A small question about how to calculate joint torque to decrease joint velocities and accelerations more quickly? #1952
Replies: 7 comments
-
Did you try removing the |
Beta Was this translation helpful? Give feedback.
-
And as I read from this issue #1095, can I use the method mentioned in this issue,
to measure the real joint torque applied to each joint after I individually set the joint torque equal to the results from the joint impedance controller? Namely,
Because I suspect, because the joint torque computed is so large, with the ctrl range limits set in each actuator, the actual joint torques applied to each actuator is actually very small, which result in the very slow decreasing speed of joint accelerations and joint velocties. Updates:
|
Beta Was this translation helpful? Give feedback.
-
Hi, Yuval, Thanks for your reply. I removed the ctrlrange defined in the actuators, but it presented the same result. Here is a minimal example of the panda xml file I used in my simulation. I used the model from the mujoco_menagerie and listed the main model definition as follows,
As for the joint impedance controller, I just implemented it with its classic theory as follows,
where qvel_current, qpos_current are read from the data, and qacc_desired, qvel_desired and qpos_desired are planned. If you have free time, could you help me check it and give me some advice on these problems including the issues I mentioned before like #1909. Thanks very much for your help and time. Really appreciate that. Btw, I also want to show great thanks to this wonderful simulation tool, your and other developers' quick and detailed responses to issues which helps me get more familiar with this powerful simulation tool to do robotic simulation, for instance, some basic questions like how to calculate the derivatives of the jacobian matrix, how to measure the contact force ect. Really appreciate. |
Beta Was this translation helpful? Give feedback.
-
I suspect your actuators are inheriting the |
Beta Was this translation helpful? Give feedback.
-
Yes, I noticed it so I commented out this line as I showed in the xml file above, but it also didn't work. Or just commented out this is not the right way to remove the inheritance relationship? |
Beta Was this translation helpful? Give feedback.
-
In that case I'm not sure... Anyway, why are you implementing you own actuator rather than using the built-in ones? |
Beta Was this translation helpful? Give feedback.
-
Well, I just hope to make the simulation close to the real scenes, I mean, hope that the joint position limits and joint torque limits listed in the official web as https://frankaemika.github.io/docs/control_parameters.html could be well meet during the simulation. You mean the actuators defined originally in the mujoco_menagerie? I tried just now, it doesn't work neither. And it seems lack some torque limits on joint torques. Thanks for your time and help, Yuval. May I just make a small summary of some questions I am confused here so that you can give me a quick check?
Btw, I also referred to some other developers' codes in panda robot simulation, however, most of them simply use methods like mocap, position-type actuators instead of motor-type actuators because we have different focus in simulation. Anyway, thanks very much for your time and help. Wishing you all the best. |
Beta Was this translation helpful? Give feedback.
-
Hi,
I'm a student and I'm trying to use MuJoCo for robotic simulation.
I'm looking for some help with the joint torque command.
I used the motor-type actuators in panda robot simulation. Here is my problem.
It could be seen that the joint acceleration decreases due to the joint torque with right signs,but what I expect is that the joint acceleration could be decreased much more quickly so that the joint velocities could reach 0 more quickly.
And, yes, the joint torque seems very high due to the high damping coefficients I set for the robot. However, I set the ctrl range in the xml file, I guess after I set the ctrl range, the actual torque sent to the joint will be automatically clipped into this range, right? Is this the reason why the joint acceleration changes not too much after mj_step?
So, how could I enable the joint velocities and accelerations decrease to 0 quickly with the right joint torque?
Thanks and wishing you all the best.
Beta Was this translation helpful? Give feedback.
All reactions