High-Tech for Farm Equipment
Yield Increase and Quality Improvement in Robot Welding by Usage of Laser Vision Systems
- Fig. 1: Laser spot vision sensor seam finding on enclosure with close up of the sensor system mounted to the robot (Picture courtesy ofCarl Cloos Schweisstechnik GmbH)
- Fig. 2: Laser line sensor seam finding on aluminum pallet
- Fig. 3: On a farm tiller welded component, a total of about 60 welds per part having an average length of 35 mm, are done in a robotic gas metal arc welding cell
- Fig. 4: Sequence from start to finish for seam find and weld cycle
The biggest cause of underperforming arc welding robots is the variability in the joint location. Traditionally the most common method used to correct for this variability is to use the weld wire, torch nozzle or separate probe to search for the joint and send the correction values to the robot to offset the pre-programmed path. Although this method can work in some situations, it can never achieve full optimization because it is very slow and inaccurate. In addition, on materials like aluminum or coated steel it is totally ineffective. Laser vision sensing is five times faster and also five times more accurate which can dramatically reduce non added value search time while improving finished weld quality immensely.
The seam finding technique is basically the same no matter which robot type is integrated to the laser vision sensor system. The seam finder can either employ a spot or line laser, depending on the exact application requirements.
High Speed with Laser Spot Sensor
The spot laser vision sensor shown in figure 1 is integrated very easily to any robot. The unique robot faceplate mounting allows the laser to reach any place the weld wire can, thus providing maximum access and flexibility. Just like with touch sensing, the robot is moved across the joint to locate and measure it. However unlike when the weld wire is used, the laser can be moved extremely fast which drastically reduces cycle time wasted. All standard robot programming deployed when doing touch sensing can be used with laser sensing thus simplifying doing a retrofit.
Long Standoff with Laser Line Sensor
The laser line sensor is typically mounted as shown in figure 2. This system uses a laser line with a field of view of about 40 mm (width) x 100 mm (depth) and a resolution of 50 microns. In addition it has a long standoff which allows one to mount the camera far from the welding arc if necessary so any tooling around the part can be avoided. These capabilities mean that any joint (tight butt, thin sheet lap fillet, groove, etc) that is within the field of view of the laser sensor can be located in less than 0.5 seconds. In addition, joint features like gap, area and mismatch can be measured and the values used to either prevent welding from starting if they are out of specification, or alternatively selecting a different weld schedule to handle the changed condition.
This combination of precise electrode location with respect to the joint and the ability to adapt the welding parameters to the actual joint condition maximizes productivity (faster travel speed) and quality. See figure 4 which explains further the sequence of seam finding from start to finish.
Increase of Productivity and Quality
An excellent example of an application that benefited tremendously from the use of laser vision seam finding is the manufacturing of farm tiller components. These are produced in a robotic Gas Metal Arc Welding cell with a total of about 60 welds per part having an average length of 35 mm. The productivity and quality levels were reasonable, but in the new economic environment this company finds itself in, more parts per hour and reduced repair rates are required from every piece of equipment they operate. The biggest place to improve in this particular robotic welding station was to reduce the time spent doing seam finding which totaled about 4.2 minutes. The touch sensing was used to try to handle the stack-up of part and fixture variability. The reason the touch sensing took so long was that the sequence required first finding the bottom of the Tee joint and then going over to find the upper part of the joint. This search routine was done at a very slow rate of 0.2 m/min because at any faster speed the search would be very inaccurate due to overshoot and wire bending. In addition, due to distortion concerns, the part was welded in sections. This meant that after touch sensing the robot would weld a section of the tiller component. Then the robot would have to move over to the wire cutter to cut it off so that the wire end was sharp to improve touch sensing performance. This extra operation wasted even more time. With the laser line sensor, this 4.2 minutes of searching was reduced by 2/3 which resulted in a big time saving. In addition, the laser searching was much more precise which allowed the travel speed to be increased 20 % and the reject rate reduced 10 %. All these improvements added together resulted in the ability to produce 70 more parts per week from the same robot cell. The payback on the installed laser vision system was less than six months.
Every company in the world needs to deal with the new economic reality requiring greater efficiencies from every piece of equipment and every person on the production floor. It is a reality that in welded components there will many times be excessive variability in the detailed parts making up the assembly and in the fixturing used to put the assembly together. For this reason, some type of sensor must be added to the arc welding robot to allow it to accommodate for this variation, either batch to batch or even part to part. Laser vision sensing, both seam finding as discussed in this paper, and real time seam tracking with adaptive control which is used for longer welds, must be employed to achieve optimized welding. This optimization will achieve maximum productivity and quality, while also helping companies become less wasteful and thus greener.