Research On Robot Welding Trajectory Correction Method Based On Two-dimensional Laser Scanning Imaging

Teaching and playback is the main mode of industrial robot’s application field, being lack of flexibility during its process of operation. When there are deviation and rotation between the actual location of the workpiece and its standard position used to teaching the welding trajectory, if not to modify the teaching welding trajectory, the robot cannot to welding of artifacts ideally, it will appear the following phenomena such as welding wandering 、 welding incompletely and even collision. It is necessary to use sensing equipment to detect the amount of deflection between the actual position of workpiece and its standard position and design trajectory correction model and algorithm for modifying the teaching welding trajectory to achieve the purpose of robot welding automatically. Considering the harsh working environment of industrial welding workshop,this paper uses two-dimensional laser sensor with strong anti-interference to replace the conventional CCD, etc. imaging equipment to detect and locate the artifacts, develops a system of robot welding trajectory correction based on laser vision sensing.Firstly,this paper performs kinematics analysis for Yaskawa NX100 industrial robot with six degrees of freedom, discussing the transformation relationship between the each joint coordinate of robot and the description of gesture in robotics, laid the foundation for later work.The second step,this paper studies the hand-eye calibration of the two-dimensional laser sensor, proposes simplified single point calibration algorithm innovatively,based on this, puts forward the single points calibration algorithm, solves the transformation relationship between the measuring coordinate system of Laser sensor and the coordinate system of the robot end flange.The third step, studying the principle of laser scanning imaging technology, laid a theoretical foundation for the follow-up work. Through putting forward the processing algorithms for point cloud data, rules out the interference of t noise data, interpolation makes the objective accuracy, precision and density of laser scanning sampled data improved. Proposing the algorithm to generate the binary image according to the point cloud data reflecting the 3d morphology, achieves the purpose of imaging for work pieces,detects the amount of work pieces’ s deflection by image matching.The fourth step, establishing the model of welding trajectory correction.Putting forward a set of algorithms, through the conversion between coordinate systems, successfully use the amount of deflection detected by machine vision to correct the teached trajectory of robot.At the end of this paper, the experiments are carried out to verify the algorithms described above, to prove their feasibility and correctness.