ADVSERVO.LIB and Adept Robots

This will be an archive of the Yahoo Adept Software Group messages

ADVSERVO.LIB and Adept Robots

Postby uoftrugby@yahoo.ca » Thu Aug 21, 2008 5:00 pm

Hello,

I have a project which requires me to go beyond V+'s general
programmability. Basically, I will have to control the robot's motors
directly (ideally with voltage) for some rather "different" tasks. That
is to say, simply moving joint ii 10 degrees, and joint ij 20 degrees
isn't going to cut it.

I've gone through the ADVSERVO documentation and noticed that its
somewhat vague on its implementation and introduction, however it did
state that ADVSERVO.LIB is limited to Joint 4 on Adept Robots. Does
this mean that ADVSERVO can NEVER be used on joints 1-3 ? How can
I "unsecure" those joints so that I don't get an "Illegal Joint Number"
error when executing the program. Is this done through SPEC?

How can I go around the security features on Adept's system to access
joints 1-3 so that I can perform ASYNC tasks with V+ ADVSERVO? I'm
afraid direct control/monitoring of the motors is the only use for this
work...

Thanks
uoftrugby@yahoo.ca
 
Posts: 3
Joined: Thu Aug 21, 2008 5:00 pm

Re: [adept_software] ADVSERVO.LIB and Adept Robots

Postby juergen.bosse@robo-technology.de » Thu Aug 21, 2008 5:00 pm

Best regards,
Juergen
uoftrugby schrieb:
>
> Hello,
>
> I have a project which requires me to go beyond V+'s general
> programmability. Basically, I will have to control the robot's motors
> directly (ideally with voltage) for some rather "different" tasks. That
> is to say, simply moving joint ii 10 degrees, and joint ij 20 degrees
> isn't going to cut it.
>
> I've gone through the ADVSERVO documentation and noticed that its
> somewhat vague on its implementation and introduction, however it did
> state that ADVSERVO.LIB is limited to Joint 4 on Adept Robots. Does
> this mean that ADVSERVO can NEVER be used on joints 1-3 ? How can
> I "unsecure" those joints so that I don't get an "Illegal Joint Number"
> error when executing the program. Is this done through SPEC?
>
> How can I go around the security features on Adept's system to access
> joints 1-3 so that I can perform ASYNC tasks with V+ ADVSERVO? I'm
> afraid direct control/monitoring of the motors is the only use for this
> work...
>
> Thanks
>
>=20=20
juergen.bosse@robo-technology.de
 
Posts: 132
Joined: Tue Jul 09, 2002 5:00 pm

Re: ADVSERVO.LIB and Adept Robots

Postby uoftrugby@yahoo.ca » Thu Aug 21, 2008 5:00 pm

Hi Juergen,

Yes I am working with a Cobra s350 so the gravity issue shouldn't be
a problem. Also, a cage will be set up around the robot in the near
future (the robot is still new!).

Basically what we need to do is feed a voltage (Torque is equivalent
i'm assuming) to joint 1 and joint 2 in "real-time" to get the robot
from a percieved position, to a target position. This requires
monitoring the position of the robot at every nth time interval, and
correcting its motion using a predetermined algorithm (based on error
vectors and a model of the plant). Unfortunately, we can't use
Adept's predetermined trajectory planning/motion planning algorithms
because this is towards developing a unique control system.

In order to do this, we will also need to feed user generated
voltages (possibly read from a file) into the robot in order to
determine how it behaves, and develop a model of the plant.

So our two main goals are to monitor the robot position in real time
(degrees), and feed voltages into the motors in order to correct any
errors associated with its position. Both vision, and the joint
positions will be used for monitoring.

Thanks.
Filip

--- In adept_software@yahoogroups.com, Juergen Bosse
wrote:
>
> Hi there,
>
> from your email I am guessing you are working with a SCARA robot,
like
> an AdeptOne or Cobra robot - is this correct?
>
> For if it is, I'd have less qualms about you messing with the robot
> joints directly than with a PUMA-type robot, like the Viper series.
>
> Basically, in both cases, you can peel off enough control layers so
that
> you can command the torque of the motors directly. For a SCARA
robot,
> this is mostly fine, but for a PUMA robot, you then have to worry
about
> gravity compensation from the first millisecond you put joints 2
and 3
> into current mode.
>
> This (and the warranty issue) is the reason Adept is usually not
> allowing access to the motors directly. There are serious operator
and
> equipment safety issues involved here, so unless you operate the
robot
> inside an enclosed area with no physical access for any human
beings,
> and unless you completely forget about the warranty of your system,
we
> really shouldn't let you mess with it.
>
> It can be done, though. And I'd be the one to do it. In fact, I
have
> done it (of course I have) :-)
>
> How about you outline your task a bit for us, and we take it from
there...?
>
>
> Best regards,
> Juergen
>
> PS: If this is a newer robot system, with sDAI, sEJI or AIB, you
may
> want to look at Cerebellum's CIDE to do low-level work on the
servos.
> It's an expensive tool, but it's worth every cent of its price.
>
>
> uoftrugby schrieb:
> >
> > Hello,
> >
> > I have a project which requires me to go beyond V+'s general
> > programmability. Basically, I will have to control the robot's
motors
> > directly (ideally with voltage) for some rather "different"
tasks. That
> > is to say, simply moving joint ii 10 degrees, and joint ij 20
degrees
> > isn't going to cut it.
> >
> > I've gone through the ADVSERVO documentation and noticed that its
> > somewhat vague on its implementation and introduction, however it
did
> > state that ADVSERVO.LIB is limited to Joint 4 on Adept Robots.
Does
> > this mean that ADVSERVO can NEVER be used on joints 1-3 ? How can
> > I "unsecure" those joints so that I don't get an "Illegal Joint
Number"
> > error when executing the program. Is this done through SPEC?
> >
> > How can I go around the security features on Adept's system to
access
> > joints 1-3 so that I can perform ASYNC tasks with V+ ADVSERVO? I'm
> > afraid direct control/monitoring of the motors is the only use
for this
> > work...
> >
> > Thanks
> >
> >
>
> --
>
_____________________________________________________________________
>
> Juergen Bosse
> Robo-Technology GmbH Sitz der Gesellschaft: Puchheim
> Benzstrasse 12 Amtsgericht München HRB 140072
> 82178 Puchheim Geschäftsf.: Jürgen Bosse, Jörn
Bosse
> Germany
>
> Tel: +49-89-8006390 http://www.robo-technology.de
> Fax: +49-89-807917 juergen.bosse@...
> Mob: +49-171-5178540
>
_____________________________________________________________________
>
uoftrugby@yahoo.ca
 
Posts: 3
Joined: Thu Aug 21, 2008 5:00 pm

Re: [adept_software] Re: ADVSERVO.LIB and Adept Robots

Postby juergen.bosse@robo-technology.de » Thu Aug 21, 2008 5:00 pm

uoftrugby schrieb:
>
> Hi Juergen,
>
> Yes I am working with a Cobra s350 so the gravity issue shouldn't be
> a problem. Also, a cage will be set up around the robot in the near
> future (the robot is still new!).
>
> Basically what we need to do is feed a voltage (Torque is equivalent
> i'm assuming) to joint 1 and joint 2 in "real-time" to get the robot
> from a percieved position, to a target position. This requires
> monitoring the position of the robot at every nth time interval, and
> correcting its motion using a predetermined algorithm (based on error
> vectors and a model of the plant). Unfortunately, we can't use
> Adept's predetermined trajectory planning/motion planning algorithms
> because this is towards developing a unique control system.
>
> In order to do this, we will also need to feed user generated
> voltages (possibly read from a file) into the robot in order to
> determine how it behaves, and develop a model of the plant.
>
> So our two main goals are to monitor the robot position in real time
> (degrees), and feed voltages into the motors in order to correct any
> errors associated with its position. Both vision, and the joint
> positions will be used for monitoring.
>
> Thanks.
> Filip
>
> --- In adept_software@yahoogroups.com=20
> , Juergen Bosse
> wrote:
> >
> > Hi there,
> >
> > from your email I am guessing you are working with a SCARA robot,
> like
> > an AdeptOne or Cobra robot - is this correct?
> >
> > For if it is, I'd have less qualms about you messing with the robot
> > joints directly than with a PUMA-type robot, like the Viper series.
> >
> > Basically, in both cases, you can peel off enough control layers so
> that
> > you can command the torque of the motors directly. For a SCARA
> robot,
> > this is mostly fine, but for a PUMA robot, you then have to worry
> about
> > gravity compensation from the first millisecond you put joints 2
> and 3
> > into current mode.
> >
> > This (and the warranty issue) is the reason Adept is usually not
> > allowing access to the motors directly. There are serious operator
> and
> > equipment safety issues involved here, so unless you operate the
> robot
> > inside an enclosed area with no physical access for any human
> beings,
> > and unless you completely forget about the warranty of your system,
> we
> > really shouldn't let you mess with it.
> >
> > It can be done, though. And I'd be the one to do it. In fact, I
> have
> > done it (of course I have) :-)
> >
> > How about you outline your task a bit for us, and we take it from
> there...?
> >
> >
> > Best regards,
> > Juergen
> >
> > PS: If this is a newer robot system, with sDAI, sEJI or AIB, you
> may
> > want to look at Cerebellum's CIDE to do low-level work on the
> servos.
> > It's an expensive tool, but it's worth every cent of its price.
> >
> >
> > uoftrugby schrieb:
> > >
> > > Hello,
> > >
> > > I have a project which requires me to go beyond V+'s general
> > > programmability. Basically, I will have to control the robot's
> motors
> > > directly (ideally with voltage) for some rather "different"
> tasks. That
> > > is to say, simply moving joint ii 10 degrees, and joint ij 20
> degrees
> > > isn't going to cut it.
> > >
> > > I've gone through the ADVSERVO documentation and noticed that its
> > > somewhat vague on its implementation and introduction, however it
> did
> > > state that ADVSERVO.LIB is limited to Joint 4 on Adept Robots.
> Does
> > > this mean that ADVSERVO can NEVER be used on joints 1-3 ? How can
> > > I "unsecure" those joints so that I don't get an "Illegal Joint
> Number"
> > > error when executing the program. Is this done through SPEC?
> > >
> > > How can I go around the security features on Adept's system to
> access
> > > joints 1-3 so that I can perform ASYNC tasks with V+ ADVSERVO? I'm
> > > afraid direct control/monitoring of the motors is the only use
> for this
> > > work...
> > >
> > > Thanks
> > >
> > >
> >
> > --
> >
> __________________________________________________________
> >
> > Juergen Bosse
> > Robo-Technology GmbH Sitz der Gesellschaft: Puchheim
> > Benzstrasse 12 Amtsgericht M=FCnchen HRB 140072
> > 82178 Puchheim Gesch=E4ftsf.: J=FCrgen Bosse, J=F6rn
> Bosse
> > Germany
> >
> > Tel: +49-89-8006390 http://www.robo-technology.de=20
>
> > Fax: +49-89-807917 juergen.bosse@...
> > Mob: +49-171-5178540
> >
> __________________________________________________________
> >
>
>=20=20
juergen.bosse@robo-technology.de
 
Posts: 132
Joined: Tue Jul 09, 2002 5:00 pm

RE: [adept_software] Re: ADVSERVO.LIB and Adept Robots

Postby alexandre.haag@cerebellum-automation.com » Thu Aug 21, 2008 5:00 pm

uoftrugby schrieb:=20
alexandre.haag@cerebellum-automation.com
 
Posts: 17
Joined: Sun May 08, 2005 5:00 pm

RE: [adept_software] Re: ADVSERVO.LIB and Adept Robots

Postby alexandre.haag@cerebellum-automation.com » Thu Aug 21, 2008 5:00 pm

uoftrugby schrieb:=20
alexandre.haag@cerebellum-automation.com
 
Posts: 17
Joined: Sun May 08, 2005 5:00 pm

Re: ADVSERVO.LIB and Adept Robots

Postby uoftrugby@yahoo.ca » Tue Sep 23, 2008 5:00 pm

Hi,

eveything is working well at the moment. Just need a few more pieces
of information. Using the advanced servo library (which i'm assuming
bipasses some routines in V+), what is the rate of information being
sent to the servo? Is it still the standard 16ms trajectory rate, or
does it change? Where can I find/set this frequency? How can the rate
be changed (SPEC/CONFIG_C)?

For example, if I have a loop which only sends DAC inputs to the
servos:

FOR 1 TO 100
CALL sl.setdac(1, TRUE, dac, error)
END

at what rate is the input being sent to the servos? At the moment the
rate is not required to be anything particular, however in the future
I will require it to run at approx 30Hz.

I am also recording the resultant position to a .dat file immediatly
after the DAC is sent to the motor.

Thanks,
Filip

--- In adept_software@yahoogroups.com, Juergen Bosse
wrote:
>
> Hi Filip,
>
> ok, sounds like some nice fundamental research!
>
> Well ok, a Cobra s350 probably won't hurt as bad as, say, a Viper
s1700
> if you ever lose control. So as long as you're observing proper
safety
> procedures, you should be ok.
>
> You should have a close look at the ex.servo.data example in the
> Advanced Servo Library reference guide. You can use it to record
all
> kinds of servo variables, like motor position and velocity, DAC
command
> (proportional to torque), etc. For a safe start you, you can have
it
> perform square waves under Adept servo control. Basically, you
could get
> all the information from that, because it excites all frequencies
at
> once. But there is also a Bode plot generator in SPEC, which is
very
> useful to learn more about the plant.
>
> If you want to program your own control loop, you will need to
account
> for the fact that V+ is being interpreted from semi-compiled code,
and
> you probably won't be able to realize a 1 kHz loop once your
filters
> become more advanced. Then again, if you know your control
frequency,
> you can still do a lot of useful things. Many years ago, we ran
servo
> loops with 5ms or even 28ms (the original Unimation Puma robots for
> example!) and it worked very well.
>
> If you need tighter loops, you will need to work with CIDE. I can
> definately recommend it for this job.
>
> But for now you can start in V+, at no extra cost and you will
already
> achieve a lot of things before you run into a performance
limitation.
>
> Best regards,
> Juergen
>
>
> uoftrugby schrieb:
> >
> > Hi Juergen,
> >
> > Yes I am working with a Cobra s350 so the gravity issue shouldn't
be
> > a problem. Also, a cage will be set up around the robot in the
near
> > future (the robot is still new!).
> >
> > Basically what we need to do is feed a voltage (Torque is
equivalent
> > i'm assuming) to joint 1 and joint 2 in "real-time" to get the
robot
> > from a percieved position, to a target position. This requires
> > monitoring the position of the robot at every nth time interval,
and
> > correcting its motion using a predetermined algorithm (based on
error
> > vectors and a model of the plant). Unfortunately, we can't use
> > Adept's predetermined trajectory planning/motion planning
algorithms
> > because this is towards developing a unique control system.
> >
> > In order to do this, we will also need to feed user generated
> > voltages (possibly read from a file) into the robot in order to
> > determine how it behaves, and develop a model of the plant.
> >
> > So our two main goals are to monitor the robot position in real
time
> > (degrees), and feed voltages into the motors in order to correct
any
> > errors associated with its position. Both vision, and the joint
> > positions will be used for monitoring.
> >
> > Thanks.
> > Filip
> >
> > --- In adept_software@yahoogroups.com
> > , Juergen Bosse
> > wrote:
> > >
> > > Hi there,
> > >
> > > from your email I am guessing you are working with a SCARA
robot,
> > like
> > > an AdeptOne or Cobra robot - is this correct?
> > >
> > > For if it is, I'd have less qualms about you messing with the
robot
> > > joints directly than with a PUMA-type robot, like the Viper
series.
> > >
> > > Basically, in both cases, you can peel off enough control
layers so
> > that
> > > you can command the torque of the motors directly. For a SCARA
> > robot,
> > > this is mostly fine, but for a PUMA robot, you then have to
worry
> > about
> > > gravity compensation from the first millisecond you put joints 2
> > and 3
> > > into current mode.
> > >
> > > This (and the warranty issue) is the reason Adept is usually not
> > > allowing access to the motors directly. There are serious
operator
> > and
> > > equipment safety issues involved here, so unless you operate the
> > robot
> > > inside an enclosed area with no physical access for any human
> > beings,
> > > and unless you completely forget about the warranty of your
system,
> > we
> > > really shouldn't let you mess with it.
> > >
> > > It can be done, though. And I'd be the one to do it. In fact, I
> > have
> > > done it (of course I have) :-)
> > >
> > > How about you outline your task a bit for us, and we take it
from
> > there...?
> > >
> > >
> > > Best regards,
> > > Juergen
> > >
> > > PS: If this is a newer robot system, with sDAI, sEJI or AIB, you
> > may
> > > want to look at Cerebellum's CIDE to do low-level work on the
> > servos.
> > > It's an expensive tool, but it's worth every cent of its price.
> > >
> > >
> > > uoftrugby schrieb:
> > > >
> > > > Hello,
> > > >
> > > > I have a project which requires me to go beyond V+'s general
> > > > programmability. Basically, I will have to control the robot's
> > motors
> > > > directly (ideally with voltage) for some rather "different"
> > tasks. That
> > > > is to say, simply moving joint ii 10 degrees, and joint ij 20
> > degrees
> > > > isn't going to cut it.
> > > >
> > > > I've gone through the ADVSERVO documentation and noticed that
its
> > > > somewhat vague on its implementation and introduction,
however it
> > did
> > > > state that ADVSERVO.LIB is limited to Joint 4 on Adept Robots.
> > Does
> > > > this mean that ADVSERVO can NEVER be used on joints 1-3 ? How
can
> > > > I "unsecure" those joints so that I don't get an "Illegal
Joint
> > Number"
> > > > error when executing the program. Is this done through SPEC?
> > > >
> > > > How can I go around the security features on Adept's system to
> > access
> > > > joints 1-3 so that I can perform ASYNC tasks with V+
ADVSERVO? I'm
> > > > afraid direct control/monitoring of the motors is the only use
> > for this
> > > > work...
> > > >
> > > > Thanks
> > > >
> > > >
> > >
> > > --
> > >
> > __________________________________________________________
> > >
> > > Juergen Bosse
> > > Robo-Technology GmbH Sitz der Gesellschaft: Puchheim
> > > Benzstrasse 12 Amtsgericht München HRB 140072
> > > 82178 Puchheim Geschäftsf.: Jürgen Bosse, Jörn
> > Bosse
> > > Germany
> > >
> > > Tel: +49-89-8006390 http://www.robo-technology.de
> >
> > > Fax: +49-89-807917 juergen.bosse@
> > > Mob: +49-171-5178540
> > >
> > __________________________________________________________
> > >
> >
> >
>
> --
>
_____________________________________________________________________
>
> Juergen Bosse
> Robo-Technology GmbH Sitz der Gesellschaft: Puchheim
> Benzstrasse 12 Amtsgericht München HRB 140072
> 82178 Puchheim Geschäftsf.: Jürgen Bosse, Jörn
Bosse
> Germany
>
> Tel: +49-89-8006390 http://www.robo-technology.de
> Fax: +49-89-807917 juergen.bosse@...
> Mob: +49-171-5178540
>
_____________________________________________________________________
>
uoftrugby@yahoo.ca
 
Posts: 3
Joined: Thu Aug 21, 2008 5:00 pm

Re: [adept_software] Re: ADVSERVO.LIB and Adept Robots

Postby juergen.bosse@robo-technology.de » Tue Sep 23, 2008 5:00 pm

Juergen Bosse
juergen.bosse@robo-technology.de
 
Posts: 132
Joined: Tue Jul 09, 2002 5:00 pm

AW: [adept_software] Re: ADVSERVO.LIB and Adept Robots

Postby Crush4size » Tue Sep 23, 2008 5:00 pm

Juergen Bosse
Crush4size
 
Posts: 12
Joined: Wed Jul 21, 2010 12:38 pm


Return to Yahoo Software Group Archive

Who is online

Users browsing this forum: No registered users and 2 guests