration control model, and obtain the collaborative pressure parameter and the collaborative suction parameter according to the force collaboration control model.

[0071] Specifically, according to the collected force sample data set (including the upper pressure, the lower suction, the difference between the two, and the assembly error label), through the detailed process described above, a force collaboration control model based on impedance control is constructed. In the model, physical parameters such as stiffness, damping, and mass are used as variables to be optimized, and parameter fitting is performed through deep learning (such as a regression neural network) to reflect the relationship between force and error during the assembly process. Using the trained force collaboration control model, the currently collected force data (such as the actual pressure value, suction value, etc.) and the assembly state information are input in real time during the actual assembly process. The model makes inferences based on the input data and outputs the optimal collaborative pressure parameter and the collaborative suction parameter.

[0072] For example, assume that in a sunroof assembly test, the following data is obtained through acquisition: upper pressure data samples: 490 N, 495 N, 500 N, 505 N, 510 N; lower suction data samples: corresponding suction data after conversion (such as about 60 kPa); assembly error labels: 0.3 mm, 0.25 mm, 0.2 mm, 0.22 mm, 0.2 mm. A regression neural network model is constructed. Initially, the stiffness of the upper robot is set to 1000 N / m, the damping to 50 N·s / m, and the mass to 200 kg. The stiffness of the lower part is set to 800 N / m, the damping to 40 N·s / m, and the mass to 150 kg. Through 2000 epochs of training, the mean squared error of the model on the validation set is reduced to about ±0.05 mm. When the training converges, the model can output the collaborative pressure parameter (about 500 N) and the collaborative suction parameter (about 60 kPa) according to the current real-time force data, guiding the robot to adjust the pressure and suction. By constructing a force collaboration control model, the stability and safety of the robot assembly process are realized, and the automation and intelligence level of the assembly process are improved.

[0073] S400: The upper pressure robot controls the pressure of the pre-assembled sunroof part according to the collaborative pressure parameter, and the lower suction robot controls the suction of the pre-assembled sunroof part according to the collaborative suction parameter to obtain a completed sunroof part.

[0074] Specifically, the collaborative pressure parameter is sent to the upper pressure robot, and the collaborative suction parameter is sent to the lower suction robot. The upper pressure robot adjusts its degree-of-freedom pressurizing manipulator and pressurizing platform according to the collaborative pressure parameter to accurately apply a predetermined pressure, so that the upper surface of the sunroof part closely fits the assembly base. At the same time, the lower suction robot controls its negative pressure generator and the honeycomb suction cups on the vacuum adsorption platform according to the collaborative suction parameter to realize the adsorption and fixation of the lower surface of the sunroof part. The uniform distribution of the suction helps to prevent the sunroof part from shifting or being damaged due to uneven stress during the pressurization process.

[0075] During the entire pressurization and adsorption process, both robots monitor the actual stress situation in real time through their respective sensors (such as force sensors and negative pressure sensors), and feed the data back to their respective control modules. The control module makes fine adjustments to the collaborative pressure and suction according to the feedback data to ensure that the actually applied pressure and suction are stable near the predetermined target, so as to ensure that the sunroof part can ensure precise alignment when stressed and will not exceed the safe operating conditions.

[0076] After the upper robot applies pressure and the lower robot completes the adsorption and fixation actions in coordination, the pre-assembled skylight part is firmly fixed on the assembly base, thus completing the assembly. During the whole process, under the coordinated action of forces by the upper and lower robots, it is ensured that each component of the skylight part is evenly stressed and accurately aligned, achieving high assembly accuracy and stability. For example, the upper pressure robot applies pressure according to the cooperative pressure parameter (such as 500N) output by the force cooperation control model, and at the same time, the lower suction robot performs suction control according to the cooperative suction parameter (such as 60 kPa). In the experiment, during the initial pressure application stage, the applied pressure of the upper robot was measured to be 498N, and it stabilized at 500N ± 2N after closed-loop adjustment; the initial negative pressure of the lower suction robot was 59 kPa, and it stabilized at 60 kPa ± 1 kPa after adjustment. The results show that after the cooperative application of pressure and suction, the pre-assembled skylight part can be firmly and evenly fixed, and the assembly error is controlled within ±0.25 mm.

[0077] Through the coordinated action of forces by the upper and lower robots, it is ensured that the skylight part is evenly stressed, avoiding local overpressure or insufficient suction, thereby reducing the risk of material damage and breakage. The upper and lower robots can quickly adjust the pressure application and suction parameters to ensure that they always remain within the predetermined target range during the assembly process, making the assembly process highly automated, significantly shortening the assembly time, and at the same time reducing rework caused by assembly errors, and significantly improving the overall production efficiency.

[0078] In summary, the skylight production and assembly control method based on the robot cooperation network provided by this application has the following beneficial effects: By obtaining the customized skylight part, aligning and pre-assembling it according to the assembly position of the skylight part to obtain the pre-assembled skylight part; starting the distributed cooperative robots, the distributed cooperative robots include an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; constructing a force cooperation control model, and obtaining the cooperative pressure parameter and the cooperative suction parameter according to the force cooperation control model; the upper pressure robot performs pressure control on the pre-assembled skylight part according to the cooperative pressure parameter, and the lower suction robot performs suction control on the pre-assembled skylight part according to the cooperative suction parameter to obtain the assembled skylight part. That is to say, by setting double-pressure assembly robots, one pressure robot is distributed on the skylight and one suction robot is distributed under the skylight. After the skylight is aligned, the two robots flexibly cooperate for force cooperation assembly to achieve precise coordination and improve the product quality and production efficiency of skylight assembly.

[0079] Embodiment 2, based on the same inventive concept as the skylight production and assembly control method based on the robot cooperation network in the foregoing Embodiment 1, this application also provides a skylight production and assembly control system based on the robot cooperation network. Please refer to the appendix Figure 2, the skylight production and assembly control system based on the robot collaboration network includes: A skylight part acquisition module 11, which is used to acquire customized skylight parts, align and pre-assemble them according to the assembly positions of the skylight parts, and obtain pre-assembled skylight parts; A robot startup module 12, which is used to start distributed collaborative robots, and the distributed collaborative robots include an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; A control model construction module 13, which is used to construct a force collaboration control model, and obtain collaborative pressure parameters and collaborative suction parameters according to the force collaboration control model; A skylight part assembly module 14, which is used for the upper pressure robot to perform pressure control on the pre-assembled skylight parts according to the collaborative pressure parameters, and the lower suction robot to perform suction control on the pre-assembled skylight parts according to the collaborative suction parameters, and obtain the assembled skylight parts.

[0080] Furthermore, the robot startup module 12 in the skylight production and assembly control system based on the robot collaboration network is further used for: The force collaboration control model is connected to an assembly path planning module. The visual extraction unit collects the assembly position image of the pre-assembled skylight parts to obtain an assembly image; performs pixel recognition on the assembly image to obtain a first pixel data set and a second pixel data set; respectively obtains the first unit coverage pixel data corresponding to the upper pressure robot and the second unit coverage pixel data corresponding to the lower suction robot. The assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data, and obtains an assembly planning path; controls the upper pressure robot and the lower suction robot according to the assembly planning path.

[0081] Furthermore, the robot startup module 12 in the skylight production and assembly control system based on the robot collaboration network is further used for: The assembly path planning module separates the first pixel data set by using the first unit coverage pixel data as a pixel step to obtain a first set of path nodes; the assembly path planning module separates the second pixel data set by using the second unit coverage pixel data as a pixel step to obtain a second set of path nodes; performs trajectory connection on the first set of path nodes to obtain a first planned path, performs trajectory connection on the second set of path nodes to obtain a second planned path, and generates an assembly planning path from the first planned path and the second planned path, where the first planned path and the second planned path are not synchronized.

[0082] Furthermore, the robot startup module 12 in the skylight production and assembly control system based on the robot collaboration network is further used for: Determine the mean pixel data based on the first unit-covered pixel data and the second unit-covered pixel data; use the mean pixel data as a pixel step to separate the first pixel data set and the second pixel data set, obtaining a first set of path nodes and a second set of path nodes; perform trajectory connection on any one of the first set of path nodes or the second set of path nodes according to a preset trajectory to obtain a synchronous planning path, and generate an assembly planning path based on the synchronous planning path.

[0083] Further, the skylight production and assembly control system based on the robot cooperation network further includes an upper pressure feedback module, and the upper pressure feedback module is further configured to: The upper pressure robot includes a six-degree-of-freedom pressurizing robotic arm, a pressurizing platform, a pressure control module, and a force sensor; send the cooperation pressure parameter to the six-degree-of-freedom pressurizing robotic arm, and the pressure control module drives the pressurizing platform to perform pressure control on the pre-assembled skylight part by controlling the six-degree-of-freedom pressurizing robotic arm; perform pressure monitoring by the force sensor, upload the obtained pressure sensing data to the pressure control module, and perform force feedback on the six-degree-of-freedom pressurizing robotic arm based on the pressure control module.

[0084] Further, the skylight production and assembly control system based on the robot cooperation network further includes a lower suction feedback module, and the lower suction feedback module is further configured to: The lower suction robot includes a negative pressure generator, a vacuum adsorption platform, and a suction control module; send the cooperation suction parameter to the suction control module, and the suction control module controls the suction of the suction cups on the vacuum adsorption platform according to the negative pressure generator.

[0085] Further, the skylight production and assembly control system based on the robot cooperation network further includes a vacuum adsorption module, and the vacuum adsorption module is further configured to: The suction cups on the vacuum adsorption platform are arranged in a honeycomb array.

[0086] Further, the control model construction module 13 in the skylight production and assembly control system based on the robot cooperation network is further configured to: Obtain the force sample dataset of the distributed collaborative robot, where the force sample dataset includes the pressure sample dataset of the upper pressure robot and the suction sample dataset of the lower suction robot, the difference dataset corresponding to the pressure sample dataset and the suction sample dataset, and the label information characterizing the assembly error; Define the parameters of the impedance control model, including the initial stiffness, initial damping, and initial mass parameters of the upper pressure robot and the lower suction robot; Train the defined impedance control model according to the force sample dataset, and when the model converges, output the force collaborative control model.

[0087] Further, the control model construction module 13 in the skylight production and assembly control system based on the robot collaboration network is further configured to: Collect the material information and geometric information of the skylight part; Perform force protection analysis according to the material information and geometric information, output the force protection threshold, and use the force protection threshold as a constraint condition for training the impedance control model to obtain the force collaborative control model.

[0088] The various embodiments in this specification are described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The foregoing Figure 1 The skylight production and assembly control method and specific examples in the first embodiment are equally applicable to the skylight production and assembly control system based on the robot collaboration network in this embodiment. Through the foregoing detailed description of the skylight production and assembly control method based on the robot collaboration network, those skilled in the art can clearly know the skylight production and assembly control system based on the robot collaboration network in this embodiment. Therefore, for the sake of simplicity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0089] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0090] Obviously, for those skilled in the art, without departing from the principle of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the present application.

Claims

1. A skylight production and assembly control method based on a robot collaborative network, characterized in that: include: Obtain a custom-made skylight component, align and pre-assemble it according to the assembly position of the skylight component, and obtain a pre-assembled skylight component; Starting a distributed collaborative robot, wherein the distributed collaborative robot includes an upper pressure robot and a lower suction robot, wherein the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; Constructing a force cooperation control model, and obtaining cooperation pressure parameters and cooperation suction parameters according to the force cooperation control model; The upper pressure robot controls the pressure of the pre-assembled skylight component according to the cooperative pressure parameters, and the lower suction robot controls the suction of the pre-assembled skylight component according to the cooperative suction parameters to obtain an assembled skylight component.

2. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 1, characterized in that: The force cooperation control model is connected with the assembly path planning module, including: The assembly position image of the pre-assembled skylight component is collected by a visual extraction unit to obtain an assembly image; Performing pixel recognition on the assembly image to obtain a first pixel data set and a second pixel data set; Respectively obtaining first unit coverage pixel data corresponding to the upper pressure robot and second unit coverage pixel data corresponding to the lower suction robot, and the assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data to obtain an assembly planning path; The upper pressure robot and the lower suction robot are controlled according to the assembly planning path.

3. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 2, characterized in that: The assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data, including: The assembly path planning module uses the first unit coverage pixel data as a pixel step to separate the first pixel data set to obtain a first group of path nodes; The assembly path planning module uses the second unit coverage pixel data as a pixel step to separate the second pixel data set to obtain a second group of path nodes; The first group of path nodes are connected by trajectories to obtain a first planned path, the second group of path nodes are connected by trajectories to obtain a second planned path, and the first planned path and the second planned path are combined to generate an assembly planned path, wherein the first planned path and the second planned path are not synchronized.

4. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 2, characterized in that: The assembly path planning module performs assembly path planning on the first pixel data set and the second pixel data set according to the first unit coverage pixel data and the second unit coverage pixel data, further comprising: Determine mean pixel data according to the first unit coverage pixel data and the second unit coverage pixel data; Separating the first pixel data set and the second pixel data set by using the mean pixel data as a pixel step to obtain a first group of path nodes and a second group of path nodes; Connect any one of the first group of path nodes or the second group of path nodes according to a preset trajectory to obtain a synchronous planning path, and generate an assembly planning path based on the synchronous planning path.

5. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 1, characterized in that: The upper pressure robot includes a six-degree-of-freedom pressure robot arm, a pressure platform, a pressure control module and a force sensor; The cooperative pressure parameter is sent to the six-degree-of-freedom pressurizing robot arm, and the pressure control module controls the six-degree-of-freedom pressurizing robot arm to drive the pressurizing platform to perform pressure control on the pre-assembled skylight component; The force sensor performs pressure monitoring, and the obtained pressure sensing data is uploaded to the pressure control module, and force feedback is performed on the six-degree-of-freedom pressurized robotic arm based on the pressure control module.

6. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 1, characterized in that: The lower suction robot comprises a negative pressure generator, a vacuum adsorption platform and a suction control module; The collaborative suction parameters are sent to the suction control module, and the suction control module controls the suction of the suction cup on the vacuum adsorption platform according to the negative pressure generator.

7. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 6, characterized in that: The suction cups on the vacuum adsorption platform are arranged in a honeycomb array.

8. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 1, characterized in that: The skylight production and assembly control method based on the robot collaborative network also includes: Acquire a force sample data set of the distributed collaborative robot, wherein the force sample data set includes a pressure sample data set of the upper pressure robot and a suction sample data set of the lower suction robot, a difference data set corresponding to the pressure sample data set and the suction sample data set, and label information characterizing assembly errors; Defining parameters of the impedance control model, including initial stiffness, initial damping and initial mass parameters of the upper pressure robot and the lower suction robot; The defined impedance control model is trained according to the force sample data set, and the force cooperative control model is output when the model converges.

9. The skylight production and assembly control method based on a robot collaborative network as claimed in claim 8, characterized in that: Acquisition force cooperative control model, also includes: Collecting material information and geometric information of the skylight component; A force protection analysis is performed according to the material information and the geometric information, and a force protection threshold is output. The force protection threshold is used as a constraint condition for training the impedance control model to obtain a force cooperation control model.

10. The skylight production and assembly control system based on robot collaborative network is characterized by: The steps for implementing the sunroof production and assembly control method based on a robot collaborative network as described in any one of claims 1 to 9, wherein the sunroof production and assembly control system based on a robot collaborative network comprises: A skylight component acquisition module, used to acquire a custom-produced skylight component, align and pre-assemble it according to the assembly position of the skylight component, and acquire a pre-assembled skylight component; A robot start module, used to start a distributed collaborative robot, wherein the distributed collaborative robot includes an upper pressure robot and a lower suction robot, and the upper pressure robot and the lower suction robot are synchronously controlled through a distributed control architecture; A control model building module, used to build a force cooperation control model, and obtain cooperation pressure parameters and cooperation suction parameters according to the force cooperation control model; The skylight component assembly module is used for the upper pressure robot to control the pressure of the pre-assembled skylight component according to the cooperative pressure parameters, and the lower suction robot to control the suction of the pre-assembled skylight component according to the cooperative suction parameters, so as to obtain the assembled skylight component.

Images