KR19980078507A - Welding bead removal method of grinding robot - Google Patents

Abstract

The present invention relates to a method for removing a welding bead by a grinding robot, wherein a contour laser sensor is mounted on a grinding robot to synchronize the robot with the sensor, and then the sensor is moved by the robot, ; ≪ / RTI > Modeling the body surface using the measurement points; Calculating a posture and a position of the robot with respect to the position of the weld bead using the modeled calculation value and grinding the weld bead by controlling the robot using the calculated value, Since the modeling is performed, it is possible to calculate the position and attitude information necessary for using the robot regardless of the shapes and sizes of the body and beads of various kinds, and the information of the body surface is inputted in a noncontact manner using the laser sensor. There is no time and difficulty in teaching and it is possible to carry out the grinding work of the body surface requiring high precision without the existing worker completely.

Description

Welding bead removal method of grinding robot

BACKGROUND OF THE INVENTION Field of the Invention [0001] The present invention relates to a method for removing welding beads of a grinding robot, and more particularly, to a grinding apparatus for grinding a workpiece without removing a welding bead, To a welding bead removing method of a grinding robot.

Conventionally, the removal of weld bead on the body of a car is performed manually by most of the experience and skill of the operator. The grinding operation requires a high-precision work technique, and the dust generated during the operation of the flame and grinding occurring during the work is very It has a bad influence. In addition, this work is a very simple repetition work, which makes the worker become tired easily, which is a big factor to make the shape of the body bad, and the productivity is lowered due to frequent rest and replacement of the worker.

Therefore, it is required to develop a grinding robot system for removing automobile body weld bead using a robot.

Conventionally, a method using a robot has mainly been performed by predicting a posture and a position to be actually taken by the robot and repeating the operation.

This is because most automation lines using robots in industry are simple repetitive assembly lines or welding lines. In case of assembling or welding line, it is possible to carry out assembly or welding process regardless of object even if moving the robot in a high position and attitude by teaching in advance. If the shape of object changes and new posture and position are needed, You can teach. The assembly or welding process is advantageous in that it can be used even if it is difficult to teach the robot at first because it can be used without changing the posture and position of the robot once and for a long period of time. .

Another method is to input the measured force signal using the F / T sensor and change the attitude and position of the robot according to the input signal. This makes it possible to change the position and position of the robot more autonomously than the robot teaching method described above.

However, in the method of teaching and controlling the position and the posture of the conventional robot described above, it is suitable for a simple assembly system, but it is very difficult to implement the method in the above-described method in a system for actually processing an object, not an assembly system. The reason for this is as follows.

First, it can not be guaranteed that the object to be machined always comes to a relatively fixed position with respect to the robot. In fact, on the factory production line, the bodywork is moved by the jig that carries the bodywork, and this jig has its own position and orientation error. Therefore, it can not be assumed that the beads to be machined always come in the same position and direction. Therefore, the desired beads can not be removed by teaching the position and orientation of the robot in advance.

The second bead line welded is different for every car body, and the amount of beads left on the car body surface is also different. Therefore, even if the body of the same shape is used, the attitude and position of the different robots are required to remove the beads. It is not possible to teach another robot every time for this purpose, and it is also impossible to remove the bead within the working time of a real car production line. Therefore, the conventional robot teaching method can not perform the desired operation.

Third, the process of removing the weld bead on the car body surface is a very important process that directly affects the quality in the automobile production line. The beveling bead on the outer surface of the vehicle body surface must be accurately removed along the vehicle body surface so that the shape of the produced vehicle is correct. If the grinding process is not properly performed, the vehicle can not properly shape the vehicle do. Therefore, the quality of the car body surface should be very good as a result of removing the bead. Therefore, in the manual process by the worker in the current automobile line, very high skill and experience of the worker are needed. For such a high - precision shape, it is necessary to input the robot 's attitude and position, which is different from the teaching used in the conventional simple repetition work.

In the method using the F / T sensor, since the robot and the object must be in contact with each other, the object may be deformed and a new algorithm using the sensor should be applied as the shape of the object changes. The algorithms using the F / T sensor vary greatly in difficulty depending on the shape of the object and the work process, and additional processing of undesired additional forces such as gravity, not the force to be measured, is additionally required.

In addition, automobile production lines are very expensive equipment, and various types of automobiles are produced using this production line. Therefore, when developing an automation system that runs on the actual production line, it should be applicable to various types of automobile bodies. In other words, it is not necessary to teach the position and position of different robots according to the automobile system, but the same robot system and algorithm should be applied to different bodies.

Therefore, the object of the present invention is to grind and remove the welding bead, so that the modeling is performed every time the vehicle body enters, so that the position and attitude information required when using the robot can be calculated irrespective of the shape and size of various kinds of body and beads Since the information of the body surface is input in a non-contact manner using the laser sensor, there is no time and difficulty in the conventional teaching, and the grinding operation of the body surface requiring high precision can be performed without any existing worker And to provide a method for removing weld beads of a grinding robot.

1 is a view showing a shape of a body surface to be measured and a position of a measurement point input from a sensor.

Fig. 2 is a view showing a coordinate system for calculating the posture of the robot. Fig.

3 is an actual configuration diagram of the controller unit;

4 is an actual structural view of an entire body grinding system.

5 is a general flow chart of a vehicle body grinding system.

DESCRIPTION OF THE REFERENCE NUMERALS

10: main processor unit 20a, 20b: grinding system

21a, 21b: CPU 22a, 22b: DSP board

30a, 30b: robot controller Ra, Rb: grinding robot

Sa, Sb: Laser sensor

In order to accomplish the object of the present invention, there is provided a method of controlling a grinding robot, comprising the steps of: mounting a contour laser sensor on a grinding robot to synchronize a robot and a sensor; measuring coordinates of an arbitrary point on the body surface while moving the sensor; Modeling the body surface using the measurement points; Calculating a posture and a position of the robot with respect to a location of the weld bead using the modeled calculation value, and controlling the robot using the calculated value to grind the weld bead. A removal method is provided.

In the step of measuring the coordinate values on the vehicle body surface, the measurement is repeated at a predetermined interval of two points, upper and lower, based on the position of the weld bead on the vehicle body surface. In the modeling step, a method of the Ferguson modeler is used. In the step of grinding the weld bead, the robot is moved by grinding at least two times.

Hereinafter, a method for removing a welding bead of a grinding robot according to the present invention will be described in detail.

First, a laser sensor is attached to a grinding robot as a non-contact sensor and is synchronized with the robot. Then, the coordinates are measured with respect to an arbitrary point on the vehicle body surface while moving the sensor by the robot.

The laser sensor has a characteristic that it can measure precisely when an arbitrary shape is measured, and can accurately measure the attitude and position of the object to be measured in an actual factory work environment. Particularly, among the laser sensors, it is preferable to use an outline laser sensor (Contour Laser Sensor) which obtains outline information by using a triangular method.

Specifications of the contour laser sensor are shown in Table 1 below.

Laser type (LASER type) Single laser-plane contour forming optics Optical Array Charge-couple device (CCD) 768 x 482 pixel array Standoff 800mm View Field 79 mm (y) x 167 mm (z) Accuracy 0.05mm or less Resolution 0.02 mm or less

In measuring the vehicle body surface by the laser sensor, since the sensor is a scanning sensor capable of measuring only one line, the upper and lower two points of the welding bead are set at an arbitrary first position, And moving the position by the robot to measure coordinates of four points at a second position spaced apart from the first position in the same manner as in the first position. In this way, the length of the weld bead .

For example, as shown in Fig. 1, when the vehicle body surface 1 is inclined at an arbitrary curvature at both sides and the weld bead 2 is at a length of about 160 mm in the transverse direction, Take 2 points on the upper side and 2 points on the lower side based on the position. Measure the four points of the body surface with one measurement and move the sensor again using the robot.

Basically, we move 16 times with a laser sensor on a car body surface and measure it to write 16 × 4 points to model car body surface later. The time of the laser sensor sensing and the robot's moving time are synchronized in advance.

The body surface is modeled using the Ferguson Modeler method using the measurement points.

That is, the vehicle body surface (1) is modeled using the Ferguson Modeler modeler method for 64 measurement points measured at four points in 16 positions.

Here, modeling the vehicle body surface 1 means calculating the position vector, tangent vector and vector at all internal points.

In the Ferguson surface modeling method, a corresponding unit surface is specified in a whole surface of a compound surface, a position vector, a tangent vector and a twist vector at four corners corresponding to the unit surface are input, and the value of a given parameter (0 ≤ u, v ≤ 1) corresponding to the curvature of the curved surface, and combining the obtained unit curved surfaces to obtain the overall curved surface. The position vector corresponding to the value of the given parameter on the curved surface in the corresponding unit curved surface is obtained by the following equation (1). Where r is the position vector at the measurement point, t is the tangential vector in the u direction, s is the tangent vector in the v direction, x is the twist vector, and u and v are the coordinate system of the unit curved surface. We obtain the tangent vector in the u direction and the tangent vector in the v direction by differentiating the vector r (u, v) on the unit vector thus obtained in the u and v directions. Then, the tangent vectors in the two directions are cross- I ask.

In this way, the model of the unit curved surface can be used to obtain the position vector and the normal vector at an arbitrary position on the vehicle body surface. Thus, using the measuring point of the laser sensor, the surface of the vehicle body surface with the weld bead can be obtained every time the bodywork enters the production line. Using these information, we determine the position and position of the robot in the place where the bead to grind is located. As shown in FIG. 2, a curved surface position vector P and a curved surface tangent vector T are defined, and the angle of grinding and the pre- The position of the robot is obtained by the following equation (2).

In order to obtain the attitude of the robot, the K vector is first obtained by the T vector and the N vector as shown in Equation (3), and the rotation matrix of the K vector is obtained as shown in Equation (4).

At this time, .

Using this, the attitude of the robot determines the following equation (5).

In this way, the body surface is modeled using the measurement point by the laser sensor, and the position and attitude of the robot are calculated at the location of the weld bead using the modeled calculation values. The welding bead is grinded and removed by controlling the robot according to the calculated posture and position value of the robot.

Hereinafter, a method of removing a weld bead of a grinding robot according to the present invention will be described as an embodiment.

Table 2 shows the configuration of the sensor of the welding bead removing robot system for grasping the welding bead removing method according to the present invention, the grinding and the control for controlling the robot and the whole system.

A specification table showing the controller configuration of the grinding system. number Name model name Specifications (Spec) Slot Quantity One PC Processor EPC-5 -CPU: 80486DX2 (66MHz) -RAM: 4Mbyte-2 RSC Serial Port-1 Parallel Port-EXM-13 Super VGAGraphic Controller-EXP200-A Hard Disk-EPC Connect S / W 4 1 set 2 PC Processor EPC-5 -CPU: 80486DX2 (66MHz) -RAM: 4Mbyte-2 RSC Serial Port-1 Parallel Port-EXM-13 Super VGAGraphic Controller-EXP200-A Hard Disk 4 × 2 = 8 2 sets 3 Serial Board EXM-8 -2 RS422 Serial Port 1/2 × 2 = 1 (3U size) 2 sets 4 Digiral I / O EXM-19-OPTO -24 Opto Isolated Input / Output-Output: 16 Channel-Input: 8 Channel 1/2 × 2 = 1 (3U size) 2 sets 5 DSP Board P4000 2 x 2 = 4 2 sets

Fig. 3 shows an actual installed image of the controller unit of the grinding robot system. The controller communicates data on the VME-bus. The MPU is responsible for the monitoring of the entire grinding system. The left and right sides of the body are processed by the left and right sensors. The left and right CPUs are connected to the left and right CPUs. Two serial ports And two Digitel I / O ports that accept emergency signals from robots and other PLC lines. And there are Nachi Robot which works with a grinder and an actual grinder, and two robot controllers that move the robot by receiving the position and position information of the left CPU and the right CPU.

The measured data from the P4000 contour laser sensor, which is a measurement sensor, is transmitted to the DSP board through the RS170 line, and the CGs TP, which filters the data on the sensor CCD camera into distance data, and the filtering process, VME Embedded 486 CPU. The measurement sensor used in this device is a Contour Laser Sensor that detects contour information using a triangular technique. The laser beam is projected to form a laser light plate in front of the sensor. The solid state camera observes the plane through the triangular angle and follows the laser line. Measure the position in the Y and Z directions. We use this to model the car body surface and download this information to the controller of the robot using RS232C serial communication. The robot moves to the given position and the grinder starts to work in the calculated posture.

Hereinafter, a specific embodiment of the present invention will be described. FIG. 4 is a diagram showing the entire structure of a grinding robot system for implementing a method of removing a welding bead of a grinding robot according to the present invention. As shown in FIG. 4, left and right grinding systems 20a and 20b are mounted on a main processor unit 10, And connecting the robots Ra and Rb to the grinding systems 20a and 20b through the robot controllers 30a and 30b and connecting the laser sensors Sa and Sb mounted on the robots Ra and Rb, Are connected to the left and right grinding systems 20a and 20b.

The left and right grinding systems 20a and 20b are each composed of CPUs 21a and 21b and DSP boards 22a and 22b and various communication ports such as com1, com2, RS232C, com3 and I / O ports are connected.

Various communication ports such as com1, com2, RS232C, RS422 and I / O ports are also connected to the left and right robot controllers 30a and 30b.

The DSP boards 22a and 22b and the laser sensors Sa and Sb are connected to each other through an RS170 port.

In the drawings, reference numerals LCa and LCb denote load cells, and PLCa and PLCb denote PIEC (programmable logic controller).

Hereinafter, the method of removing the welding bead of the grinding robot according to the present invention will be described in each step.

Step 1

The robots Ra and Rb and the laser sensors Sa and Sb are synchronized with each other to sense and sense the robots Ra and Rb from the CPUs 21a and 21b.

The laser sensors Sa and Sb moved by the DSP boards 22a and 22b transmit the measured distance data to the CPUs 21a and 21b via the VME bus. The CPUs 21a and 21b create an MDI file which is a robot step file at a position for sensing the sensors Sa and Sb and transmits the MDI file to the PC controllers 30a and 30b, Converts it into a machine constant file by a teaching command in the CATS, and downloads it to the robot controllers 30a and 30b. The laser sensors Sa and Sb are moved several times (for example, 16 times) to input measurement points, and when the sensing is completed, the robots Ra and Rb are stopped and the sensors Sa and Sb are stopped.

Step 2

The vehicle body surface shape is modeled from the coordinate values of the vehicle body surface 1 received in the first step and the positions and postures for removing the beads 2 are modeled using the method described above to the robots Ra and Rb Download by RSC232C communication. First, the distance data of the vehicle body surface 1 measured by the laser sensors Sa and Ab is converted into a base coordinate system (Robot Base Coordinate) of the robots Ra and Rb. After all necessary distance data information is measured, the CPU 21a and 21b model the vehicle body surface with the Ferguson Surface Modeler using these distance data.

Step 3

Next, the position and posture of the robot to be moved for grinding are created using a Ferguson modeler, and then converted into a * .SRC file according to the used Nazi robot. This * .SRC file is converted into a machine constants file according to the MDI file format by the teaching command in the PC-CATS of the Nazi robot.

The converted machine constant file is downloaded to the robot controllers 30a and 30b of the Nazi robot and the robots Ra and Rb are moved to grind the weld bead 2 with the grind disk. Since the flexible razor discs can not remove all of the weld beads 2 by moving the robots Ra and Rb once, the robots Ra and Rb are moved four times to grind the weld beads 2. After completion of the grinding, the robots Ra and Rb delete the previously downloaded input values and return to the reference initial position. When a signal for exchanging a flexible grind disk enters the CPUs 21a and 21b, the robots Ra and Rb move to a disk changer (not shown), exchange disks, and return to their original positions.

In addition, check robot operation error when RS-232C serial communication between CPU processor and robot controller, check error when program download and program kill by PC-CATS The CPU receives the interrupt for processing. The MPU processor responsible for monitoring the entire grinding system communicates and communicates the various interrupts and parameters required between the left and right CPUs on the VME bus. Fig. 5 shows the overall flow chart of the body grinding system.

According to the present invention, when grinding and removing the welding beads on the vehicle body surface, the modeling is performed every time the vehicle body enters. Therefore, the position and attitude information necessary for using the robot can be calculated regardless of the shapes and sizes of the various bodies and beads Since the information of the car body surface is inputted by the non-contact method using the laser sensor, there is no time and difficulty in the conventional teaching, and as a result, the car body surface grinding operation requiring high accuracy can be performed without any existing worker will be.

Claims (4)

A step of synchronizing the robot and the sensor by attaching the contour laser sensor to the grinding robot and measuring coordinates of an arbitrary point on the body surface while moving the sensor by the robot; Modeling the body surface using the measurement points; Calculating a posture and a position of the robot with respect to a location of the weld bead using the modeled calculation value, and controlling the robot using the calculated value to grind the weld bead. Removal method. The method according to claim 1, wherein in the step of measuring the coordinate values on the body surface, the weld bead is repeatedly measured at a predetermined interval by two points on the basis of the position of the weld bead on the body surface. The method of claim 1, wherein the modeling step uses a method of the Ferguson modeler. The method of claim 1, wherein the step of grinding the welding bead comprises grinding the robot by moving the robot more than twice.