{"id":84007,"date":"2023-01-13T11:00:56","date_gmt":"2023-01-13T03:00:56","guid":{"rendered":"https:\/\/www.tm-robot.com\/?post_type=docs&p=84007"},"modified":"2023-03-27T11:07:49","modified_gmt":"2023-03-27T03:07:49","slug":"deviation-calibration-for-robot-duplicate","status":"publish","type":"docs","link":"https:\/\/www.tm-robot.com.cn\/de\/docs\/deviation-calibration-for-robot-duplicate\/","title":{"rendered":"Deviation Calibration for Robot Duplicate"},"content":{"rendered":"
There exists robot differences because of axial differences between each robot coordinate system, TCP posture differences and absolute accuracy errors. So if you directly run the project imported from the golden robot, the duplicate robot will get a different result.<\/p>\n
When you need to replace a new robot after the golden robot crashes, or have request of robot duplicate, such as AMR use, the traditional way is adjusting robots one by one.<\/p>\n Now just imagine, you have 100 robots and want them to do the same actions on 100 working stands, and if you spend T hours for adjusting the first robot, it still needs T*99 hours for adjusting other robots.<\/p>\n As you can see, this requires lots of time and manpower. To solve this problem, we provide this method to reduce adjusting time for duplicate.<\/p>\n To solve the axial difference problem, you can use a Landmark to build the vision bases on it, then edit points on this base.<\/p>\n After that, choose a visual point of the golden robot as the calibration point, then re-adjust this point as a new point on the duplicate robot. Calculate the point offset and compensate to the duplicate robot\u2019s TCP. Points applied this modified TCP will get the same compensation.<\/p>\n Modified TCP can only be used for points with the same posture of the calibration point.<\/p>\n By measurement, it\u2019s expected that the point error after calibration can be under 1 mm, but the points should follow the rule: points that are within 20 cm above the calibration point and \u00b110 cm in the X-Y directions with the calibration point at the center.<\/p>\n However, the error and range still depend on how you do calibration and the calibration accuracy at the calibration point. If the error is larger, the range would narrow down.<\/p>\n Here we edit a project of picking up a Front Opening Unified Pod (FOUP), and will use two robots to show how to do deviation calibration.<\/p>\n [The following can be omitted if no need]<\/strong><\/p>\n To let you directly observe the calibration result, we install two screws on the TCP, and setup an offline calibration platform. There also installs another two screws on the platform.<\/p>\n Two TM robots with the same model [The following can be omitted if no need]<\/strong><\/p>\n A platform can be setup on the working stand Note:<\/strong><\/span><\/p>\n The first robot for editing the projects is the golden robot. We will edit the project with it in the working area, then move it to the calibration stand for calibration.<\/p>\n Before editing the project, navigate to Settings<\/strong> > TCP Settings<\/strong> to set a new TCP.<\/p>\n Steps:<\/p>\n Install the screws on the TCP and move the robot to the calibration stand, then attach a landmark on the stand and redo the landmark positioning tasks. Choose a point from the project as the calibration point and then guide the robot to arrive at this point.<\/p>\n Here we choose the point of picking up the FOUP as the calibration point. Then adjust positions of screws on the calibration stand to let two sets of screws be aligned.<\/p>\n Import the project and TCP from the golden robot. First we directly run the project at the calibration stand, and find that the point deviation happens. Then we follow the steps below to do deviation calibration.<\/p>\n Q1. Why the result is not satisfactory after the deviation calibration?<\/strong><\/p>\n A1.<\/strong> The reasons are probably as follows:<\/p>\n Q2. Can I use this modified TCP for all project points?<\/strong><\/p>\n A2.<\/strong> No. See A1<\/strong>(3). Besides, even though the postures are the same, it\u2019s still limited to correction range. Remember that the farther the distance between working and calibration points is, the larger the error is.<\/p>\n Q3. If the points spread too widely and would exceed the correction range, how to solve it?<\/strong><\/p>\n A3.<\/strong> You need to divide the points into several groups, and choose a point from each group for calibration (notice the point posture). Then save the modified TCP as a new TCP, and use Tool Shift<\/strong> to change TCP of each group as the correspondingly new one.<\/p>\n
<\/a><\/p>\n
<\/a><\/p>\nIntroduction<\/h1>\n
<\/a><\/p>\n
<\/a><\/p>\nLimitations of use<\/h1>\n
Point<\/h2>\n
Accuracy & Correction Range<\/h2>\n
<\/a><\/p>\nAn Example of an AMR Application<\/h1>\n
<\/a><\/p>\n
<\/a><\/p>\nRequirements<\/h2>\n
\nA FOUP
\nA fork for picking up FOUPs
\nA landmark<\/p>\n
\nA landmark
\nTwo screws (on the platform)
\nTwo screws (on the TCP)<\/p>\n\n
Golden Robot<\/h2>\n
Set a TCP<\/h3>\n
<\/a><\/p>\nEdit project<\/h3>\n
\n
\n
<\/a><\/li>\n\n
\n
<\/a><\/li>\n
\n
<\/a>
\nIf the scores are too low or float greatly, you can re-adjust the robot observation angle or go back to flow page to re-adjust the camera parameters.
\n
<\/a><\/li>\n<\/ul>\n<\/li>\n
\n
<\/a><\/li>\n<\/ol>\nAdjust calibration platform [can be omitted]<\/h3>\n
<\/a><\/p>\nBackup or Duplicate Robots<\/h2>\n
\n
\n
<\/a>
\n
<\/a><\/li>\n
\n
<\/a><\/li>\n\n
\n
<\/a><\/li>\n
\n
<\/a><\/li>\n
\n
<\/a><\/li>\n<\/ul>\n<\/li>\n
\n
<\/a><\/li>\n
\n[Can be omitted if no need]<\/strong>
\nWe first validate the result at calibration stand, and find that the screws are almost aligned after calibration.
\n
<\/a>
\nThen we move the robot to the working area to pick up a FOUP. As you can see, if we use the original TCP to run, the robot catches it with a tilt angle. But change to modified TCP, it gets a FOUP with correct posture.
\n
<\/a><\/li>\n<\/ol>\nQ&A<\/h1>\n
\n
Demo Video<\/h1>\n