Inverse dynamics validation for cable-driven bar using spatial tendon (forward vs inverse consistency question)

[JorgeBonilla99]

Intro

Hi!

I am a graduate engineering student working on a rehabilitation robotics prototype. Im new to the use of MuJoCo to model a cable-driven mechanism where a bar is actuated through a tendon routed like a pulley system.

My setup

I am using:

MuJoCo (latest version)
Python
MJCF XML model
Windows
No contact friction in the model (idealized system)

My question

I am trying to validate inverse dynamics vs forward dynamics consistency for a cable-driven system.
My model consists of:

• A rotating bar attached to a wall through a hinge joint (no actuator torque at the hinge).
• A sliding box connected to the bar through a spatial tendon.
• The tendon is routed via an intermediate site acting as a pulley-like routing point.
• The system is frictionless and all actuation comes from the sliding box.
  My goal is to perform three steps:
  1. Inverse Dynamics:
  I define a desired angular trajectory θ(t) for a rotating bar (for example up to +-25° periodic motion). Using inverse dynamics, I want to compute the horizontal force F(t) that the sliding box must apply (through the tendon) to produce exactly this motion.(NO torque in the bar)
  2. Forward Dynamics (validation):
  Using the computed F(t), I then apply this force as an input on the sliding box and run forward dynamics. My expectation is that the resulting bar trajectory should reproduce the same θ(t) used in inverse dynamics. However, I am not getting consistent results. In theory, both should match if everything is correct.
  3. Play with τ(t) +F(t) GOAL:
  With the box removed after inverse and forward, I compute the joint torque τ(t) required to follow the same trajectory θ(t) at the bar. My goal is to compare required joint torque τ(t) at bar without assistance of the box so that later I can play around how much assistance the cable+box provides. example: Desired θ(t)= %torque +%Force

BoxRopeBar.xml
InverseFoward Simulation.py

Minimal model and/or code that explain my question

I'm not sure whether the issue is related to the XML file parameters or the time step. From the documentation, I understand that the default integrator is Euler with 0.002time, so I'm not sure what the problem might be. Any help would be greatly appreciated. Thanks in Advance

Confirmations

• I searched the latest documentation thoroughly before posting.
• I searched previous Issues and Discussions, I am certain this has not been raised before.