Opinion on using K-roset for simulating Zeroing joint axis?

    I've been tasked with training my electrical maintenance team in robotics. My boss and I have had internal discussions on the merit of training them on mastering/zeroing joint axis and he posed the question as to whether or not we can perform this training in K-Roset in order to avoid negative consequences to production equipment. I advised against it at first but after testing it out it appears to work. Understandably there wouldn't be any witness(zero) marks to have them jog too but I thought about creating a lab where we first zero off of the current zero value ( in leu of using witness marks) so they become familiar with the general concept/pose position. Then as a second step have them master jt1 10-20 degrees off and have them observe the differences in poses. Thoughts or suggestions?

    Not a lot to add as far as opinion goes...…..but Absolutely...…….:top:


    The majority of functions are freely usable in K-Roset, therefore anything like this, K-Roset can be used for where training on a live robot is restricted.

    However, as you've prudently pointed out, no witness marks to view in K-Roset, but you can at least go through the options for zeroing and maybe draw up a procedure for re-zeroing by using/demonstrating both ways of zeroing using either:

    1. Teach GUI

    2. Keyboard entry method using ZZERO 10x and ZZERO x commands.


    You could go a further step and modify the stl models used and incorporate witness marks in the model, but this would be a lot of effort for little return really.

    (I wouldn't consider this an option, but it could be done).


    However, nothing beats using a live robot as it introduces health and safety awareness and considerations, whereas in simulation, this does not exist.


    Just a 'point' to add.....(this is also an intentional pun).

    By using POINT command and a precision point that doesn't exist, will create a NULL Precision Point - ie a posture location set at 0 degrees for all joints.

    So you could adjust the zeroing - ie make it out of zero.

    Then create this NULL posture, drive the robot to that posture to demonstrate that the zeroing is incorrect.

    ie set an offset like you suggested...….then get them to create this null position, create a simple program with JMOVE #zero (for example) and then get them to re-zero it from there.


    You could also adopt this method in a live robot too, in order to check the zeroing.

    ie

    Get the trainees in a position of creating a program called zero_chk

    Have 2 commands:

    JMOVE #zero_chk

    HALT


    Then POINT #zero_chk in the keyboard, this will set the NULL precision point.

    Then use the program in Teach/Check to manually move the robot to where it considers it's zero to be as confirmation/verification.

    You can then re-zero to your hearts content, and use the program to check and verify.

    Oh man kwakisaki, you gave me a lot to ponder/unpack. There's a few terms I'm unfamiliar with, maybe you can explain further.

    "1. Teach GUI"

    I'm assuming this is referring to the method using the teach pendant and Aux.0501? What does GUI stand for?

    "2. Keyboard entry method using ZZERO 10x and ZZERO x commands."

    I had not thought of that! I've never used AS Commands to zero out a Robot. I'll have to look into the ZZERO commands and do some testing for myself first.

    "You could go a further step and modify the stl models used and incorporate them in the model, but this would be a lot of effort for little return really."

    What do you mean by stl models?

    "However, nothing beats using a live robot as it introduces health and safety awareness and considerations, whereas in simulation, this does not exist.


    Just a 'point' to add.....(this is also an intentional pun).

    By using POINT command and a precision point that doesn't exist, will create a NULL Precision Point - ie a posture location set at 0 degrees for all joints.

    So you could adjust the zeroing - ie make it out of zero.

    Then create this NULL posture, drive the robot to that posture to demonstrate that the zeroing is incorrect.

    ie set an offset like you suggested...….then get them to create this null position, create a simple program with JMOVE #zero (for example) and then get them to re-zero it from there.


    You could also adopt this method in a live robot too, in order to check the zeroing."


    This suggestion is pure gold. I've been in the robotics/controls field for a few years now but am new to Kawasaki's ( 6 months experience) and this is really the first time one of my core responsibilities is work force(skilled trades) development so I've been largely teaching as I learn it myself. I took a AS Language course about two months ago and that really pushed me up a notch and has made me more comfortable. I'll attach an example of a Lab I've done with my guys here recently just for anyone interested. I will also note that I am guiding my technicians through the lab so forgive me if the wording doesn't necessarily make sense, its loosely a lesson plan for myself as well and most of my technicians have taken E-Ops class already.

    'GUI' - Graphical User Interface - Yes, The Teach pendant Aux 0501 functions - But you can only specify integers using that method.

    ZZERO commands are what are detailed in the maintenance manuals, and you can use decimals aswell as integers.

    So it's better to cover them both. when training zeroing techniques.


    Any robot model in K-Roset is made up of separate STL models and is 'constructed' during project creation.

    So if you are a bit of a techie geek, and a dab hand at CAD, you could add witness marks to these in the applicable place and have them as witness marks.

    But like I said, a lot of effort for little return.


    I'll not comment on your lab (if you check my profile and website there, you'll see why as this is what I do, except I cover the product as well as applications).

    Except to say, it looks well laid out and I assume the topics you are covering are relative to your environment as opposed to 'product training'.

    Nice work...…………:top:


    Just a thought, I would also include precision points so you can highlight the differences between those and transformation values when it comes to having a different attached EOAT.

    Quote from kwakisaki

    Except to say, it looks well laid out and I assume the topics you are covering are relative to your environment as opposed to 'product training'.

    Nice work...…………:top:


    Just a thought, I would also include precision points so you can highlight the differences between those and transformation values when it comes to having a different attached EOAT.

    Thanks for the feed back, when you say precision points I'm assuming you mean what I refer to as "#Joint" poses? I'll start using the correct terminology if that is the case then. Precision points aren't any more accurate then Trans points correct? The main difference being how the pose data is recorded correct? Straight encoder values versus pose data transformed from the base cartesian?

    Yes, in the manuals you will see '#' has only one purpose and that is for a precision point or joint displacement value or what a lot of people refer to as an absolute position.


    Precision points and Transformation values have no difference in terms of accuracy but their fundamental difference is very important to understand.

    Have a look at ELBOW, ABOVE, BELOW, UWRIST and DWRIST in the AS Manual.

    This will give you an idea about 'posture' and 'transformation' differences.


    # is where the robot is moving the Joints from start angle to finish angle.

    non # is where you are transforming the current TCP relative to the current BASE from one coordinate to another.


    Or what I say to my clients 'Hash is only good for moving the joints'.

    You need to remember, when people are moving the robot around, they are looking at 'physical position'.

    The Robot however is either moving the TCP relative to the BASE coordinate or Joints.

    Will do, thank you for clarifying.

Advertising from our partners