Posts by rf103

Quote from failarmada

Although I'm not sure why it would default to FALSE...

Because it can lead to "strange" behaviour.

The controller will now read ahead (a few lines) in your program and essentially make a copy of whatever value you have in your register when it encounters the motion instructions.

This copy is the value that will be used.

If you have a program where the register values are constantly updated, making copies like that can lead to lost updates, where an old value is used instead of the new one.

Setting $RGSPD_PREXE to true must be a decision made by the robot programmer as it has consequences for all programs and ultimately the behavior of the robot.

Quote from massula

At first, I was trying to re-serialize a virtual M900iA/260L to a M900iB/280L, and the 280L didn't appeared in the selection window. This particular robot uses a custom FRVRC.

Just out of curiosity: what is custom about it?

Press shift and disp, then choose one of the options.

Quote from Nation

I've got a V420 compiler I got from an old guy early in my career. I've attached the compiled code, along with the compiler itself. I had to spin up a windows xp virtual machine to run the compiler though. Windows 7 flat out refuses to run it, regardless of compatibility settings.

Thanks for the V4.20 compiler.

dosbox is quite happy to run it as well. No need for a virtual machine.

Btw Nation: all the files, excep tsiodim's program and pc file have the hidden attribute set, which makes the files invisible on my machine. Had to remove the attribute using attrib.

Could you post the binary here if you do manage to compile it? I'd like to take a look at it.

Quote from tsiodim

Unfortunately no, it cannot be loaded (same error as previously).
Either can not be copied correctly.
The copy procedure terminates successfully, but the file is "null", size 0.
Thank you.

Then I can't help you right now unfortunately.

I only have a V2.22 that is older, but that won't compile your program.

Quote from massula

rf103, did You used WinOLPC?

I used the compiler infrastructure from an old winolpc version yes. But not winolpc itself.

Lowest version I can find right now: 5.11-1. See attachment.

Not sure it'll help, but the attached pc file was compiled with 6.40: still a long way from 4.3, but at least closer than 8.10.

As far as I know the zero distance error is not caused by specifying 0 distance offsets, but the motion that leads to DPM being activated cannot be zero distance.

So for stationary tracking for instance, the FINE motion after the Track DPM statement cannot be a zero distance motion.

Changing stop tolerances does not influence this (and sounds like something that should not be haphazardly done).

As far as I know there are no threads in karel, only programs, but disregarding that: see the manual for a description of the semaphore functions. Not exactly the same, but probably the closest you'll get with karel.

Only thing I've been able to find: https://github.com/ros-industr…evel...isys-vision:ikfast.

They seem to be doing the same thing (ikfast appears to be an ik solver generation framework).

Thanks for the suggestion. We're already using that. Going to full velocity control over all axes will make things possible we cannot do with plain position control (even with DPM).

Does anyone know whether a complete robot can be reconfigured with all axes as process axes?

I'm trying to figure out whether it'd be possible to control an entire robot in velocity mode. With an external computer doing the FK/IK, this would allow for highly dynamic motions (based on camera input for instance). Think visual servoing, dynamic obstacle avoidance, etc. Position control mode doesn't work for those cases, and process axes seem to be the only way of achieving velocity control.

Before I try and reconfigure an entire robot I thought I'd ask whether someone had already tried doing that.

Quote from HawkME

Fanuc Basic Process Axis has only speed control, no position control.

Exactly

What if I'd like to have velocity control over all axes?

Does anyone know whether it is possible to reconfigure an entire robot (so all 6 axes) such that all axes become process axes?

Not saying that you shouldn't do what you are after, but just mentioning that RoboDK appears to be able to do just this.

https://www.robot-forum.com/ro…ons/robodk-g-code-import/

Quote from Velcro

Is this correct? There seems to be a lot of people using socket messaging but does it not work on windows?

This is a strange thing to ask, as "socket messaging" is a technique, not a program that runs on either your robot or a PC.

The option on the robot is called User Socket Messaging, but that just provides you with the necessary Karel functions to open and close sockets.

And I can understand why fanuc would say that the PCDK is recommended: it takes away some of the low-level details of communicating with your controller (so you can immediately read and write registers, positions, ios and so forth. With socket messaging you're going to have to do some more work.

This will not work.

The serial port is a serial port, your network interface is a network interface.

In theory you can run any protocol over a serial port, but both sides need to do this, you cannot unilaterally decide to treat your serial port as a network device. To my knowledge, the Fanuc controller does not support this, so regardless of what you do on your laptop you cannot use the program you mentioned and expect it to work.

ERBU will also not work, or it will at least not be able to establish an FTP connection over a plain serial port.

Quote from Onequickpuck

I haven't been able to pull it off. I even brought in my IT guy, but he can't get his head around the robot NOT having an IP address LOL.

I'm rather surprised your it guy didn't know this.

I don't have an image of an older controller available right now, but another alternative would be to look at featfil.dat (in os\frs).