New to tracking

48 views
Skip to first unread message

nchuk...@gmail.com

unread,
Sep 24, 2014, 9:46:30 AM9/24/14
to velmex-...@googlegroups.com
I am new to Velmex bislide. I would like to build a tracking system for the Velmex. Essentially have the velmex follow an object smoothly. Can anyone help me get started on this? Matlab or Visual C++ is preferred. I know I need some kind of PiD but I'm not sure whatelse.

rob....@thermofisher.com

unread,
Sep 24, 2014, 12:58:02 PM9/24/14
to velmex-...@googlegroups.com, nchuk...@gmail.com
Well, I'm going to be the first to say it:

RTFM

I don't mean that to be overly insulting.  But really, read it.  It's a good manual.  One of the better ones I've read in the industry.  There are also really great examples you can follow for various languages.  Just dive in.

When you hit a road block, then come back with specific questions.

nkirukau...@gmail.com

unread,
Sep 26, 2014, 10:41:04 AM9/26/14
to velmex-...@googlegroups.com, nchuk...@gmail.com
Hi Rob,

I have read the manual as I already have the system working where it can follow an object of interest. The motion, however, is jerky. I would like to make the motion smooth by controlling the speed and acceleration not just the position: having the bislide move to this/that position at a constant speed.

I thought a PID over the serial port would be the way to go to control speed and acceleration of the bislide. Hence, my question earlier to see if anyone has done a PID before. Mark suggests using their analog input and modifying the accel profile. I am looking for any more ideas from anyone while I work on analog.

So, if anyone has done a PID before, what did you do?

Thanks

Nikky

nkirukau...@gmail.com

unread,
Sep 26, 2014, 10:42:19 AM9/26/14
to velmex-...@googlegroups.com, nchuk...@gmail.com, nkirukau...@gmail.com
correction: not Mark but Manny

Thanks

Mitchel Evans

unread,
Sep 26, 2014, 5:49:14 PM9/26/14
to velmex-...@googlegroups.com
Take a look at this Application Note:  http://www.bislide.com/controls/vxm/downloads/documentation/application_notes/an105.pdf
The analog jog input responds to a 0-5V input to control speed and direction.


Best Regards,

Mitch

--
You received this message because you are subscribed to the Google Groups "Velmex Controls" group.
To unsubscribe from this group and stop receiving emails from it, send an email to velmex-contro...@googlegroups.com.
To post to this group, send email to velmex-...@googlegroups.com.
Visit this group at http://groups.google.com/group/velmex-controls.
For more options, visit https://groups.google.com/d/optout.

velmexcontrols

unread,
Sep 26, 2014, 6:23:05 PM9/26/14
to velmex-...@googlegroups.com
Agreed,
Analog input is the closest thing to a "dynamic-speed-change".
From my understanding you have a 3.6:1 gearbox so I must ask, what sort of moves are you making? and at what speeds???
100mm = 1 rev = (400 x 3.6) = 14400 steps
So it should be pretty smooth compared to the "jerking" that you are describing.

Perhaps if you can attach a video showing what you describe we may have other ideas?

nkirukau...@gmail.com

unread,
Sep 27, 2014, 9:44:52 AM9/27/14
to velmex-...@googlegroups.com
Thank you all.

I will read up on the analog input. Thanks.

We are doing object tracking with a depth camera. We have the depth locations, convert it to locations the bislide understands, and do an absolute move to that location: essentially this line
"G, //no echo, online

C, //clear commands

setDM129, //set paraments

PM-4, //clear program 4

S1M5200, //set speed for motor 1
S3M5000, //set speed for motor 2
IA3M[_locZ], //move motor 3 to _locZ
IA1M[_locX]" //move motor 1 to _locX

Jerky might not be the right word for the motion but this line is essentially move-stop-move program at the speed we have specified. I can upload a video on Monday to show what I mean.

In the meantime, read up on analog.

Cliff

unread,
Sep 27, 2014, 10:36:13 AM9/27/14
to velmex-...@googlegroups.com
Somehow I am under the impression that you are running a belt-drive? (Not sure where it was mentioned in passing with Manny or read it somewhere?) but under that impression I am guessing the the equipment being used and the sort of moves being made are sort of like the following?

1.) Assumption from the code that you are using a VXM-1-1 ???
2.) Code in general could have some flaws due to speed.
G,	  	;On-Line mode with echo off, grouping a <cr> (carriage return) with responses
		;Not often I see anyone using G instead of F but just personal preference
C,   		;clear commands from current program
setDM129, 	;set VXM to run program 4 first after a reboot before running intended program
;
;
;
PM-4,    	;select and clear program 4
S1M5200, 	;set speed for motor 1 ;<--- Speed is a little high if you ask me?
S3M5000, 	;set speed for motor 3 ;<--- Speed is a little high if you ask me?
IA3M[_locZ], 	;move motor 3 to _locZ
IA1M[_locX],	;move motor 1 to _locX
;
;
;
PM0                        ;Select Pgm 0 (default)
;
;
;
From that code its hard to tell what the normal movements are, but I can guess its less than 1 rev = 1440 steps at a high rate of speed (which it will never get to that speed in such short moves) resulting in what you are seeing as being "Jerky"

rob....@thermofisher.com

unread,
Sep 29, 2014, 8:44:36 AM9/29/14
to velmex-...@googlegroups.com, nchuk...@gmail.com, nkirukau...@gmail.com
My apologies.  It seemed from your post that you hadn't already gotten your bislide up and running. 

nkirukau...@gmail.com

unread,
Sep 29, 2014, 9:26:17 PM9/29/14
to velmex-...@googlegroups.com, cl...@velmex.com
This is a VXM1-1 as Cliff assumed.

I did not think about the speed but I will look at it as I work on the analog input.

Here is the video I promised:
https://dl.dropboxusercontent.com/u/45127361/test_roboticArm.wmv

From the video, the tracking is satisfactory. The start-and-stop motion of the bislide while the object being tracked is still moving is something we would like to fix. That may be in our program (speed modifications like Cliff suggested) or someplace else. Any suggestions are helpful.

And @rob, apology accepted. It was my fault for not explaining myself adequately.

Cliff

unread,
Sep 30, 2014, 11:52:33 AM9/30/14
to velmex-...@googlegroups.com, nkirukau...@gmail.com
Send me the CO# that is marked on the side of the slide near the motor.

Also if you can send me a video of just the slides as they move

nkiruka uzuegbunam

unread,
Sep 30, 2014, 12:41:19 PM9/30/14
to Cliff, velmex-...@googlegroups.com

I can get a video tomorrow.  But the # you asked for is C043310.  What are you thinking?

Nkiruka Uzuegbunam nee Chuka~Obah

Cliff

unread,
Sep 30, 2014, 2:02:43 PM9/30/14
to velmex-...@googlegroups.com, nkirukau...@gmail.com
From the video, I believe I can see what you are doing now and the equipment involved.
It would appear that the camera is attempting to make a mirrored image profile of your movements in the background and sending those commands to the VXM as it reads them?

If I am correct in that assumption, then that will never work. The only thing to "slightly" smooth out the motion is to generate all the commands ahead of time and send the whole thing as a program instead of "one-line-chunks"
Reply all
Reply to author
Forward
0 new messages