Our top priority

Hi!
I’ve been working with force control and have a question:

What are the optimal settings for the inertia and damping parameters. Could someone provide some combinations that work well with the robot? I’ve been playing with these settings and can get the inertia parameters for translation to work like I would like them too, but for the rotational DOFs they are either too stiff or jumpy.

Is it possible to keep the orientation of the robot end effector constant with respect to the world coordinate system but have the position under admittance control?

I’d also like the default settings for the robot.

Thanks
Guru

Hi Guru

Here are my answers to your questions:

Q: What are the optimal settings for the inertia and damping parameters. Could someone provide some combinations that work well with the robot? I’ve been playing with these settings and can get the inertia parameters for translation to work like I would like them too, but for the rotational DOFs they are either too stiff or jumpy.
A: I can’t give you the perfect combination. Normally the default parameters work fine, but not in the case of your robot apparently. Here are some debugging steps before trying to change the inertia and damping parameters:
1) Insure your torque sensors are well calibrated. To do so, follow the ‘Zero torque’ calibration procedure detailed in the ‘Advanced specification guide’.
2) Use the optimal gravity mode. To do so, you will have to perform an automatic calibration trajectory with your robot in order to get some parameters (RunGravityZEstimationSequence). Then you will have to send these parameters to the robot and to activate the optimal gravity mode (SetOptimalZParam and SetGravityType)
3) If both of these previous steps are not enough, you can try to tweak orientation admittance parameters, while specifying inertia and damping values close to the default values.

Q:Is it possible to keep the orientation of the robot end effector constant with respect to the world coordinate system but have the position under admittance control?
A: You could increase a lot the damping and inertia parameters in orientation while keeping those in translation to their default value. This way, the admittance orientation command will always be close to zero. Also, please insre your arm is set in Fixed mode (and not in Rotating mode) if you want to keep the orientation constant. By default, the arm is the Rotating mode.

Q: I’d also like the default settings for the robot.
A: Do you have a Jaco or a Mico arm?
For Mico:
Damping_X=20.8 N/(m/s)
Damping_Y=20.8 N/(m/s)
Damping_Z=20.8 N/(m/s)
Damping_ThetaX=0.779 Nm/(rad/s)
Damping_ThetaY=0.779 Nm/(rad/s)
Damping_ThetaZ=0.779 Nm/(rad/s)

Inertia_X=5.0 N/(m/s^2)
Inertia_Y=5.0 N/(m/s^2)
Inertia_Z=5.0 N/(m/s^2)
Inertia_ThetaX=0.14 Nm/(rad/s^2)
Inertia_ThetaY=0.14 Nm/(rad/s^2)
Inertia_ThetaZ=0.14 Nm/(rad/s^2)

For Jaco V2
Damping_X=18.18 N/(m/s)
Damping_Y=31.82 N/(m/s)
Damping_Z=54.55 N/(m/s)
Damping_ThetaX=0.85 Nm/(rad/s)
Damping_ThetaY=0.85 Nm/(rad/s)
Damping_ThetaZ=0.85 Nm/(rad/s)

Inertia_X=6.36 N/(m/s^2)
Inertia_Y=11.14 N/(m/s^2)
Inertia_Z=19.10 N/(m/s^2)
Inertia_ThetaX=0.21 Nm/(rad/s^2)
Inertia_ThetaY=0.21 Nm/(rad/s^2)
Inertia_ThetaZ=0.21 Nm/(rad/s^2)

Q: If I use a different end effector, ie, change the gripper to something else, how do I modify force control so it knows the new kinematics/dynamics of the robot? Would RunGravityZEstimationSequence function help estimate the new parameters?
A: Yes, you can run the automatic calibration trajectory using RunGravityZEstimationSequence, and this function will output you the right gravity parameters with your new gripper. Of course, run RunGravityZEstimationSequence with your new gripper. Another method would be to use the TorqueConsole and specify the mass and center of mass of your new gripper in the Payload section.

Hope this helps you 🙂
Martine

Hi Martine,
Thanks again for your help. I have another question:

When setting the torque zero using the Torque zero GUI option, is this reflected in the Angular Monitoring screen? When I tried, it did not change.

Thanks again for your help,
Guru

Hi Martine,
Thanks for your support, its been really helpful. I think I’ve come across a bug, but I might be using the API incorrectly.

When I command a cartesian position and then enable force control, the robot shifts to force control but with impedance set at default values w.r.t the angular coordinates of the joints. I can kind of get around this by pressing the home button on the joy stick when its in force control. I tried many things including the function: SetCartesianControl, but this doesn’t seem to work. If this is a bug is there a work around? If it isn’t please let me know what I’m doing wrong.

Thanks a lot!
Guru

Hi Martine,
I’m sorry I wasn’t clear, maybe a code snippet will help:

result = MyInitAPI();
MyMoveHome();
float OptimalParam[OPTIMAL_Z_PARAM_SIZE] = {0.86285,0.0180729,-0.00496937,-0.854213,
                                            0.00472874,0.482024,0.00124092,0.151529,-0.00096519,
                                            -0.00816589,-0.329973,-0.128842,0.238235,0.058439,-0.0345066,0.111202
                                           };
MySetGravityOptimalZParam(OptimalParam);
MySetGravityType(OPTIMAL);
MyInitFingers();
TrajectoryPoint pointToSend;
pointToSend.InitStruct();
pointToSend.Position.Type = CARTESIAN_POSITION;
pointToSend.Position.CartesianPosition.X = -0.3;
pointToSend.Position.CartesianPosition.Y = -0.2;
pointToSend.Position.CartesianPosition.Z = 0.2;
pointToSend.Position.CartesianPosition.ThetaX = PI;
pointToSend.Position.CartesianPosition.ThetaY = 0;
pointToSend.Position.CartesianPosition.ThetaZ = 0;

pointToSend.Position.Fingers.Finger1 = 6400;
pointToSend.Position.Fingers.Finger2 = 5500;
pointToSend.Position.Fingers.Finger3 = 0;

MySendBasicTrajectory(pointToSend);
usleep(1000);

pointToSend.Position.CartesianPosition.X = -0.3; 
pointToSend.Position.CartesianPosition.Y = 0; 
pointToSend.Position.CartesianPosition.Z = 0.1;

MySendBasicTrajectory(pointToSend);
usleep(1000);

cin.ignore(); // Key press after robot reaches required location

result = (*MyEraseAllTrajectories)();
result = (*MySetCartesianControl)();
result = (*MySetFrameType)(0);
result = MyStartForceControl();
for(int a =1; a != 0;)
    cin >> a;// Press any key except 1 to exit
result = (*MyStopForceControl)();
result = (*MyCloseAPI)();

I believe the manipulator is in a valid position. When I activate forcecontrol in the home position and then move it to the desired location manually by physically moving it, it stays in forcecontrol. But when I ask the robot to reach the desired position via code ie:

MySendBasicTrajectory(pointToSend)

and then activate forcecontrol, it flips to joint space impedance. If I press the home button on the joystick for a brief moment just so the robot barely moves, the robot provides Cartesian impedance. I’m using the button press on the joystick as a work around.

Thanks again for your help!
Guru

• This reply was modified 2 years ago by  gurusub90.

HI Guru

I would like to answer your previous questions:

Hi Martine,
Thanks for the response! It was very helpful! Is there any way to have gravity compensation on with the robot in a different orientation? My robot is a Mico2. R: For the moment, only the manual gravity compensation model is available for other orientation of the robot. This mode is activated by default on the robot. You just have to specify the new gravity vector using the Torque Console or the API. I am currently working on a version of the Optimal gravity compensation model that would work with the robot in any orientation.

Do you have documents specifying the volume of usable space around the robot, locations of singularities etc, this would be very useful information. R: I have a quick sheet that describes the singularities. For the workspace, I do not have a formal sheet that shows you the volume of usable space around the robot. The Advanced Specification Guide tells you the angle limits of the robot. The reach or a Mico arm is 70 cm, which means the robot will be able to reach a maximum height of about 1 m over its base, and more or less 40 cm under. Does this helps you? I’ll see if I can provide a more visual representation of the workspace around the robot.

Thank you!

Martine