1
$\begingroup$

I'm facing the problem of model calibration for a large-scale manipulator (e.g. example image) with a half-spherical workspace of around 40m diameter.

The typical approach would be computer vision, which is difficult in this case due to:

  1. large scale and therefore huge distance to cameras,
  2. only possible outdoors, where the sunlight might be a problem.
  3. calibration of 5-6 cameras takes some time, the measurement should be finished in one day.

So my ideas are the following:

  1. use triangulation of radio frequency signals (e.g. RF position sensor
  2. use extra IMUs
  3. develop a sensor fusion scheme, like kalman filter

I want to ask you for alternative approaches, expected accuracy of RF position sensors without sensor fusion, problems with RF reflections (the manipulator is constructed from steel) or any other input.

thanks in advance!

Improve this question
$\endgroup$
2
  • $\begingroup$ What kind of calibration? $\endgroup$
    – Ben ♦
    Jul 5, 2019 at 16:35
  • $\begingroup$ calibration in a way that my mathematical model fits the reality as accurate as possible. I have a complex model, having the actuator torques/forces as inputs and the joint angles/joint positions as outputs. By measuring these quantities on the real machine it is possible to improve the accuracy of the mathematical model. My question aimed for the problem of measuring the position of certain points on the manipulator. $\endgroup$
    – madn
    Jul 8, 2019 at 7:38

1 Answer 1

Reset to default
1
$\begingroup$

You are not the first person in the world that wants to calibrate a larger robot, so there are commercially available solutions :)

The tool of choice in your case is a laser tracker. You attach one or multiple targets (laser reflectors) to your machine. The laser tracker consists of a laser that measures the distance to a target, two motors to pan and tilt the laser so that it finds and hits the targets.

Improve this answer
$\endgroup$
8
  • $\begingroup$ Thanks for the answer, indeed I also considered laser tracking but I'm facing some problems there as well: 1) of course line-of-sight is mandatory for the system to work. Since it is best to calibrate by exploiting the whole workspace, without using a lot of trackers I fear I can't ensure proper line-of-sight 2) laser trackers are somehow super expensive about your first comment. Could you find any literature on that topic, I couldn't. Also, I wasn't able to find any dealers selling laser trackers. $\endgroup$
    – madn
    Jun 6, 2019 at 7:19
  • $\begingroup$ You don't have to buy a laser tracker, just rent it for a day (e.g. industrialmeasurementservices.com/Services/…). Line of Sight could be solved by moving the tracker around. You could use some fixed targets to later merge the different set of measurements in one coordinate system. $\endgroup$
    – FooTheBar
    Jun 6, 2019 at 7:54
  • $\begingroup$ I contacted some companies renting laser trackers. One additional issue, I didn't have in mind before is that with a laser tracker you can only monitor one spot at the time. So I would have to do multiple measurements and replacing the reflector every time. Then I have strange data... $\endgroup$
    – madn
    Jun 11, 2019 at 8:44
  • $\begingroup$ Did they tell you that only a single marker is possible? Why isn't it possible to mount several markers and get their position (even though you'd have to search them again every time as you can't track them) $\endgroup$
    – FooTheBar
    Jun 11, 2019 at 9:06
  • 1
    $\begingroup$ I've used this system for large volume metrology. nikonmetrology.com/en-gb/product/igps It is very accurate over very large volumes. And it works slightly different than a laser tracking system as referenced above. (It has different pros/cons). $\endgroup$
    – Ben ♦
    Aug 5, 2019 at 19:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.