{"id":374,"date":"2014-08-18T18:01:04","date_gmt":"2014-08-18T22:01:04","guid":{"rendered":"http:\/\/coro.etsmtl.ca\/blog\/?p=374"},"modified":"2014-08-26T10:17:13","modified_gmt":"2014-08-26T14:17:13","slug":"new-low-cost-wireless-3d-position-probe-for-measuring-the-repeatability-of-industrial-robots","status":"publish","type":"post","link":"https:\/\/coroblog.etsmtl.ca\/?p=374","title":{"rendered":"Measuring position repeatability of industrial robots"},"content":{"rendered":"

\"3DPosition repeatability<\/em>\u00a0is one of the most important performance criteria of\u00a0industrial robots.\u00a0The\u00a0ISO 9283:1998<\/a> norm defines it as\u00a0the “closeness of agreement between the attained [positions] after n<\/em> repeat visits to the same command pose in the same direction.”\u00a0Position repeatability\u00a0is the only positioning performance indicator\u00a0that industrial robot manufacturers specify in their brochures and varies between 0.010\u00a0mm and 0.100\u00a0mm. Methods for measuring repeatability\u00a0were presented in ISO\/TR 13309:1995<\/a>. However, metrology has changed since then, so what are today’s methods for measuring position repeatability?<\/p>\n

<\/p>\n

If you are lucky like us\u00a0and have plenty of state-of-the-art metrology equipment, you may use a laser interferometer. The latter is highly accurate (measurement uncertainty of 0.001\u00a0mm), but is expensive (about US$50,000), difficult to set up\u00a0and measures only one coordinate at a time. You can also use a CMM, but that’s obviously not practical. You can\u00a0use various types of 3D measurement equipment, but I\u00a0am pretty\u00a0certain that industrial robot manufacturers use laser trackers since they already use these for calibrating some of their robots<\/a>.\u00a0The problem with laser trackers (other than their prohibitive cost) is that they are not sufficiently accurate to measure position variations of the order of\u00a00.010\u00a0mm. So should we believe the repeatability data supplied by industrial robot manufacturers?<\/p>\n

One of the basic methods\u00a0presented in\u00a0ISO\/TR\u00a013309:1995 relies on the use of three orthogonally\u00a0arranged\u00a0position sensors. I recently bought one such fine\u00a0device: the Trinity probe<\/a> from IBS Precision Engineering. The latter is wireless, light and compact, and not that expensive (about US$15,000). Its measurement uncertainty is about 0.001 mm, and its measurement range is 3.5\u00a0mm. Unfortunately, it uses eddy-current sensors that call for special datum spheres, which cost more than US$500 each and are mounted on slim stems that are very easy to break.<\/p>\n

I therefore\u00a0adopted the idea of\u00a0Prof. Jean-Fran\u00e7ois Breth\u00e9 from Universit\u00e9 du Havre (France), who used three orthogonally placed digital indicators. However, I didn’t like the fact that his indicators are all cabled and opted for a more compact, wireless\u00a0solution.<\/p>\n

Our 3D probe consists of three\u00a0ID\u2013C112X <\/a>(543-390B) digital\u00a0indicators, three wireless transmitters U\u2013WAVE\u2013T<\/a> and\u00a0one U\u2013WAVE\u2013R<\/a> receptor, three U\u2013WAVE cables, and the U\u2013WAVEPAK software, all from Mitutoyo. It also consists of three 0.5″ magnetic nests<\/a> from HUBBS, three V\u2013blocks and a hollow 0.5″ precision sphere from Bal-tec, and two custom precision machined supports from aluminium. The cost of the complete device, including machining, is less than US$6,000. The device, including an adaptor and a QC\u20135 tool changer weighs only 1.2\u00a0kg.<\/p>\n

A very important\u2014and novel\u2014component of our 3D probe is its calibration plate. The latter is essentially a kinematic platform using the principle of three spheres on three V-grooves. It allows us to position over and over again a 0.5″ precision\u00a0sphere in exactly the same place with respect to the 3D probe and set the zero for all three indicators.<\/p>\n

We tested our 3D probe on a FANUC LR Mate 200iC industrial robot (lent to us by GE Aviation) and on an ABB IRB\u00a0120 robot. For our test, we use a custom-made pivoting base with three 0.5″ datum spheres, each pair separated about 300 mm apart. Using very simple Matlab code, we are able to retrieve the position data from each digital indicator and then send it to the robot controller via Ethernet. The controller of each robot then runs a program\u00a0that performs the testing automatically. This testing procedure is as follows.<\/p>\n

For a given end-effector orientation, the robot automatically aligns the 3D probe to each of the three datum spheres through several readjustments until each\u00a0indicator is at zero (\u00b10.002\u00a0mm). Then we record the end-effector’s position (actually the robot’s configuration) and send the robot 30 times to each datum sphere, first in a clockwise order\u00a0and then in counter clockwise order.\u00a0We tested the same set of five different end-effector’s orientations on each robot.<\/p>\n

 <\/p>\n