Different approaches for converting from cartesian to joint in KAREL

    I was looking into converting cartesian positions (represented as XYZWPREXT variables) into joint positions (JOINTPOS variable) via KAREL. I found two other posts (this one and this one) which basically contain two approaches.

    Approach 1

    Let the KAREL built-in casting do the work: P_joint = P_cart

    Approach 2

    Use the built-in routine POS2JOINT(_, P_cart, _,_ ,_ ,_ , _, P_joint, _)


    I used to use approach 1 because of its simplicity, but I've now found that it doesn't always convert into the correct joint representation (when comparing to the 'golden standard' of setting a cartesian PR and changing the representation to joint). So I switched to the second approach; seems to work fine.. but it raises the question of why these approaches differ? Any ideas?

    Does the automatic casting (approach 1) not use $MNUFRAMENUM and $MNUTOOLNUM perhaps?

    Quote from hoitzing

    Approach 2

    Use the built-in routine POS2JOINT(_, P_cart, _,_ ,_ ,_ , _, P_joint, _)

    I've tried to use this a while ago in order to check the expected J6 angle of a point provided by vision equipment.

    It works ok if you will do a J movement to that position, but it will not if you will execute an L movement even if you provide the function with the desired solution it must use (config_ref = NO_TURNS for a solution calculated moving the J4/5/6 axis the minimum, like a linear movement will do)

    So if you will use this to check if a cantesian position is reachable with a max J4 or J6 angle take this into account... nothing I tried with that built-in routine provided me the correct joint angles for an L movement all the time... sometimes it would match, and sometimes it would not.

    Finally I was so sick that I've made a BG logic routine to check the J6 angle and cancel the movement if J6 angle was exceeded.

Advertising from our partners