With the rapid development of industry 4.0 and the rapid development of intelligent manufacturing, the flexible automatic assembly technology of aircraft based on industrial robots has attracted extensive attention . Among them, AGV (automated guided vehicle) mobile drilling robot has become a research hotspot because it can better adapt to the characteristics of large size, multi variety and small batch of aircraft products. The industrial robot itself has low absolute positioning accuracy. At the same time, the AGV carrying the robot has elastic deformation, which is directly related to the robot's position and posture, and is also affected by the robot's motion speed, motion path, kinematic posture, load deformation  and inertia, so large errors will be introduced at the robot's mounting base At the end, it will be further enlarged after the transmission of each link of the robot. As a result, the absolute positioning accuracy of the system is far from meeting the accuracy requirements of the aircraft automatic assembly system in the offline programming mode . Therefore, the research on precision compensation technology to improve the absolute positioning accuracy of the system is the key to the application of AGV type mobile drilling robot in flexible automatic assembly for aviation manufacturing [5-6].
The traditional precision compensation technology, such as robot kinematics calibration , is based on the error model of robot kinematic parameters. Using the limited positioning error sampling data, the kinematic parameter errors of each link of the robot are identified by optimization algorithm . The Levenberg Marquardt (L-M) method is widely used . However, the error model only includes the geometric error sources of the robot. When more error sources need to be included, more error parameters must be added. For AGV type mobile drilling robot, it is very difficult to describe the change of AGV elastic deformation by mathematical model, which will lead to the increase of computational complexity and computational complexity.
In view of the shortcomings of kinematic parameter calibration methods, many scholars have explored other precision compensation methods to improve the absolute positioning accuracy of the robot [10-12]. Zeng et al. [13-14] established a spatial mapping model between robot positioning error and joint rotation angle by using the similarity of robot positioning error, and then linear unbiased optimal estimation of compensation point was conducted. This method only focuses on the relationship between the end positioning error and the joint angle input, and does not pay attention to the specific error source, so it has strong universality to a certain extent. However, this method has some limitations in AGV type mobile drilling robot. The limitation is that when the end positioning error caused by AGV elastic deformation increases to a certain extent, the mapping model depending on robot joint angle may be difficult to establish.
Other methods to improve the positioning accuracy of robots, such as online error compensation [15-16], need to install a real-time feedback device at the end to adjust the robot's position and posture, which usually can obtain higher compensation accuracy. However, the installation device is not easy to operate in some complex scenes, and the cost of the installation is higher than that of offline calibration technology. Therefore, under the background of multi forms and rapid development of robot application, it is urgent to innovate the robot precision compensation technology. Referring to the robot positioning error compensation method based on the inverse distance weighting method proposed by Zhou Wei et al. [17-18], using the similarity of positioning errors, the AGV type mobile drilling robot is regarded as a "black box", which only cares about the relationship between the end position input and the positioning error output, and estimates the positioning error of the compensation point with the known similarity point positioning error.
Based on the spatial similarity of positioning error, this paper discusses a space interpolation and compensation method based on inverse distance weighting, which is suitable for AGV type mobile drilling robot. Through the proposed method of changing station in frame coordinate system of AGV mobile drilling robot, the sampling data of precision compensation station can be used in other processing stations. The drilling system of KUKA kr480 industrial robot equipped with AGV is tested and verified. The optimal grid step size is selected through the test. The compensation results show that the method can compensate the influence of AGV elastic deformation on the robot end positioning error, and improve the absolute positioning accuracy of AGV mobile drilling robot.
1 AGV mobile drilling robot system
1.1 System workflow
The AGV mobile drilling robot system is shown in Figure 1, and the system workflow is shown in Figure 2.
The hardware part of the system is mainly composed of KUKA kr480 industrial robot, AGV, multifunctional end effector and other auxiliary equipment. The system software part is mainly divided into offline programming software and integrated control software. The precision compensation algorithm in this paper is integrated into the integrated control software. The work flow of the system is as follows: firstly, the off-line programming software generates NC program which can be analyzed by the integrated control software, then the integrated control software controls the AGV to locate to the station to be processed and carries out the benchmark detection. Then, the precision compensation is made for the hole position to be processed. After the robot is in place, the end effector will compress and make holes, and cycle until all the processing tasks are completed.
1.2 System positioning error
In order to observe the elastic deformation degree of AGV and the end positioning error of the robot, the measuring rod is installed on the electric spindle of the end actuator, and the target ball seat is installed at the other end to fix the target ball of the laser tracker (target ball 1 in Figure 1); a target base is fixed at the robot base, and the fixed target ball (target ball 2 in Figure 1) is used as the observation point of AGV deformation. When the robot is in the default home position, the observation is measured The three-dimensional coordinates of the measuring points are (383.236, - 287.460, 63.901) mm. The coordinate values in this paper are based on the robot frame coordinate system. KUKA industrial robot stipulates that the origin of the frame coordinate system is located in the center of the base, the X direction points to the positive direction of the robot, the Z axis is vertical upward, the Y axis is determined by the right hand rule, and the frame coordinate system determines the specific position and posture of the robot.
478 sampling points are randomly selected in the working space of practical engineering application of the system. The sampling points are in X Y Z, and the range of direction values is shown in Table 1. Take the theoretical value of the sampling point as the input, drive the robot to move at a speed of 10%. Use the laser tracker to measure the end positioning error and the three-dimensional coordinates of the observation point. The data statistics are shown in Table 1, and the deviation line diagram of the three-dimensional coordinates of the observation point is drawn as shown in Figure 3.
Please download the rest