[EGM] External Guided Motion

Developer Tools
1

Dear ABB Support,

I am currently trying to implement the new EGM control in a .Net desktop application.

  • I compiled the .proto
  • I compiled the .cs into a dll
  • I referenced this dll and imported the abb.egm namespace in a new VB.net project
  • I installed Google Protocol Buffers via NuGet in my project
  • I translated the demo file (in C#) to VB.Net, and activated the “remove integer overflow check” in order to have an equivalent cast of DateTime.Now to proper 32B UInt without having an overflow (as used in the demo message header).

Everything seems to work fine: I receive and send messages, and my virtual robot in RS 6.0 x64 is not triggering any timeout, so it’s receiving my messages too. The timestamps & sequance numbers are matching the ones of the demo as well. But nothing is moving (while in the demo a movement is triggered with these same values), and I get no error at all. Is there any way to activate some kind of error handling for EGM? Because I have no way to find the problem…

Also, this could come from the protobuf library itself (protobuf-csharp-port, as used in the example files, seems to be C# only) → Do you have an exemple using protobuf net, which seems to be compatible with VB? .proto files are not used by this second library.

Thank you in advance,
Best regards,
Thibaut

2

I’ve got EGM “working”, but the movement is VERY slow. I based my code on the C example provided with RobotStudio.
I had to change the rapid code a bit, the example code does not provide movement all the time, put the Act and Run commands in an infinite while loop

3

I got it working in C# as well but my question concerning .Net protobuf support in general is still open. The RAPID exemples are indeed triggering extremely slow convergence, and are inexact, but it’s not so difficult to write your RAPID procedure to get much better results.
It seems that a lot of things are still missing though (no feedback messages, buggy speedref messages).

4

Hi,

I try to do a rapid programm only with EGM RAPID code following the examples writed on documentation… but I cannot reach to understand what is the relationship with frames amb movements

when I am doing changes in Group Inputs. The mechanical unit is moving but it is doing to wrong directions :-((.

The question is, the robot must follow the path and change it base on the value of group input in this case?

Here is the code:

MODULE maimMod

TASK PERS tooldata tFroniusCMT:=[TRUE,[[12.3313,-0.108707,416.142],[0.903899,-0.00320735,0.427666,0.00765917]],[2.6,[-111.1,24.6,386.6],[1,0,0,0],0,0,0.072]];

TASK PERS wobjdata wobj_EGM1:=[FALSE,TRUE,“”,[[150,1320,1140],[1,0,0,0]],[[0,0,0],[1,0,0,0]]];

TASK PERS wobjdata wobj_EGM2:=[FALSE,TRUE,“”,[[0,1000,1000],[1,0,0,0]],[[0,0,0],[1,0,0,0]]];

CONST jointtarget jpos10:=[[0,0,0,0,35,0],[9E9,9E9,9E9,9E9,9E9,9E9]];

CONST robtarget p30:=[[1500,0,1500],[0.322151,-0.601023,0.672381,0.287914],[0,0,-1,0],[9E9,9E9,9E9,9E9,9E9,9E9]];

CONST robtarget p20:=[[150,1320,1140],[0.000494947,0.662278,-0.749217,-0.00783173],[0,0,-1,0],[9E9,9E9,9E9,9E9,9E9,9E9]];

VAR egmident egmID1;

VAR egmident egmID2;

CONST egm_minmax egm_minmax_lin1:=[-100,100];

CONST egm_minmax egm_minmax_rot1:=[-2,2];

CONST egm_minmax egm_minmax_joint1:=[-0.1,0.1];

! CONST jointtarget jpos10:=[[0,0,0,0,35,0],[9E+09,9E+09,9E+09,9E+09,9E+09,9E+09]];

CONST pose posecor:=[[1200,400,900],[1,0,0,0]];

CONST pose posesens:=[[12.3313,-0.108707,416.142],[0.903899,-0.00320735,0.427666,0.00765917]];

! corr-frame: world, sens-frame: world

VAR pose posecor0:=[[1500,0,1500],[1,0,0,0]];

VAR pose posesen0:=[[1500,0,1500],[1,0,0,0]];

!TASK PERS tooldata tFroniusCMT:=[TRUE,[[12.3313,-0.108707,416.142],[0.903899,-0.00320735,0.427666,0.00765917]],[2.6,[-111.1,24.6,386.6],[1,0,0,0],0,0,0.072]];

TASK PERS loaddata load1:=[5,[0,1,0],[1,0,0,0],0,0,0];

! corr-frame: wobj, sens-frame: wobj

! TASK PERS wobjdata wobj_EGM1:=[FALSE,TRUE,“”,[[150,1320,1140],[1,0,0,0]],[[0,0,0],[1,0,0,0]]];

VAR pose posecor1:=[[0,0,0],[1,0,0,0]];

VAR pose posesen1:=[[0,0,0],[1,0,0,0]];

! TASK PERS wobjdata wobj_EGM2:=[FALSE,TRUE,“”,[[0,1000,1000],[1,0,0,0]],[[0,0,0],[1,0,0,0]]];

VAR pose posecor2:=[[150,320,0],[1,0,0,0]];

VAR pose posesen2:=[[150,320,0],[1,0,0,0]];

PROC main()

MoveAbsJ jpos10\NoEOffs,v1000,fine,tFroniusCMT;

testAO;

ENDPROC

PROC testAO()

! Get two different EGM identities. They will be used for two different eGM setups.

EGMGetId egmID1;

! EGMGetId egmID2;

! Set up the EGM data source: Analog output signals and configuration “default”

! One guidance using Pose mode and one using Joint mode

EGMSetupGI ROB_1,egmID1,“default”\Pose\giR1x:=giX\giR2y:=giY\giR3z:=giZ;

!EGMSetupAO ROB_1,egmID2,“default”\Joint\aoR1x:=ao_MoveX\aoR2y:=ao_MoveY\aoR3z:=ao_MoveZ\aoR4rx:=ao_RotX\aoR5ry:=ao_RotY\aoR6rz:=ao_RotZ;

! Move to the starting point - fine point is needed.

MoveJ p30,v1000,fine,tool0;

! Set the signals

! Correction frame is the World coordinate system and the sensor measurements are also relative to the world frame

! No offset is defined (posecor0 and posesen0)

EGMActPose egmID1\Tool:=tFroniusCMT\WObj:=wobj0,posecor0,EGM_FRAME_WORLD,posesen0,EGM_FRAME_WORLD\x:=egm_minmax_lin1\y:=egm_minmax_lin1\z:=egm_minmax_lin1\rx:=egm_minmax_rot1\ry:=egm_minmax_rot1\rz:=egm_minmax_rot1\LpFilter:=20\SampleRate:=16\MaxPosDeviation:=1000;

! Run: keep the end position without returning to the start position

EGMRunPose egmID1,EGM_STOP_HOLD\x\y\z\RampInTime:=0.05\PosCorrGain:=16;

! Move to the starting point - fine point is needed.

MoveJ p20,v1000,fine,tFroniusCMT;

! Set the signals

! Run with the same frame definitions: ramp down to the start position after having reached

! the EGM end position

EGMRunPose egmID1,EGM_STOP_RAMP_DOWN\x\y\z\RampInTime:=0.05\PosCorrGain:=16;

! Move to the starting point - fine point is needed.

MoveJ p30,v1000,fine,tool0;

! Set the signals

EGMReset egmID1;

ENDPROC

ENDMODULE

Thank you

BR

Jordi

5

Hi, I got EGM working based on the sample code provided by RobotStudio. As a first test it’s getting a constant signal from an UdPUC device. I’ve tried EGM in joint and in pose mode and it always gets to the correct position. However, the movements are extremely slow no matter in which direction or to which destination. Changing the position correction gain or the default proportional gain seems to have no influence at all. It feels like there is some kind of speed limitation because it always moves at the same speed. Does anybody know how to get the EGM movements considerably faster?

6

Did anyone manage to solve the problem of slow EGM motions?

7

To speed things you can use “\MaxSpeedDeviation”, see example below.
But if this is set to high (bigger then 10) I get a: “Dynamic load too high” error. Is there someone who know how I can fix this without reducing the speed?

EGMActPose egmID1\Tool:=UISpenholder \WObj:=wobj0, posecorTable,EGM_FRAME_WOBJ, posesenTable, EGM_FRAME_WOBJ
\x:=egm_minmax_lin1 \y:=egm_minmax_lin1 \z:=egm_minmax_lin1
\rx:=egm_minmax_rot1 \ry:=egm_minmax_rot1 \rz:=egm_minmax_rot1\LpFilter:=20\Samplerate:=4\MaxSpeedDeviation:= 10;

8

To speed things you can use “\MaxSpeedDeviation”, see example below.
But if this is set to high (bigger then 10) I get a: “Dynamic load too high” error. Is there someone who know how I can fix this without reducing the speed?

EGMActPose egmID1\Tool:=UISpenholder \WObj:=wobj0, posecorTable,EGM_FRAME_WOBJ, posesenTable, EGM_FRAME_WOBJ
\x:=egm_minmax_lin1 \y:=egm_minmax_lin1 \z:=egm_minmax_lin1
\rx:=egm_minmax_rot1 \ry:=egm_minmax_rot1 \rz:=egm_minmax_rot1\LpFilter:=20\Samplerate:=4\MaxSpeedDeviation:= 10;

9

Was able to avoid the “Dynamic load too high” error by reducing the bandwidth of the low- pass filter.

10

What values for low pass filter and gain are you using, if I may ask?

11

My value for the lowpass -filter is 5. Haven’t set a value for gain, so I guess it is set to one by default

12

Where can I find more info about EGM? Like a manual or a code example? Google doesn’t reveal it that easily..;/

13

The module is quite new so there is not a lot of information out on the web. I’m writing a master’s thesis about EGM and my primary source of information has been the "

Application manual - Controller software IRC5", document ID: 3HAC050798-001.

All the instructions and functions you need in RAPID are listed in “Technical reference manual RAPID Instructions, Functions and Data types” .

Last semester I did a project on EGM where I created two programs which you are welcome to use as an example, they are both available on Github. See the links at the bottom.

Both programs where implemented on an Irb 140 robotic arm, and the code is written in C# and RAPID.

In the first program I’ve made a slider to adjust the height (z-cord) of the robot while it moves in a square (xy- plane) .

In the second program a Kinect sensor is used to control the movements of the robot in a 2D - plane.

The source code in both projects are a mess (we had limited time on the project so I did not have time to clean it up), but they will give you a good idea of what’s needed to create an application using EGM.

14

I’m grateful!
Just a thought - is it possible to combine PC SDK and EGM ?

15

It should be possible, but I don’t know enough about PC SDKs and to make them, so I can’t be 100% sure. Maybe someone else knows the answer.

16

As long as you have both options enable in the system, it will work. The next question is the resource that you are trying to modify in the robot controller. EGM as today allows you to access motion data, while PC SDK gives you access to much more information.

Do you have an example of what you are trying to do?

17

At the moment I am just changing x, y position based on the computer vision application I have built, and some times I interrupt the execution of “MoveL” with a variable named “state” from a windows application.
Shortly:
RAPID:
…

CONNECT perslint WITH iroutine1;
IPers state, perslint;
…
MoveL Offs(home, x_pos, y_pos, 0), velocity, fine, puck\WObj := wobj0;
…
C#:
…

private ABB.Robotics.Controllers.RapidDomain.RapidData state_robot_rapid = null;
private ABB.Robotics.Controllers.RapidDomain.RapidData y_robot_target_rapid = null;
private ABB.Robotics.Controllers.RapidDomain.RapidData x_robot_target_rapid = null;
…
private Mastership mastership = null;
…
tRob1 = ctr1.Rapid.GetTask(“T_ROB1”);
…
mastership = Mastership.Request(ctr1.Rapid);
…

if (interrupt)
{
if (state_cp.Value == 1)
{
state_cp.Value = 0;
state_robot_rapid.Value = state_cp;
}
else
{
state_cp.Value = 1;
state_robot_rapid.Value = state_cp;
}
}
y_robot_target_rapid.Value = y_pos_cp;
x_robot_target_rapid.Value = x_pos_cp;

The result I get is 0.5 s delay before the robot starts the movement after the values are changed.
Since, the project I am working on requires a fast response from a robot(objects are moving up to 10 m/s) I came to the conclusion that it might be impossible to accomplish it with an ABB robot, only today have I found out about so called EGM. Still, I have no idea if it requires some sort of licence to implement it as my university is short of them.

18

EGM will help you to get a better performance. The first question I have is related to your hardware to check if it supports this robotware option.

Whicn robotware version do you have installed in your controller? Can you share the serial number, and if possible, a picture of the main computer.

BR

19

The lab staff told me that the current version installed was 6.02 as it was on the computer, but I’ve just checked it myself using flexpendant and I can see that it’s RW5.15_02.00.2005
Serial no: 120-501087
Probably it’s the reason why it says in the rapid program “Identifier not found: egmident”.

20

EGM is a Robotware option that is available in RW 6.x. You could check if the robot has another system or backup with RW 6.02, and then check if EGM is enabled in that system. If not, you will need a new license.