Research On A Rotational Mirror 3D Laser Scanning Vision Measurement-Based Robotic Vision Inspecting System

The welding technique is one of the four body manufacturing process,and welding dimension quality control is the basis for assuring the subsequent general assembly quality,thereby affecting the overall quality of the automobile.Generally,measuring points which reflect the assembling function are chosen to be measured,and the welding dimension quality is evaluated through the measuring data.The measuring data combined with welding dimension quality control method are used to improve the consistency of BIW and enhance improve the quality of BIW.Although the traditional quality control systems play a very important role in assuring the BIW manufacturing consistency and enhance the manufacturing quality,the traditional methods cannot satisfy the in-line quality control requirement of the high yield,high speed and high flexibility of BIW welding production line at the same time.The robotic vision inspecting system faces the following challenges: a 3D vision system with field scanning function is developed to measuring a variety of geometrical features,what is more,it should also satisfy the real-time requirement of on-line inspection.The complex measuring environment results in the nonuniform optical surface,and the extraction accuracy and robustness of the measured features in BIW is still a problem.In the actual production process,the inspection accuracy of the error-compensated robotic vision inspecting system is affected due to self-heating and change of temperature,and the maintaince of the inspecting accuracy of the robotic vision insepecting system in the actual production process is a problem to resolve.Aim to meet the demand of the BIW welding production line,the paper studies the theory and technology of the robotic vision inspecting system,and focus on the mathematical model and calibration method of the rotational 3D laser scanning vision measurement system,the in-line extraction algorithm of the BIW assembly features,the in-line error compensation technology for the robotic vision inspecting system,determination of optimal measurement configurations for self-calibrating the robotic vision inspecting system with multiple point constraints and the integration application between the robotic vision inspecting system and the BIW welding production line.The paper presents a rotational mirror 3D laser scanning vision measurement system which uses a rotational reflective mirror as its mechanical device to drive the laser stripe to sweep along the object,and then the function map between the rotational angle and the reflective laser stripe plane is established mainly based on POE and the reflection theory.The parameters of the function map are identified by minimizing the distance from the control points to the corresponding reflective laser stripe plane.Finally,the calibration method of the rotational 3D laser scanning vision measurement system is proposed on the basis of the two-step calibration method.The experimental results verify the effectiveness of the proposed rotational 3D laser scanning vision measurement system and the calibration method.The paper proposes a density-based spatial clustering assembly features extraction algorithm for the BIW.The algorithm first transforms the fitting point cloud to the parameter space of the fitting feature and select the parameter which is nearest to the center of the largest cluster by using the DBSCAN clustering algorithm as the initial identified parameter.Outliers are removed from the fitting point cloud by using the gross error evaluation standard.The geometric fitting algorithm is implemented to the point cloud without outliers to obtain the final identified parameters.The experimental results show that the proposed algorithm has high accuracy and robust to input parameters.In order to maintain the accuracy of the robotic vision inspecting system during the real production line,the paper proposes the in-line error compensation method for the robotic vision inspecting system.Firstly,the error compensation model of the robotic vision inspecting system is established mainly based on the MD-H model and the error compensation model contains the hand-eye relationship,the robot itself and the robot exterior relationship.The redundant parameters are eliminated after analyzing the structure of the robotic vision inspecting system,and the error constraint equations are established by measuring the spherical calibration targets which are placed within the measurement envelope of the inspecting system.The proposed error compensation method is a self-calibration method and it can meet the requirement of calibrating the inspecting system automatically.The mathematical model contains all the kinematic parameters and identified them in one step to avoid error propagation rather than identified the hand-eye relationship,the robot itself and the robot exterior relationship separately.The paper analyzes and eliminates redundant kinematic parameters in the kinematic model to enhance the efficiency and accuracy of parameter estimation.The experimental results show that the accuracy of the robotic vision inspecting system can be controlled within 0.3mm after error compensation.The optimal measuring configurations of the robotic vision inspecting system are constrained by multiple points and it cannot be solved using the existing optimal configuration chosen algorithms.To solve the aforementioned problem,the paper proposed a determination of optimal measurement configurations for self-calibrating a robotic vision inspecting system with multiple point constraints.Firstly,the measuring configuration set with one point constraint is obtained using the robot inverse kinematic algorithm which is based on the screw theory.The best combination of measuring configuration is chosen from the measuring configuration set using DETMAX algorithm.The spherical calibration targets can be placed anywhere in the measuring envelope of the inspecting system,so the PSO algorithm is used to optimize the position of the spherical calibration target one by one.The contrast experiment results show that the error compensation accuracy of the proposed algorithm is 18.6% higher than that using the randomly selected measurement configurations.On the basis of the aforementioned theory and technology research,the robotic vision inspecting system is developed by ourselves and integrated with the BIW welding line for quality control.After finishing a variety of precision test,the quality of parts and fixture adjustment are analyzed through the measuring data using the robotic vision inspecting system to verify the significance of the inspecting system to improve the quality of the BIW.