Method and system for automatically extracting track of upper surface robot

Abstract

The invention discloses an upper surface robot track automatic extraction method and system, and the main working principle comprises the steps: 1), collecting images of an upper surface at a plurality of angles through a 3D camera, and generating a plurality of 3D point clouds of the upper surface and a track line thereof; (2) selecting an ROI (Region of Interest) area of each section of trajectory, and extracting the trajectory on the upper surface; 3) converting the plurality of 3D point clouds to the same coordinate system; 4) splicing the 3D point clouds of the multiple sections of trajectory lines to generate a complete robot trajectory; (5) the robot track is sent to a robot workstation, and the robot processes the upper face according to the generated robot track; the method has the advantages that the robot track of the processed shoe upper surface can be quickly generated, and the remodeling production efficiency of an intelligent line is improved.

Description

Technical Field

[0001] This patent belongs to the field of artificial intelligence, machine vision, intelligent manufacturing and robotics. Background Art

[0002] Currently, the robot trajectory generation for the upper glue line is extracted by manual robot teaching. These traditional methods are inefficient and result in a long time for shoe model change production. Usually, the trajectory extraction of a shoe requires manual control of the robot arm to teach and record each trajectory point one by one. This method takes an hour for the robot trajectory extraction, and the time for changeover production of multiple sizes will reach several days. Summary of the invention

[0003] In order to solve the problem of rapid extraction of robot trajectories of shoe uppers, the present invention provides an automatic extraction method and system for robot trajectories of shoe uppers, and the main working principles include: 1) a 3D camera captures images of the shoe upper at multiple angles to generate multiple 3D point clouds of the shoe upper and its trajectory lines; 2) an ROI area of ​​each trajectory line is selected to extract the trajectory lines on the shoe upper; 3) multiple 3D point clouds are converted to the same coordinate system; 4) the 3D point clouds of multiple trajectory lines are spliced ​​to generate a complete robot trajectory; 5) the robot trajectory is sent to a robot workstation, and the robot processes the shoe upper according to the generated robot trajectory.

[0004] The advantage of the present invention is that the robot trajectory of the processed shoe upper can be quickly generated. The robot trajectory of a shoe can be extracted within 1 minute, and the accuracy of the robot trajectory is within 0.5mm, thereby greatly improving the production efficiency and production quality of the intelligent line.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] The patented system comprises: the system comprises: 1) a 3D camera and its multi-angle shooting mechanism; 2) a trajectory extraction computer; 3) a robot workstation, wherein the trajectory extraction computer is connected to the 3D camera and the robot workstation via a network.

[0007] The 3D point cloud acquisition method of the 3D camera and its multi-angle shooting mechanism is to collect the pre-drawn rubber lines on the upper surface around four positions of the upper surface, and then generate the robot trajectory; preferably, in order to ensure the accuracy of extracting the robot trajectory and reduce the influence of the error of the mechanism movement on the trajectory accuracy, four fixed cameras are deployed around the upper surface;

[0008] Preferably, in order to reduce costs, a 3D camera can be used, and a precisely designed rotating motion mechanism can be used to drive the 3D camera to four photo taking positions to collect the 3D point cloud of the shoe upper surface.

[0009] Furthermore, for the stitching of the four point clouds collected at the four positions, it is preferred that the camera coordinates of the four cameras are uniformly converted to the robot coordinate system, and hand-eye calibration is performed in advance to obtain the conversion parameters of the four cameras to the robot coordinate system: XYZ coordinates, and rotation angles around the XYZ axis; the trajectory extraction computer converts the four point clouds to the robot coordinate system according to the conversion parameters from their respective camera coordinate systems to the robot coordinate system, and then completes the stitching of the robot trajectory line 3D point cloud, and the trajectory calculation software generates a complete robot trajectory according to the rules; the robot trajectory is composed of multiple trajectory points located on the robot trajectory line, and each trajectory point contains: 3D coordinates (X, Y, Z) and robot posture (Rx, Ry, Rz).

[0010] Preferably, the trajectory calculation software takes robot trajectory points at equal intervals on a continuous trajectory line 3D point cloud according to a rule; preferably, the trajectory calculation software takes robot trajectory points on a continuous trajectory line 3D point cloud according to a rule according to the characteristics of the shoe shape rubber line; preferably, the trajectory calculation software sets the robot posture for each trajectory point according to the characteristics of the shoe shape.

[0011] The trajectory extraction computer includes: 1) a communication interface connected to a 3D camera to control the 3D camera to take photos and receive image information collected by the 3D camera; 2) an upper surface ID identification sensor to read the upper surface ID information; 3) a trajectory calculation software to complete the 3D point cloud collection, splicing and trajectory extraction of the upper surface and its trajectory line; 4) a robot trajectory database to store the corresponding robot trajectory according to the upper surface ID information; 5) a communication interface connected to a robot workstation to send the upper surface robot trajectory to the robot workstation.

[0012] Preferably, the robot trajectory corresponding to the shoe upper surface ID information storage means that each shoe upper surface ID information includes: 1) shoe model information; 2) size; 3) left and right markings; 4) unique code.

[0013] During the production process of the robot workstation, each upper surface on the intelligent line uses an ID information. The robot workstation reads the ID information of the current upper surface to be processed, reads the robot trajectory corresponding to the ID from the robot trajectory database, and then performs processing.

[0014] Preferably, an RFID chip is used to store the ID information of the upper surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of the working principle of this patent

[0016] Figure 2 To apply the first embodiment of this patent, the robot trajectory automatic extraction system

[0017] Figure 3Using the second embodiment of this patent, the trajectory line extraction method corresponding to the glue line DETAILED DESCRIPTION

[0018] Attached Figure 1 Schematic diagram of the working principle of this patent

[0019] 101. The 3D camera collects images of the rubber line of the upper surface at multiple angles. The rubber line is drawn in advance, and the rubber line is imaged as the trajectory line. The robot trajectory generated by the system coincides with the rubber line. This embodiment uses a high-resolution 3D camera, and the error between the automatically generated robot trajectory and the pre-drawn rubber line is less than 0.1mm;

[0020] 102. Extracting multiple 3D point cloud trajectory lines on the upper surface in the ROI area: First, collect them in the 3D modeling workstation; Second, set up a 3D camera in the intelligent line robot workstation environment to collect 3D point cloud features of the upper surface;

[0021] 103. In the same coordinate system, multiple segments of trajectory line 3D point clouds are spliced ​​to form a continuous trajectory line 3D point cloud. The trajectory points are determined by the trajectory calculation software on the continuous trajectory line 3D point cloud, and then the robot trajectory is generated. In this embodiment, the coordinate system of the four cameras is converted to the robot coordinate system;

[0022] 104. Enter the robot trajectory generated in 103 into the trajectory database according to the shoe upper ID information;

[0023] 105. During the production process, the robot workstation reads the ID information of the shoe upper to be processed, and obtains the robot trajectory of the shoe upper from the trajectory database using the ID information as an index.

[0024] Attached Figure 2 :For the first embodiment of the present invention, the robot trajectory automatic extraction system

[0025] This embodiment is composed of four 3D cameras fixed at four different angles, a shoe upper template station and a computing device. Figure 2 Generate a robot trajectory working environment design diagram for the 3D point cloud collection of the upper glue line in the shoemaking scene.

[0026] 201.4 3D cameras set at multiple angles collect 3D point clouds of the upper rubber line and transmit the collected point cloud information to the trajectory extraction computer;

[0027] 202. Upper placement station;

[0028] 203. The trajectory extraction computer reconstructs the point cloud of the upper glue line in 3D, generates four ROI areas for the point cloud of four angles, extracts the precise 3D point cloud of the glue line, generates the robot trajectory, establishes the robot trajectory database based on the upper ID information as an index, and transmits it to the robot workstation on the intelligent assembly line;

[0029] 204 RFID reading and writing device writes the shoe upper surface ID information for each shoe upper surface.

[0030] Figure 3 Using the second embodiment of this patent, the trajectory line extraction method corresponding to the glue line

[0031] In this embodiment, a 3D camera collects the black glue lines on the upper surface drawn on the white masking sticker in advance, and accurately extracts the 3D point cloud of the glue lines to form trajectory lines, thereby generating the robot trajectory.

[0032] 301 extracting the ROI area of ​​the trajectory line, the trajectory calculation software marks the ROI area near the trajectory line for the collected trajectory line 3D point cloud image, the trajectory calculation software extracts the 3D trajectory line coordinates of the trajectory line in the ROI area, and then generates the robot trajectory based on the trajectory line;

[0033] 302 The glue line drawn manually on the masking sticker on the upper surface is a continuous black line. The 3D camera generates a trajectory line 3D point cloud after acquisition.

[0034] The robot trajectory automatically generated by the 303 system is a robot trajectory composed of multiple trajectory points, each of which contains three X, Y, and Z coordinates and three Rx, Ry, and Rz postures.

Claims

1. A method and system for automatically extracting the trajectory of a shoe upper robot, the system comprising: 1) a 3D camera and its multi-angle shooting mechanism; 2) Trajectory extraction computer; 3) Robot workstation; The system working process includes the following steps: Step 1: The 3D camera collects images of the upper surface at multiple angles, generates multiple 3D point clouds of the upper surface and its trajectory, and transmits them to the trajectory extraction computer; Step 2: On the trajectory extraction computer, select the ROI area of ​​each trajectory line and extract the trajectory line on the upper surface; Step 3: Convert multiple 3D point clouds to the same coordinate system and splice the 3D point clouds of multiple trajectory lines to generate the robot trajectory; Step 4: Send the robot trajectory to the robot workstation, and the robot processes the upper surface according to the generated robot trajectory.

2. The system according to claim 1, characterized in that The 3D point cloud acquisition method of the 3D camera and its multi-angle shooting mechanism at least includes: 1) four 3D cameras at fixed positions surround the upper surface; 2) a 3D camera moves around the upper surface to four fixed positions; 3) a camera is fixed and the upper surface rotates to four positions.

3. The method and system according to claim 1, characterized in that: Converting multiple 3D point clouds to the same coordinate system and stitching 3D point clouds of multiple trajectory lines refers to converting multiple camera coordinate systems at different positions of 3D cameras to the same coordinate system, and converting 3D point clouds of multiple trajectory lines to the same coordinate system, thereby completing the stitching of multiple trajectory lines. The method at least includes: 1) calibrating multiple camera coordinate systems to the coordinate system of the robot workstation; 2) selecting a fixed position of a camera, and calibrating the coordinate systems of other multiple cameras to the coordinate system of this fixed camera.

4. The system according to claim 2, characterized in that Selecting the ROI region of each track line and extracting the track line on the upper surface means covering the ROI region with the track line drawn in advance on the upper surface collected by the 3D camera.

5. The system according to claim 1, characterized in that The trajectory extraction computer includes: 1) a communication interface connected to a 3D camera to control the 3D camera to take photos and receive image information collected by the 3D camera; 2) an upper surface ID identification sensor to read the upper surface ID information; 3) a trajectory calculation software to complete the 3D point cloud collection, splicing and robot trajectory extraction of the upper surface and its trajectory line; 4) a robot trajectory database to store the corresponding robot trajectory according to the upper surface ID information; 5) a communication interface connected to a robot workstation to send the upper surface robot trajectory to the robot workstation.

6. The system according to claim 5, characterized in that Storing the corresponding robot trajectory according to the shoe upper surface ID information means that each shoe upper surface ID information includes: 1) shoe model information; 2) size; 3) left and right marks; 4) unique code.

7. The system according to claim 1 and claim 5, characterized in that Processing the shoe upper surface according to the shoe upper surface ID information and the robot according to the generated robot trajectory means that the robot workstation reads the shoe upper surface ID information to be processed currently, and reads the robot trajectory corresponding to the ID from the robot trajectory database.

8. The system according to claim 5, characterized in that The storage carrier of the shoe upper ID information includes: 1) RFID chip; 2) QR code image.

9. The system according to claim 1 and claim 5, characterized in that The method of generating robot trajectory by trajectory calculation software refers to determining the robot trajectory points on the trajectory line 3D point cloud, and then generating a complete robot trajectory. The method of determining the robot trajectory points includes: 1) determining the robot trajectory points according to the characteristics of the shoe type rubber line; 2) selecting trajectory points at equal intervals; 3) setting the robot's working posture for each trajectory point.

Images