Intelligent gas filling system and method

Abstract

The invention discloses an intelligent gas filling system and method. The intelligent gas filling system comprises an image acquisition unit, a center shaft positioning unit, a path planning unit and a tail end execution unit, the image acquisition unit is used for acquiring a gas cylinder pose image and a bottleneck position image in the gas cylinder pose image; the central axis positioning unit is used for screening out a plurality of positioning points according to the bottle opening position image and determining a central axis based on the plurality of positioning points; the path planning unit is used for deploying a plurality of end effector operation paths for the mechanical arm based on the central axis, and determining an optimal target path in the operation paths; and the tail end execution unit is used for enabling a tail end executor to move the inflated connector to the position of a bottle opening by taking the optimal target route as a motion path and connecting the inflated connector. The technical problems that due to the fact that connection and positioning of a gas cylinder valve port and a gas charging connector are not accurate, butt joint angle deviation and uneven clamping force are caused, then poor contact of a sealing face or local stress concentration is caused, and finally gas leakage or connector damage is caused are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas filling, and in particular to a gas intelligent filling system and method. Background Art

[0002] Gas filling is the process of filling gas cylinders according to regulations. It strictly adheres to the GB 14194-2006 safety standard and covers aspects such as cylinder inspection and cleaning, pressure and temperature monitoring, and leak testing. During operation, attention must be paid to gas compatibility, explosion and toxicity protection, equipment calibration, and personnel qualifications. This process is widely used in industries such as industry, healthcare, and firefighting. Cylinder compliance must be confirmed before filling, and the filling volume and pressure must be controlled during the filling process to ensure leak-free storage and transportation.

[0003] Currently, according to the requirements of the gas cylinder valve outlet connection standard, the sealing performance and connection strength of the interface must strictly meet the working pressure and airtightness requirements. However, during the traditional gas cylinder filling process, the connection between the gas cylinder valve port and the charging connector requires manual adjustment of the angle and force. For example, during actual operation on the workshop, inaccurate connection positioning of the gas cylinder valve port and the charging connector can lead to deviation in the connection angle and uneven clamping force, which in turn causes poor contact between the sealing surfaces or localized stress concentration, ultimately resulting in gas leakage or damage to the interface. Summary of the Invention

[0004] The purpose of the present invention is to provide a gas intelligent filling system and method to solve the technical problems of poor sealing surface contact or local stress concentration, which ultimately cause gas leakage or interface damage, caused by inaccurate connection positioning between the gas cylinder valve port and the charging connector, resulting in docking angle deviation and uneven clamping force.

[0005] The technical solution of the present invention is achieved as follows:

[0006] In one aspect, the present invention provides a gas intelligent filling system, comprising an image acquisition unit, a central axis positioning unit, a path planning unit, and an end-effector unit;

[0007] The image acquisition unit is used to acquire the gas cylinder posture image and the bottle mouth position image in the gas cylinder posture image;

[0008] The central axis positioning unit is configured to screen out a plurality of positioning points according to the bottle mouth position image, and determine the central axis based on the plurality of positioning points;

[0009] The path planning unit deploys a plurality of operation routes for the end effector of the robot arm based on the central axis, and determines an optimal target route among the operation routes;

[0010] The end effector unit is used for the end effector to move the inflatable joint to the bottle mouth position and connect it with the optimal target route.

[0011] A further technical solution is that the image acquisition unit includes:

[0012] Multi-view image acquisition module, used to synchronously capture 3D images of gas cylinders from multiple angles using multiple industrial cameras;

[0013] A bottle mouth area positioning module, based on edge detection technology, automatically identifies the pixel range of the bottle mouth area in the three-dimensional image and marks its specific coordinates in the three-dimensional image;

[0014] An image processing module is used to perform brightness adjustment, contrast enhancement and noise reduction on the three-dimensional image to generate a posture image;

[0015] The image data output module is used to output the posture image and bottle mouth position information in a standardized format to provide a data basis for guiding the operation of the robotic arm.

[0016] A further technical solution is that the bottle mouth area positioning module specifically includes:

[0017] An image preprocessing module, configured to perform grayscale conversion, histogram equalization, and Gaussian filtering on the three-dimensional image;

[0018] An edge detection execution module is used to perform edge detection on the pre-processed three-dimensional image using a Canny operator, extract the gradient change area of the bottle mouth contour, and generate an edge pixel set;

[0019] A bottle mouth region segmentation module is used to select a region that meets the bottle mouth shape characteristics from the edge pixel set and determine its pixel range;

[0020] A three-dimensional coordinate mapping module, used to convert the pixel range into three-dimensional space coordinates through a camera, and calculate the spatial position of the center point and edge points of the bottle mouth in combination with the depth information;

[0021] The coordinate marking module is used to display the pixel range and the corresponding three-dimensional coordinates of the bottle mouth area in the three-dimensional image by a marking method.

[0022] A further technical solution is that the central axis positioning unit includes:

[0023] The positioning point screening module is used to extract multiple candidate positioning points from the bottle mouth position image and select a set of positioning points that meet the symmetry characteristics of the bottle mouth through geometric consistency;

[0024] A geometric feature analysis module is used to verify whether the spatial distribution pattern of the positioning point set conforms to the geometric structure of the bottle mouth;

[0025] A central axis fitting module, based on the positioning point set, uses Hough transform to fit a straight line passing through the positioning point set as the central axis of the bottle mouth, and calculates its center coordinates and direction vector;

[0026] The central axis marking module is used to mark the central axis in the bottle mouth position image and output its pixel coordinate range and corresponding geometric parameters in the bottle mouth position image.

[0027] A further technical solution is that the path planning unit includes:

[0028] The preliminary path generation module is used to generate multiple sets of candidate operation routes based on the geometric characteristics of the central axis and the kinematic model of the robot end effector;

[0029] a path evaluation module for evaluating the path length, the deviation from the central axis, and the stability of the end effector posture of the plurality of candidate running routes;

[0030] a path optimization module, which eliminates path jitter or redundant segments for the plurality of candidate operation routes based on the evaluation results of the path evaluation module, thereby ensuring the dynamic response capability of the end effector during the filling process;

[0031] The optimal path selection module comprehensively considers the evaluation results of the path evaluation module and the optimization results of the path optimization module, adopts a decision algorithm to select an optimal target path from all the candidate operation paths, and outputs its coordinate sequence as the operation instruction of the end effector.

[0032] A further technical solution is that the path evaluation module specifically includes:

[0033] a path length calculation module, configured to quantify the total length of each candidate running route using a continuous path integration method, and output a length value of each candidate running route;

[0034] a deviation evaluation module, configured to calculate the maximum deviation distance between each candidate running route and the central axis by a projection method, and evaluate whether the candidate running route is aligned with the central axis direction;

[0035] The posture stability analysis module is used to evaluate whether singular points or posture jitter occur during the execution of the candidate running route based on the posture parameters of the end effector, and output a stability score.

[0036] A further technical solution is that the deviation evaluation module specifically executes the following steps:

[0037] Step S11: Project each path point in the candidate running route onto the central axis, and calculate the coordinates of the projection point on the central axis;

[0038] Step S12: Calculate the distance between each path point and the projection point, and count the maximum deviation distance of the entire path;

[0039] Step S13: judging whether the candidate running route is aligned with the central axis direction based on the maximum deviation distance, wherein if the maximum deviation distance is less than a preset threshold, it is judged to be aligned;

[0040] Step S14: output the maximum deviation distance and fit evaluation result of each candidate running route.

[0041] A further technical solution is that the path optimization module specifically includes:

[0042] The path jitter elimination module smoothes the path segments with high-frequency jitter in all candidate running routes;

[0043] A redundant segment elimination module, configured to eliminate repeated or invalid path segments from all candidate running routes;

[0044] The dynamic response optimization module performs time parameterized adjustment on the candidate running route in combination with the joint acceleration and angular velocity constraints of the robotic arm.

[0045] A further technical solution is that the dynamic response optimization module specifically executes the following steps:

[0046] Step S21: converting the joint angular velocity and joint acceleration constraints of the robotic arm into a mathematical model, and defining the feasible motion boundary of each optimized candidate running route;

[0047] Step S22: Based on the mathematical model, time scaling is performed on the candidate operation route to adjust the execution time of each candidate operation route segment;

[0048] Step S23: Verify whether each candidate running route satisfies all the joint angular velocity and acceleration constraints. If there is a conflict, trigger an alarm and return to a safe state.

[0049] In another aspect, the present invention provides a method for intelligent gas filling, comprising the following steps:

[0050] Step 101: collecting a gas cylinder posture image and a bottle mouth position image in the gas cylinder posture image;

[0051] Step 102: Filter out a plurality of positioning points according to the bottle mouth position image, and determine a central axis based on the plurality of positioning points;

[0052] Step 103: deploying a plurality of end effector operation routes for the robotic arm based on the central axis, and determining an optimal target route among the operation routes;

[0053] Step 104: The end effector moves the inflatable joint to the bottle mouth position and connects it using the optimal target route as a movement path.

[0054] The beneficial effects of the present invention are:

[0055] The present invention uses the image acquisition unit to accurately capture the posture and bottle mouth position image of the gas cylinder through high-precision three-dimensional imaging technology, providing a reliable data basis for subsequent positioning; the central axis positioning unit screens the positioning points and calculates the central axis based on the bottle mouth image, ensuring that the end effector of the robotic arm moves along the geometric center axis of the bottle mouth, greatly reducing the docking angle deviation; the path planning unit generates multiple candidate paths in combination with the central axis and screens the best target route, and optimizes the motion trajectory through time parameterization adjustment to avoid jitter or impact during high-speed movement of the robotic arm; under the guidance of the optimal path, the end execution unit dynamically adjusts the clamping force of the inflation joint to ensure uniform fit of the sealing surface, eliminates local stress concentration, and realizes high-precision docking between the gas cylinder valve mouth and the inflation joint, thereby solving the technical problems of docking angle deviation and uneven clamping force caused by inaccurate connection and positioning of the gas cylinder valve mouth and the inflation joint, resulting in poor contact of the sealing surface or local stress concentration, and ultimately causing gas leakage or interface damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A block diagram of a gas intelligent filling system provided by the present invention;

[0057] Figure 2 This is a flow chart of the steps of the intelligent gas filling method provided by the present invention. DETAILED DESCRIPTION

[0058] In order to better understand the technical content of the present invention, specific embodiments are provided below, and the present invention is further described in conjunction with the accompanying drawings.

[0059] Example 1

[0060] See also Figure 1On the one hand, the present invention provides a gas intelligent filling system, including an image acquisition unit, a central axis positioning unit, a path planning unit and an end execution unit; the image acquisition unit is used to acquire a gas cylinder posture image and a bottle mouth position image in the gas cylinder posture image; the central axis positioning unit is used to screen out multiple positioning points according to the bottle mouth position image, and determine the central axis based on the multiple positioning points; the path planning unit deploys multiple end effector operation routes for the robot arm based on the central axis, and determines the best target route among the operation routes; the end execution unit is used for the end effector to move the inflated joint to the bottle mouth position and connect it with the best target route as the movement path.

[0061] It should be noted that the gas cylinder posture image includes the image of the overall outline and posture of the gas cylinder, and the bottle mouth position image includes the image of the bottle mouth area.

[0062] In an embodiment of the present invention, the present invention uses an image acquisition unit through high-precision three-dimensional imaging technology to accurately capture the posture and position image of the gas cylinder and the bottle mouth position, providing a reliable data basis for subsequent positioning; the central axis positioning unit screens the positioning points and calculates the central axis based on the bottle mouth image, ensuring that the end effector of the robotic arm moves along the geometric center axis of the bottle mouth, greatly reducing the docking angle deviation; the path planning unit generates multiple candidate paths in combination with the central axis and screens the best target route, and optimizes the motion trajectory through time parameterization adjustment to avoid jitter or impact during high-speed movement of the robotic arm; under the guidance of the optimal path, the end execution unit dynamically adjusts the clamping force of the inflation joint to ensure that the sealing surface fits evenly, eliminates local stress concentration, and achieves high-precision docking between the gas cylinder valve mouth and the inflation joint, thereby solving the technical problem of docking angle deviation and uneven clamping force caused by inaccurate connection positioning between the gas cylinder valve mouth and the inflation joint, resulting in poor contact of the sealing surface or local stress concentration, and ultimately causing gas leakage or interface damage.

[0063] Preferably, the image acquisition unit includes: a multi-view image acquisition module, which is used to synchronously capture three-dimensional stereo images of gas cylinders from multiple angles through multiple industrial cameras; a bottle mouth area positioning module, which automatically identifies the pixel range of the bottle mouth area in the three-dimensional stereo image based on edge detection technology, and marks its specific coordinates in the three-dimensional stereo image; an image processing module, which is used to perform brightness adjustment, contrast enhancement and denoising on the three-dimensional stereo image to generate a posture image; an image data output module, which is used to output the posture image and bottle mouth position information in a standardized format to provide a data basis for guiding the operation of the robotic arm.

[0064] In an embodiment of the present invention, the multi-view image acquisition module uses at least three industrial cameras to synchronously capture three-dimensional stereo images of the gas cylinder from different angles, such as the front, side, and oblique views. The camera must have a high resolution of ≥2 million pixels and a high frame rate of ≥30fps, and be configured with a timestamp synchronization or trigger signal synchronization mechanism to ensure the consistency of the multi-view images in the time domain and space domain. The acquired original image contains the overall outline of the gas cylinder, the texture of the bottle body, and the geometric features of the bottle mouth, providing basic data for subsequent positioning. The bottle mouth area positioning module performs gradient calculation and threshold segmentation on the three-dimensional stereo image based on edge detection technology to extract the pixel range of the bottle mouth edge. The image processing module optimizes the acquired three-dimensional stereo images, including: brightness adjustment: enhancing the contrast between the bottle mouth area and the background through histogram equalization; contrast enhancement: using the histogram equalization algorithm to dynamically adjust the contrast of the local area to highlight the edge features of the bottle mouth; denoising: using wavelet transform denoising to eliminate salt and pepper noise or motion artifacts in the image and retain the integrity of the bottle mouth geometric structure. The processed image is used as the cylinder pose image, which contains the pitch angle, yaw angle, roll angle and precise coordinates of the cylinder mouth position. The image data output module outputs the processed pose image and bottle mouth position information in XML standardized format. The format file includes: the image size, timestamp, camera intrinsic parameter matrix of the pose image; the pixel coordinates of the bottle mouth position and depth value , combined with the camera calibration parameters to convert into three-dimensional space coordinates The diameter and tilt angle of the center point of the bottle mouth are used to guide the motion path planning of the end effector of the robotic arm.

[0065] Preferably, the bottle mouth area positioning module specifically includes: an image preprocessing module for grayscale conversion, histogram equalization and Gaussian filtering of the three-dimensional stereo image; an edge detection execution module for edge detection of the preprocessed three-dimensional stereo image using the Canny operator, extracting the gradient change area of the bottle mouth contour, and generating an edge pixel set; a bottle mouth area segmentation module for screening out the area that meets the bottle mouth shape characteristics from the edge pixel set and determining its pixel range; a three-dimensional coordinate mapping module for converting the pixel range into three-dimensional space coordinates through a camera, and calculating the spatial position of the bottle mouth center point and edge points in combination with depth information; and a coordinate marking module for displaying the pixel range of the bottle mouth area and its corresponding three-dimensional coordinates in the three-dimensional stereo image by a marking method.

[0066] In this embodiment of the present invention, the image preprocessing module performs the following processing on the 3D image: Grayscaling: Converts the RGB color image into a single-channel grayscale image, using a weighted averaging method to reduce computational complexity while preserving the brightness characteristics of the bottle mouth outline; Histogram Equalization: Dynamically adjusts the contrast of local regions through a histogram equalization algorithm, enhancing the grayscale difference between the bottle mouth edge and the background, and improving the robustness of edge detection; Gaussian filtering: Smoothes the image using a two-dimensional Gaussian kernel to suppress noise interference while preserving the bottle mouth edge information. The edge detection execution module uses the Canny operator to calculate the image gradient magnitude and direction in the x and y directions to generate a gradient map. It then performs local maximum filtering along the gradient direction, refining the edge line width to a single pixel level. It then sets high and low thresholds and connects edge breakpoints using a hysteresis technique to generate a continuous set of edge pixels. This set contains the gradient variation region of the bottle mouth outline, providing basic data for subsequent segmentation. The bottle mouth area segmentation module selects the area that meets the bottle mouth shape characteristics from the edge pixel set, and then removes small area noise points through expansion and corrosion to retain the closed contour; based on the geometric characteristics of the bottle mouth, that is, the circularity C = 4πA / P², where A is the contour area and P is the perimeter, the contour with a circularity close to 1 is selected and its minimum circumscribed circle is calculated to determine the diameter range D∈ [100, 300] of the bottle mouth area. The three-dimensional coordinate mapping module converts the pixel range of the bottle mouth area into three-dimensional space coordinates, and establishes the projective transformation relationship between pixel coordinates and three-dimensional space coordinates based on the camera intrinsic parameter matrix and extrinsic parameter matrix; then, combined with the depth map provided by the depth camera, the depth map is calculated by the formula Get the depth value of the center point and edge point of the bottle mouth; use the internal and external parameters of the camera and the depth information to convert the pixel coordinates Convert to three-dimensional space coordinates , the conversion formula is:

[0067]

[0068] in 、 is the image center coordinate, 、 is the focal length parameter.

[0069] The coordinate marking module displays the pixel range of the bottle mouth area and its corresponding three-dimensional coordinates in the three-dimensional image through a visual marking method, draws the minimum circumscribed circle and edge contour line of the bottle mouth area in the image; and calculates the bottle mouth center point. The edge point coordinates are superimposed on the image in text form for the robot arm path planning module to call.

[0070] Preferably, the central axis positioning unit includes: a positioning point screening module, which is used to extract multiple candidate positioning points from the bottle mouth position image, and screen out a positioning point set that meets the symmetry characteristics of the bottle mouth through geometric consistency; a geometric feature analysis module, which is used to verify whether the spatial distribution law of the positioning point set conforms to the geometric structure of the bottle mouth; a central axis fitting module, which uses Hough transform to fit a straight line passing through the positioning point set based on the positioning point set as the central axis of the bottle mouth, and calculates its center coordinates and direction vector; a central axis marking module, which is used to mark the central axis in the bottle mouth position image, and output its pixel coordinate range and corresponding geometric parameters in the bottle mouth position image.

[0071] In this embodiment of the present invention, the positioning point screening module uses morphological operations such as dilation and erosion to remove noise points based on the edge pixel set output by the bottle mouth area positioning module, retaining the closed contour, and then extracts candidate positioning points at the inflection points or intersection points of the symmetry axes of the edge contour. Based on the symmetry characteristics of the bottle mouth axis or central symmetry, the positioning points are screened according to the following conditions:

[0072] 1. Calculate the symmetry axis deviation of the candidate point relative to the center point of the bottle mouth;

[0073] Second: The distance between the candidate points and the center point of the bottle mouth is within a reasonable range, and finally a set of positioning points that meet the symmetric characteristics is generated (for example, N ≥ 10 points).

[0074] The geometric feature analysis module verifies the spatial distribution of the positioning point set to ensure that it conforms to the geometric structure of the bottle mouth. The least squares method is used to calculate the minimum circumscribed circle of the positioning point set to verify the matching degree of its radius with the actual radius of the bottle mouth. Then the symmetry axis direction of the positioning point set is calculated and compared with the symmetry axis direction of the bottle mouth. The deviation must meet The central axis fitting module uses Hough transform to fit a straight line passing through the set of positioning points as the central axis of the bottle mouth, and calculates its center coordinates and direction vector. The set of positioning points is mapped to the parameter space in the polar coordinate system. ,in is the vertical distance from the line to the origin, is the angle between the normal direction and the x-axis; then initialize the accumulator matrix, for each positioning point Calculate its corresponding parameter curve and vote in the accumulator; detect the peak value through the sliding window method, and select the parameter group with the highest number of votes as the polar coordinate representation of the central axis. Convert the polar coordinate parameters into the straight line equation in the rectangular coordinate system, and calculate the center coordinates of the central axis, that is, the intersection point and direction vector of the straight line and the center point of the bottle mouth. The central axis marking module marks the central axis in the bottle mouth position image and outputs its pixel coordinate range and geometric parameters. Draw the central axis of the red dotted line in the image (such as), mark the center coordinates, and output the pixel coordinate range of the central axis (starting point , end point ) and geometric parameters (direction vector , the angle with the horizontal axis ).

[0075] Preferably, the path planning unit includes: a preliminary path generation module, which is used to generate multiple groups of candidate running routes based on the geometric characteristics of the central axis and the kinematic model of the end effector of the robotic arm; a path evaluation module, which evaluates the path length, deviation from the central axis, and stability of the end effector posture of the multiple groups of candidate running routes; a path optimization module, which eliminates path jitter or redundant segments for the multiple groups of candidate running routes based on the evaluation results of the path evaluation module to ensure the dynamic response capability of the end effector during the filling process; an optimal path selection module, which integrates the evaluation results of the path evaluation module and the optimization results of the path optimization module, adopts a decision algorithm to select an optimal target route from all candidate running routes, and outputs its coordinate sequence as the running instruction of the end effector.

[0076] In the embodiment of the present invention, the path preliminary generation module obtains the linear equation of the central axis from the central axis fitting module. , center point coordinates , direction vector The geometric parameters are used as the basis for path planning; based on the forward and inverse kinematics model of the manipulator, the posture constraints of the end effector are set, that is, the maximum angular velocity , maximum acceleration α_max = 1.5 rad / s² ; Then, the spline interpolation method is used to generate multiple sets of candidate paths. The starting point of the path is the current position of the robot arm, and the end point is a safe operating area with a radius of r = 5mm near the center point of the central axis. It is ensured that the path is collision-free in both joint space and Cartesian space. The path evaluation module quantitatively evaluates multiple sets of candidate operation routes. First, path length: calculate the total length of each path, and give priority to shorter paths to reduce filling time; second, deviation evaluation: calculate the angular deviation between the path and the central axis to ensure that the path is consistent with the direction of the central axis to avoid poor contact of the filling port due to offset; third, posture stability: evaluate the stability through the posture change rate of the end effector, and give priority to paths with smooth posture changes to avoid shaking of the robot arm. The path optimization module dynamically optimizes the candidate paths based on the evaluation results. Specifically, it uses the time-optimal trajectory optimization algorithm to adjust the acceleration curve of the path to reduce the vibration of the robot arm during high-speed movement; it detects redundant inflection points based on the path curvature, smoothes the path through curve fitting, and shortens invalid motion segments; it adjusts the acceleration distribution of the path based on the force data of the end effector fed back by the six-dimensional force sensor to ensure that the end stiffness of the robot arm during the filling process meets the process requirements. The optimal path selection module comprehensively evaluates and optimizes the results, and uses the target decision algorithm to select the best target route from the candidate paths. The weight of the evaluation index is set according to the filling process requirements, that is, the path length , deviation , posture stability ; Calculate the comprehensive score for each path; Finally, select the path with the highest score and add its coordinate sequence point set [ ]Converted into the joint angle sequence of the robot arm[ ] and output it as a running instruction.

[0077] Preferably, the path evaluation module specifically includes: a path length calculation module, which is used to quantify the total length of each candidate running route using a continuous path integration method, and output the length value of each candidate running route; a deviation evaluation module, which is used to calculate the maximum deviation distance between each candidate running route and the central axis through a projection method, and evaluate whether the candidate running route is in line with the direction of the central axis; a posture stability analysis module, which is used to evaluate whether the candidate running route has singular points or posture jitter during execution based on the posture parameters of the end effector, and output a stability score.

[0078] In the embodiment of the present invention, the path length calculation module uses the continuous path integration method to quantify the total length of each candidate running route, specifically the Cartesian coordinate sequence of the candidate path [ ] is discretized into a continuous point set, and the Euclidean distance between adjacent points is accumulated. , calculate the total path length The deviation evaluation module calculates the maximum deviation distance between the candidate path and the central axis through the projection method to evaluate its fit. Specifically, the linear equation of the central axis is obtained from the central axis fitting module. ,in As a starting point, Is the direction vector; for each point on the candidate path , calculate its nearest projection point on the central axis and calculate the vertical distance; traverse the distance values of all points, take the maximum value, and set a tolerance threshold. If the threshold is exceeded, it is marked as excessive deviation. The posture stability analysis module evaluates the singularity risk and posture jitter during the path execution process based on the posture parameters of the end effector. Specifically, the Euler angle of the end effector is obtained in real time through the robot arm encoder and the six-dimensional force sensor; then the Jacobian matrix is calculated. The determinant of ,when When it is determined as a singular point, the path segment is marked as a high-risk area; by calculating the angular velocity change rate of adjacent posture points ,like , and there is a continuous points , it is judged as posture jitter and the stability score is output ,in The maximum allowed jitter times.

[0079] Preferably, the deviation evaluation module specifically performs the following steps:

[0080] Step S11: Project each path point in the candidate running route onto the central axis and calculate the coordinates of the projection point on the central axis. The specific central axis direction vector is , starting from , waypoints Projection point to the central axis Calculated by the following formula:

[0081]

[0082] Step S12: Calculate the distance between each path point and the projection point, and calculate the maximum deviation distance of the entire path. The single point deviation distance is calculated using the formula:

[0083] ;

[0084] The maximum deviation distance is expressed as:

[0085]

[0086] Step S13: Based on the maximum deviation distance, determine whether the candidate running route is aligned with the central axis direction. If the maximum deviation distance is less than a preset threshold, it is determined to be aligned. The preset threshold is , the fit evaluation condition formula is expressed as:

[0087] like Otherwise, it deviates.

[0088] Step S14: Output the maximum deviation distance and fit evaluation result of each candidate running route.

[0089] In this implementation, the position information of each path point in a candidate route is converted into three-dimensional coordinate data. The coordinates of the starting point and direction vector of the central axis are obtained from the central axis fitting module. Subsequently, a mathematical method is used to calculate the closest projection point of each path point on the central axis. This method finds the perpendicular point from the path point to the central axis to determine its corresponding position on the central axis. This process ensures that the projection calculation accuracy error does not exceed 0.05 mm to meet the requirements of high-precision operation scenarios. By comparing the three-dimensional coordinates of each path point with its projection point, the straight-line distance between the two is calculated to obtain the deviation value of the path point. After traversing all path points along the entire route, the maximum deviation value is calculated as the maximum deviation distance of the candidate route. This value directly reflects the overall deviation of the path from the central axis and provides a basis for subsequent judgment. The maximum deviation distance is determined based on a preset process threshold (for example, the tolerance range of the gas cylinder filling port is set to 2 mm). If the maximum deviation distance is less than the threshold, the candidate route is determined to be aligned with the central axis direction; otherwise, it is marked as excessive deviation. This threshold is set based on the actual operating conditions of the robot arm's end-effector joint angular velocity limits and cylinder model variations to ensure that the evaluation results meet actual production requirements. The calculated results for each candidate route, including the maximum deviation distance and fit determination status, are output in a data table for use by the path optimization module. The output data includes a unique path identifier, maximum deviation distance value, and fit / unfit status.

[0090] Preferably, the path optimization module specifically includes: a path jitter elimination module, which smoothes the path segments with high-frequency jitter in all candidate running routes; a redundant segment elimination module, which is used to eliminate repeated or invalid path segments in all candidate running routes; and a dynamic response optimization module, which combines the joint acceleration and angular velocity constraints of the robot arm to perform time parameterized adjustment on the candidate running routes.

[0091] In this embodiment of the present invention, the path jitter elimination module analyzes the continuous displacement change rate of path points and, in combination with the angular velocity limit of the robot end-effector, identifies high-frequency jitter areas. A low-pass filtering algorithm is then used to locally smooth jittered path segments, reducing high-frequency noise and ensuring trajectory continuity. An error threshold is also introduced to balance path accuracy and smoothing effectiveness, preventing excessive filtering from causing the trajectory to deviate from the target. The redundant segment elimination module identifies duplicate or invalid path segments by determining the spatial distribution density of path points. For example, based on a distance threshold (0.1 mm) between adjacent path points, segments with displacements less than the threshold are marked as redundant. A clustering algorithm is used to merge redundant points or remove duplicate path segments, retaining valid turning points to reduce computational complexity. After redundant elimination, the path must be verified to still meet process accuracy requirements, for example by combining the trajectory interpolation error criterion of the robot end-effector. The dynamic response optimization module uses the robot's joint angular velocity and acceleration constraints to perform time-parameterized adjustments on candidate paths. Optimization constraints are established by obtaining the upper angular velocity limits (2 rad / s) and acceleration limits (10 rad / s²) from the robot's dynamic model. Polynomial interpolation is used to assign time parameters, ensuring smooth changes in joint motion within the constraints. Time parameters are dynamically adjusted based on real-time feedback of the manipulator's load status to avoid trajectory instability caused by sudden load changes.

[0092] Preferably, the dynamic response optimization module specifically performs the following steps:

[0093] Step S21: convert the joint angular velocity and joint acceleration constraints of the manipulator into a mathematical model, and define the feasible motion boundary of each optimized candidate running route;

[0094] Step S22: Based on the mathematical model, time scaling is performed on the candidate running routes to adjust the execution time of each candidate running route segment;

[0095] Step S23: Verify whether each candidate running route meets all joint angular velocity and acceleration constraints. If there is a conflict, trigger an alarm and return to a safe state.

[0096] In this embodiment of the present invention, the joint angular velocity and acceleration constraints of the robotic arm are converted into a mathematical model, and the feasible motion boundaries of each optimized candidate route are defined. Specifically, by analyzing the maximum angular velocity and acceleration limits of the robotic arm's joints and combining them with the range of joint motion, a rule system describing the joint motion range is established. For example, based on the actual performance of the robotic arm (with an angular velocity upper limit of 2 rad / s and an acceleration upper limit of 10 rad / s²), the executable range of velocity and acceleration for each path segment is determined, thereby defining the motion boundaries and preventing the trajectory from exceeding the physical capabilities of the robotic arm. Based on the above mathematical model, the candidate routes are time-scaled and the execution time of each path segment is adjusted. By dynamically allocating time parameters for the path segments, joint motion is smoothly transitioned within the speed and acceleration constraints. For example, if the original execution time of a path segment would cause the joint velocity to exceed the upper limit, the execution time of that segment is extended to reduce the peak velocity. If the load changes (such as an increase in end effector weight), the time parameters are further adjusted to ensure that the acceleration does not exceed the limit. The candidate routes are then verified to ensure that they meet all joint angular velocity and acceleration constraints. The optimized trajectory is checked point by point. If the speed or acceleration of a path segment exceeds the preset limit (such as the acceleration of a joint reaching 12 rad / s², while the upper limit is 10 rad / s²), an alarm signal (audio and visual prompts) is immediately triggered, and the robot arm is retracted to the preset safe state (joint angles return to zero, end effector stops) through the safety protocol to avoid equipment damage or operational risks.

[0097] Example 2

[0098] See also Figure 2 On the other hand, the present invention provides a gas intelligent filling method, comprising the following steps:

[0099] Step 101: collecting a gas cylinder posture image and a bottle mouth position image in the gas cylinder posture image;

[0100] Step 102: Filter out multiple positioning points based on the bottle mouth position image, and determine the central axis based on the multiple positioning points;

[0101] Step 103: deploy multiple end effector operation routes for the robotic arm based on the central axis, and determine the optimal target route among the operation routes;

[0102] Step 104 : The end effector moves the inflatable joint to the bottle mouth position and connects it using the optimal target route as the movement path.

[0103] In the embodiment of the present invention, the working principle of the method is the same as the working principle of the system, which will not be described in detail here.

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A gas intelligent filling system, characterized in that: It includes image acquisition unit, central axis positioning unit, path planning unit and end execution unit; The image acquisition unit is used to acquire the gas cylinder posture image and the bottle mouth position image in the gas cylinder posture image; The central axis positioning unit is configured to screen out a plurality of positioning points according to the bottle mouth position image, and determine the central axis based on the plurality of positioning points; The path planning unit deploys a plurality of operation routes for the end effector of the robot arm based on the central axis, and determines an optimal target route among the operation routes; The end effector unit is used for the end effector to move the inflatable joint to the bottle mouth position and connect it with the optimal target route.

2. A gas intelligent filling system according to claim 1, characterized in that: The image acquisition unit includes: Multi-view image acquisition module, used to synchronously capture 3D images of gas cylinders from multiple angles using multiple industrial cameras; A bottle mouth area positioning module, based on edge detection technology, automatically identifies the pixel range of the bottle mouth area in the three-dimensional image and marks its specific coordinates in the three-dimensional image; An image processing module is used to perform brightness adjustment, contrast enhancement and noise removal on the three-dimensional image to generate a posture image; The image data output module is used to output the posture image and bottle mouth position information in a standardized format to provide a data basis for guiding the operation of the robotic arm.

3. A gas intelligent filling system according to claim 2, characterized in that: The bottle mouth area positioning module specifically includes: An image preprocessing module, configured to perform grayscale conversion, histogram equalization, and Gaussian filtering on the three-dimensional image; An edge detection execution module is used to perform edge detection on the pre-processed three-dimensional image using a Canny operator, extract the gradient change area of the bottle mouth contour, and generate an edge pixel set; A bottle mouth region segmentation module is used to select a region that meets the bottle mouth shape characteristics from the edge pixel set and determine its pixel range; A three-dimensional coordinate mapping module, used to convert the pixel range into three-dimensional space coordinates through a camera, and calculate the spatial position of the center point and edge points of the bottle mouth in combination with the depth information; The coordinate marking module is used to display the pixel range and the corresponding three-dimensional coordinates of the bottle mouth area in the three-dimensional image by a marking method.

4. The intelligent gas filling system according to claim 1, characterized in that: The central axis positioning unit includes: The positioning point screening module is used to extract multiple candidate positioning points from the bottle mouth position image and select a set of positioning points that meet the symmetry characteristics of the bottle mouth through geometric consistency; A geometric feature analysis module is used to verify whether the spatial distribution pattern of the positioning point set conforms to the geometric structure of the bottle mouth; A central axis fitting module, based on the positioning point set, uses Hough transform to fit a straight line passing through the positioning point set as the central axis of the bottle mouth, and calculates its center coordinates and direction vector; The central axis marking module is used to mark the central axis in the bottle mouth position image and output its pixel coordinate range and corresponding geometric parameters in the bottle mouth position image.

5. The intelligent gas filling system according to claim 1, characterized in that: The path planning unit includes: The preliminary path generation module is used to generate multiple sets of candidate operation routes based on the geometric characteristics of the central axis and the kinematic model of the robot end effector; a path evaluation module for evaluating the path length, the deviation from the central axis, and the stability of the end effector posture of the plurality of candidate running routes; a path optimization module, which eliminates path jitter or redundant segments for the plurality of candidate operation routes based on the evaluation results of the path evaluation module, thereby ensuring the dynamic response capability of the end effector during the filling process; The optimal path selection module comprehensively considers the evaluation results of the path evaluation module and the optimization results of the path optimization module, adopts a decision algorithm to select an optimal target path from all the candidate operation paths, and outputs its coordinate sequence as the operation instruction of the end effector.

6. The intelligent gas filling system according to claim 5, characterized in that: The path evaluation module specifically includes: a path length calculation module, configured to quantify the total length of each candidate running route using a continuous path integration method, and output a length value of each candidate running route; a deviation evaluation module, configured to calculate the maximum deviation distance between each candidate running route and the central axis by a projection method, and evaluate whether the candidate running route is aligned with the central axis direction; The posture stability analysis module is used to evaluate whether singular points or posture jitter occur during the execution of the candidate running route based on the posture parameters of the end effector, and output a stability score.

7. A gas intelligent filling system according to claim 6, characterized in that: The specific execution steps of the deviation evaluation module include: Step S11: Project each path point in the candidate running route onto the central axis, and calculate the coordinates of the projection point on the central axis; Step S12: Calculate the distance between each path point and the projection point, and count the maximum deviation distance of the entire path; Step S13: judging whether the candidate running route is aligned with the central axis direction based on the maximum deviation distance, wherein if the maximum deviation distance is less than a preset threshold, it is judged to be aligned; Step S14: output the maximum deviation distance and fit evaluation result of each candidate running route.

8. The intelligent gas filling system according to claim 5, characterized in that: The path optimization module specifically includes: The path jitter elimination module smoothes the path segments with high-frequency jitter in all candidate running routes; A redundant segment elimination module, configured to eliminate repeated or invalid path segments from all candidate running routes; The dynamic response optimization module performs time parameterized adjustment on the candidate running route in combination with the joint acceleration and angular velocity constraints of the robotic arm.

9. The intelligent gas filling system according to claim 8, characterized in that: The specific execution steps of the dynamic response optimization module include: Step S21: converting the joint angular velocity and joint acceleration constraints of the robotic arm into a mathematical model, and defining the feasible motion boundary of each optimized candidate running route; Step S22: Based on the mathematical model, time scaling is performed on the candidate operation route to adjust the execution time of each candidate operation route segment; Step S23: Verify whether each candidate running route satisfies all the joint angular velocity and acceleration constraints. If there is a conflict, trigger an alarm and return to a safe state.

10. A filling method for a gas intelligent filling system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 101: collecting a gas cylinder posture image and a bottle mouth position image in the gas cylinder posture image; Step 102: Filter out a plurality of positioning points according to the bottle mouth position image, and determine a central axis based on the plurality of positioning points; Step 103: deploying a plurality of end effector operation routes for the robotic arm based on the central axis, and determining an optimal target route among the operation routes; Step 104: The end effector moves the inflatable joint to the bottle mouth position and connects it using the optimal target route as a movement path.

Images