Part testing with FT300

Sebastien 2016-11-22

Hi pros,
we were asked, during our last seminar, if we can do part/quality testing with the FT300 sensor mounted on UR. Yes we can! Below is a video of a quick proof of concept that we made. In the video, we are measuring the weight of the water bottle and checking how much it is filled with water.
During the process, we first zero the sensor at the measuring position with no bottle gripped. Then we pick up a full bottle and an empty bottle as part of a 'calibration process'. Then based on the difference between the full and empty value measured we can calculate a filled_ratio of the water bottle that we pick.

filled_ratio≔((weight-weight_empty)/(weight_full-weight_empty))*100

Note that we are calculating weight based on the moment values and lever arms. We are getting values that are repeatable at +-2% on the filled_ratio. 0% being the empty bottle measurement and 100% being the full bottle measurement.
@Annick_Mottard what key aspects should we look for to help achieve the best precision we can?

https://youtu.be/f0WqHGUrlDI

Tyler_Berryman 2016-11-24

@Sebastien Very nice proof of concept! I like how you are using the variables to set an empty bottle weight and a full bottle weight. This makes your program flexible and easy to adapt to different sizes of bottles.

Just a thought : Did you verify that the position used to calibrate the sensor and the position used to measure the bottle's weight is the same? It is possible that slight variations in the measurements could be linked to the position of the sensor's calibration when you ran the calib.tool.urp program after installing the sensor on the UR's wrist. It would be interesting to run a few calibrations with slight variations in the sensor's position to see if this has an impact on the measurement.

Sebastien 2016-11-24

@Tyler_Berryman
Good idea! I will try to give it a shot when I get a minute!

Annick_Mottard 2016-11-28

@Sebastien that is a nice application!

Are you using the force or moment feedback to measure the weight of the bottle? My only concern with using the moments in this case would be the variable distance between the sensor and the bottle. This distance depends on how the gripper picks the bottle. You need to take it into account when getting the actual weight measured using the moment - if that is the method you are using.

Other than that, using the moment feedback results in a smaller signal noise, which is a plus!

Sebastien 2016-11-28

@Annick_Mottard
yes we are using the moments with their lever arm to calculate the weight.
As this was a simple proof of concept we did not spend much time on the gripper positioning when picking the bottle. So I agree with you that this would make my measurement vary!