Tool Frame Issues

    Hello All,

    I am currently setting up my tool frame using the six point(XY) method but when I teach the robot the points my TCP isn't correct. When I rotate the robot in tool mode after the TCP moves around still. I currently am going to my approach point one then recording it. Rotating J6 about 180 in joint mode then going to world mode and moving it to my reference point for approach point 2. For third approach point I am rotating my J4 and J5 in joint mode alittle under 90 degrees, then moving the robot in world to the reference point. For origin I am moving to approach one then moving the robot up in the +Z direction and recording it then moving robot in +X and +Y. This doesn't seem to be working for me though. Could someone help me understand what I am doing wrong or missing?

    What I have to do is pick up these little pushpin parts that look like two Christmas trees stuck together with the trunks stuck together (<(((=|=)))>).<(((= is the Christmas tree and there is a small circle divider between the two trees. What I have for a fixed pointer is a steel rod with one end sanded down so there is a point on it. That rod is then bolted to the side of a table. What I am using for my pointer on the robot is the gripper used to pick up the pins with a pin in it, the bottom of the pin sticks out so the point is right where the gripper closes together around the pin. When picking up the pins, the robot is on until I rotate the pin 180 degrees and its about 5mm off.

    Looks like you are working with small parts, 5 mm is a huge error.

    Beside the point that there's human error (you teaching the setup points) which I believe you are doing your best (still, that is an error), the other problem could be that the robot is not calibrated properly.

    Maybe this is not the best of the answers, but start adjusting your result coord until it works. You will see , the 5 mm are going to get smaller or bigger, then change the values accordingly. A lot of people do that when the it is hard to define on your tool what exactly is the TCP

    Retired but still helping

    Thank you for the replies, The pin is symmetrically made but unfortunately the gripper is not centered on the robot's faceplate, one end is the gripper and the other end is a camera. When setting up your tool frame what is the most effective way to teach the approach points.

    Do you really need the 6 point method? The 3 point method will give you just as accurate of x,y&z, but with w,p,&r = 0. I only do 6 point if I really need a tool orientation that isn't pointing away from the faceplate. And even then there are better methods such as VTCP or direct entry from CAD.

    Sent from my VS985 4G using Tapatalk

    Edited once, last by HawkME (August 16, 2016 at 1:10 PM).

    VTCP (Vision TCP) is an paid option where you use a fixed mounted 2D camera to take pictures of a target that is held by your end effector. The robot moves through an automatic sequence at different angles and distances, then using the vision system calculates an accurate 6 point TCP. By 6 point I mean it gives you x, y, z, w, p & r values in your tool definition.

    I understand unfortunately that would not work with me because the range of motion for the robot is a very small window. What is the best way/method to setup a tool frame.

    Quote from Fabian Munoz


    Maybe this is not the best of the answers, but start adjusting your result coord until it works. You will see , the 5 mm are going to get smaller or bigger, then change the values accordingly.

    Either write down existing values, or copy to another tool - then try adjusting values(X,Y,Z...) by a mm or two. There will probably be interaction, so start w/ one value at a time.

    The best method will depend on the geometry of your tool and what it has to do. We know you have a gripper, but how is it oriented, is the gripper orthogonal to the faceplate? Can you post a sketch or picture?

    Generally a 3 point method will work well for a gripper, but to be accurate you need a sharp pointer centered in the gripper jaws and a sharp pointer fixed on the work surface.

    Sent from my VS985 4G using Tapatalk

    Well I am trying the direct method but I am confused on how to exactly set it. I tweaked the numbers and got the gripper to go to the spot where I needed it to for pick up, but as soon as I rotate the part its off, am I missing something? :wallbash: :wallbash:

    An old chinese proverb says.

    "Parts designed with computer software are dimensionally perfect, parts created by a milling machines and assembled together by humans are not"

    Ok reading you post , seems to me that you need to keep tweaking the numbers.

    "but as soon as I rotate the part its off", then tweak again, and rotate again, and tweak again and rotate again and tweak again and.............

    This is the time when good robot programmers show their patient.

    Retired but still helping

    Edited once, last by Fabian Munoz (August 19, 2016 at 1:15 AM).

    fyofolkdown:

    TCP using the 3 point or six point method requires good calibration, I understand your problem, no matter how precise you teach the points the TCP
    wanders when changing orientation.

    Theoretically every point can be reached in 8 different configurations
    If you manipulate the CNF (configuration) of a taught point and move in joint
    You would notice that the position is not reached exactly, because the axis are not mastered at a good zero.

    There is a procedure to improve the accuracy, even new factory calibrated robots are not THAT accurate

    But all depends on the application, most application do not need THAT precision
    Even if it is not precisely calibrated you only need to add more points to your path

    First I want to start off by saying thank you to everyone that has helped me power through my problem. What I did was did the six point method but then went in and touched up the numbers that I got. Everytime I change the numbers should I be touching up my offset and then move part and do the same over until my TCP is precise? I am new to this direct entry style.

    If you touch up the TCP, you should be able to step to the same point and see the effect... albeit there are some hoops to jump through to step to the same point. There may be easier ways - I usually move the cursor, acknowledge the warning, then move the cursor back to the original point, acknowledge warning and step the robot.

Advertising from our partners