Delayed MuJoCo feedback: can an old observation predict now?

01 / QUESTION

Twenty-four paired traces compare delay, position noise and constant-velocity prediction, retaining incomplete recoveries.

METHOD

Freeze protocol and source before execution, save raw records, then independently recompute metrics and verify hashes.

RESULT

Eight of 24 rollouts pass; all 80ms-delay conditions fail stability despite reduced overshoot with prediction.

OBSERVATIONS

Eight of 24 rollouts pass; all 80ms-delay conditions fail stability despite reduced overshoot with prediction.

CONCLUSION

Conclusions are limited to the stated variables and acceptance criteria, retaining failures and unverified scope.

LIMITATIONS

Teaching-scale fixed conditions, not online model quality, production throughput or hardware safety. The complete assistant and VL01 remain unfinished.

REPRODUCE

The article provides a pinned revision, environment, runner and independent audit commands.

1. Engineering cases

   Delayed MuJoCo feedback: can an old observation predict now?

   Twenty-four paired traces compare delay, position noise and constant-velocity prediction, retaining incomplete recoveries.

   2026-10-079 MINPublished
2. PROJECT / 003

   Hohoo Embodied Experiments

   Two MuJoCo teaching scenes—pick-and-place and planar force control—record contracts, tuning, clearance and delayed-feedback outcomes, including failures.

   2026-09-28Experiment
3. Engineering cases

   MuJoCo obstacle clearance: a point path is not a body path

   Six contact trajectories compare direct, edge and clearance paths while separating geometry from controller tracking.

   2026-10-079 MINPublished