Description
Intro
Hi!
I am a graduate student at HKU, I use MuJoCo for my research on robotic manipulation.
My setup
MuJoCo 3.2.7, Python and C, Linux
What's happening? What did you expect?
While experimenting with a hinge joint that includes frictionloss, I encountered an unexpected and unphysical behavior. After simplifying the model, I found that when using a single-joint setup with all of the following parameters set to values greater than 0—noslip_iterations, frictionloss, and damping—the deceleration of the joint becomes excessively slow.
Specifically, when I load the model and keyframe it, I expect the cylinder to decelerate to zero speed quickly due to the effects of friction and damping. This behavior works as expected when any one of the three options is turned off. However, when all three are active, the cylinder continues rotating, but the deceleration is very gradual.
Questions:
- Is this slow deceleration the expected behavior (documentation mentions that noslip solver "no longer be solving a well-defined optimization problem")?
- Could I be misusing one or more of these options?
- Or, is this potentially indicative of a bug in the simulation?
Steps for reproduction
Load the model, and load keyframe 0.
Minimal model for reproduction
<?xml version="1.0" encoding="utf-8"?>
<mujoco>
<compiler angle="radian"/>
<option timestep="0.01" integrator="implicit" noslip_iterations="1" />
<worldbody>
<body>
<joint type="hinge" axis="0 0 1" frictionloss="0.01" damping="0.1" />
<geom type="cylinder" size="0.0125 0.005"/>
<site pos="0.0125 0 0" rgba="0 1 0 1"/>
</body>
</worldbody>
<keyframe>
<key qpos="0" qvel="10"/>
</keyframe>
</mujoco>Code required for reproduction
No response
Confirmations
- I searched the latest documentation thoroughly before posting.
- I searched previous Issues and Discussions, I am certain this has not been raised before.