Discussion

Left ArrowBack to discussions page
mjwelchmjwelch Posts: 5Partner Apprentice
I am setting up a proof of concept for a customer where I am picking up 6 containers each full weighing 340g each using a vacuum gripper. I would like to be able to sense if one of the lids pops off thus leaving the full container behind using a force torques sensor. I have tried using get_sensor_fz() to store the weight then comparing it to what I have proven a full load would be. This seems a bit unreliable. Does anyone have a better recommendation for this process?

Comments

  • matthewd92matthewd92 Posts: 826Founding Pro, Tactile Sensor Beta Testers Handy
    To make it reliable, you will need to stop and stabilize before measuring the load on the arm.  We have done similar things with FT sensors in the past, but we had a bit larger weight difference, closer to a kg but it was reliable in practice.  Our toughest aspect was cycle time was super critical so we didn't get long to stabilize the load before having to measure it. Also, make sure to bias the sensor prior to lifting so you know how much force you lifted.  What sensor are you using?
  • mjwelchmjwelch Posts: 5Partner Apprentice
    I am using the sensor built into the UR5e, I haven’t considered an external sensor as of yet. What I am noticing is that after the wrist joint warms, the feedback fluctuates not matter how much setting time I allow. The containers are in a configuration of two rows of three and are approx 125mm square, with the TCP being between the two center containers. Depending on which is empty, I expect a slight rotational force but still expect a large enough difference in Z to detect. Thus far, this has not been reliable.
  • Nicholas_FrancoeurNicholas_Francoeur Posts: 93Wrist Camera URCap 1.3.0, Vacuum Beta tester Handy
    edited March 1
    Are you using the zero_ftsensor() function right before picking up the container? The built-in sensor is made to work for spontaneous measurements only. A zero sensor must be done each time you want to read it’s data.
    Nicholas Francoeur
    Co-Founder | Vanguard Robotics
    e:   [email protected]
    w:  vanguard-robotics.com
  • mjwelchmjwelch Posts: 5Partner Apprentice
    Yes, I zero the sensor before picking the containers. I have to pick twice to get all containers. After the second pick, I settle and read the load. This is where the numbers are varying. 
  • Nicholas_FrancoeurNicholas_Francoeur Posts: 93Wrist Camera URCap 1.3.0, Vacuum Beta tester Handy
    Can you give us some more details about that variation? How much?
    Nicholas Francoeur
    Co-Founder | Vanguard Robotics
    e:   [email protected]
    w:  vanguard-robotics.com
  • mjwelchmjwelch Posts: 5Partner Apprentice
    6 full - 21.324628
    5 full - 20.209324
    4 full - 20.389290
    3 full - 20.956726
    2 full - 21.423036
    1 full - 20.566676

    These results are from consecutive cycles, removing a package after each. I am not sure that I understand the unit of measure when using get_sensor_fz()

    The cycle is very simple, move above, wait 0.5 seconds, zero, pick 3 containers, move down the conveyor, pick 3 more containers, wait 0.5 seconds,  Weight:=get_sensor_fz(), popup Weight.
  • Nicholas_FrancoeurNicholas_Francoeur Posts: 93Wrist Camera URCap 1.3.0, Vacuum Beta tester Handy
    What is the weight of a container? The units are newtons.
    Nicholas Francoeur
    Co-Founder | Vanguard Robotics
    e:   [email protected]
    w:  vanguard-robotics.com
  • mjwelchmjwelch Posts: 5Partner Apprentice
    Individual containers weigh 340 grams - 3.33 newtons. Tooling weight is 690 grams - 6.7 N
Sign In or Register to comment.
Left ArrowBack to discussions page