Hi,
We’re looking for a means to verify the performance of an optical motion tracking device for 3D movements. To do this we’re considering using something like the Lynx 6 to perform known 3D motions of an object that is being independently tracked. The aim is to see if the known movements are extracted from the tracking device.
My questions are:
The Lynx 6 specs suggest a 0.09 degree accuracy for joint movements. I assume this is at the joint origin - so the error at the end of the arm is more. Is this a reliable spec from your experience?
(We’re aiming for ~0.2 mm accuracy for translations and ~0.1 degree accuracy for rotations.)
The second question is related to the first: will the Lynx 6 be a good option for achieving reproducible 3D orientations of our target object?
-ensure the PWM timings from your controller are what you expect
the type of controller you choose may dictate your resolution (although doubtful since the controller will usually offer more steps/increments than the servo could possibly accommodate)
-coreless hobby servo motors have a higher resolution than the cored hobby servo motors. The coils are glued in the coreless design and not wrapped around iron as in the cored deisgn. This yields a lighter setup. result - smoother operation yielding better resolution
-use titanium gears instead of all others - over time the titanium wears the least thus the manufactured tolerances are maintained over a much longer period of time. Nylon and other materials will flex (flexing is bad for high resolution applications) and wear much faster resulting in decreasing resolution.
Im guessing you will not meet those tolerances (in particular the 0.2mm translational accuracy) using plastic servo horns and flexible lexan as the arm material and I think you will have more luck with Lynmotion’s all aluminum SES (servo erector set) arm instead.
I am not sure of the rotational accuracy for the setup but the SSC-32 at 0.09 degrees is definitely a reasonable number - but remember, it is the servo that has to be able to responds to the increments of the controller and I highly doubt any hobby servo has a 0.09 degree accuracy.
I am looking about some robot arm, but with really an hobby application in view.
Yet, if i use some robot arm, i need also some accuracy, which is about the Tracker’s ones : about 0.2 mm in translation.
But the movements of the grip are confined in a cube of 1, or maxi 2cm side (perhaps important to precise this point)
So, what is the repeatability of the Lynx robots, with such conditions of working ?
Is it possible to increase it, if not suffisant, without spending millions of bucks ?
I also read about this 0.09° of precision in angles with L6, so that’s why i come in until here.
I want to precise that my status of roboteer is really untrue.
Noob de chez noob should be more appropriate.
The “Accuracy of motion per axis = Servo controller dependant (SSC32=.09 degrees)” is only what the servo controller puts out, and not actual servo performance. In my testing an unloaded standard servo was only capable of ~.4 deg of accuracy. The same may be true of digital servos also.
I have to say that my goal is to reproduce all the movements of a master brush, and reproduce them to paint a miniature, not necessarly plane…
The robotics enter in the dance, because slave path is exactly the same of the master path, but with a reducing factor of 3, or 5 or 10…
So, if i am sure that the path of grip is controlled, at all points, with an accuracy of 0.1 or 0.2mm, my goal could be reached.
The path is confined in a cube of 1 or 2cm, so perhaps it’s possible to reach such a precision, i don’t know.
I am totally newbie, but perhaps the servo performance is not the only parameter.
I say this, because i saw this video :
easyrobotics . fr/videos/Bras%206%20axes.wmv
and obviously, the structure of the robot has some importance, in the accuracy of the path.
wow that thing (the arm in the video) wobbles all over the place… needs to cut down on the coffe intake perhaps.
while I suspect an Lynxmotion SES arm would certainly be more mechanically robust, and being familliar with the art of painting miniatures myself, I think you would need yet an even more solid structure and higher precision than you will get with hobby servos to accomplish an electo-mechanical pantograph capable of replicating the process of painting miniatures.
I agree with Eddie, my thoughts exacly. The precission required would call for custom machine work at high tolerances and robust software code to run it.
I am sorry but I don’t have a canned solution off the top of my head for this level of application. The kind of resolution you are talking about holding at the end effectors of a 4 or 5 axis control is better than a few thousandths of an inch (0.2mm = 0.00787in, 0.1mm = 0.00394in) in 3-space. This isn’t a trivial design mechanically or electronically. Even if you are only moving something weighing a few ounces, every joint needs to be tight and have a high resolution of position measurement to achieve that degree of precision. I would consider the type of controls required to implement this to be industrial products designed for high performance closed loop servo control. In short I doubt you are talking about only several hundred dollars to implement this design.
How about a test? The idea is repeatability not absolute positioning. Drawing with a pencil on paper? With a felt marker? What kind of test would prove/disprove the accuracy can be achieved?
oz
