Objective In the fields of mold manufacturing, automobile assembly and aviation manufacturing, 3D measurement is an important link to verify whether the product shape accuracy meets the design specifications. With the continuous development of technology, manual measurement has been unable to meet the needs of modern production applications. Therefore, automatic measurement technology with higher detection accuracy and measurement efficiency has gradually come to the fore. However, the current automatic measurement method has the problem of lack of data in complex structure areas when dealing with characteristic regions with complex curvature changes, so it is difficult to be directly applied to the scanning measurement of large and complex surfaces. The methods to solve these problems are usually lack of universality, and the path planning method needs to be selected manually, which leads to low scanning efficiency. Therefore, it is necessary to design an efficient automatic measurement method for this kind of complex surface. For this reason, through extensive research, this paper proposes a feature adaptive automatic scanning path planning method based on the scanner's own scanning constraints (Fig.4).
Methods First of all, on the basis of constructing the scanning constraint conditions of the laser scanner, this paper analyzes its influence on the scanning accuracy (Fig.3). Secondly, aiming at the problem of scanner attitude planning, a new measurement inclination constrained clustering algorithm is proposed to cluster the sampling points obtained from discrete complex surfaces to realize scanner attitude planning (Fig.10). On this basis, the scanning path points are obtained (Fig.13). Thirdly, aiming at the scanning path planning problem, the approximate algorithm is improved by introducing the normal vector angle matrix as the penalty matrix, and the scanning path planning that meets the requirements is realized (Fig.15). Finally, taking the car door as the scanning object, the measurement process is as follows: first, according to the measurement task planning, the automatic measurement system enters the preset task station. Several measuring target points are installed on the car door, the relative position and pose of the car door and the flexible measurement system are determined by using the laser tracker, and the standing position of the system is fine-tuned to meet the accuracy requirements of pose estimation. Then the scanning path is generated by the feature adaptive scanning method designed in this paper. The scanning path data is converted to the robot flange coordinate system based on the system coordinate transfer model, and the task instructions that can be executed by the measuring system are obtained. Finally, based on the analysis and processing of the scanning data, the measurement results are obtained, that is, the single complete measurement of the object to be tested is realized (Fig.17).
Results and Discussions The effectiveness of the proposed method is verified by building a flexible measurement system, in which the manipulator is an Erbidi LT1500-C-6 universal robot, its workspace is a spherical area within the range of 1 500 mm around the base joint, the payload 10 kg, and the repeated positioning accuracy is ±0.05 mm. The scanning equipment is Hexcom's Leica T-Scan5 line laser scanner, which uses the Leica laser tracker AT-960M to track and position the scanner. The execution module is decoupled from the measurement module to ensure that the accuracy of the measurement data is not affected by the cumulative error of execution (Fig.16). The experiments are compared with the traditional line-cut scanning method and manual scanning method from the four dimensions of scanning efficiency, scanning accuracy, scanner attitude transformation times and scanning integrity. The experimental results show that compared with the line-cut scanning method, the attitude transformation times of the scanner are reduced by 54% (Tab.4), the measurement accuracy is improved by 64.5% (Tab.6), and the scanning integrity is close to that of the manual scanning method (Tab.2). Automatic scanning measurement for complex surfaces can be realized.
Conclusions A feature adaptive path planning method for automatic measurement of large and complex surfaces is proposed, and the automatic acquisition of 3D measurement data of large surfaces is realized. The attitude of the scanner is planned based on the scanning constraints of the scanner itself, and the scanning trajectory and scanning posture are optimized, which improves the execution efficiency and universality of path planning. Meanwhile, it can maintain good scanning integrity for the areas with complex curvature changes on the measured surface. The proposed method is tested in the design scene, and the traditional automatic measurement method and manual measurement method are compared with the traditional automatic measurement method and manual measurement method from the four dimensions of scanning efficiency, scanning accuracy, scanner attitude transformation times and scanning integrity. The results show that the scanning efficiency of the proposed method is higher than that of manual scanning, and the scanning accuracy and scanning integrity are improved. The number of attitude changes in the scanning process is much less than that of the line-cutting method, and it can replace manual automation to complete the scanning measurement of complex surfaces.
DownLoad: