Reading the Verdict: Did It Settle?
Learn that touching the target once is not passing — add damping so the device truly SETTLES (velocity → 0), and read the simulator's settle verdict correctly.
Try this first — before any explanation.
The device now has the right mass and stiffness, so it reaches about -0.10 m — but with almost no damping it overshoots to -0.114 m and keeps bouncing. The verdict reads 'reached the target but still moving.' A real device has to come to REST. Add damping until it settles.
Edit the MJCF device model and simulate it on real MuJoCo physics — tune the parameter so the part settles at the spec.
The idea, built visually.
Here's the trap: the load swings through -0.10 on its first bounce, so a naive check would call it done. But watch the velocity — it's still flying. A real part has to stop there, not ring forever. Damping is what bleeds the energy out: too little and it oscillates for ages, enough and it eases to rest. So the verdict has two parts, not one — is it AT the target, AND has it STOPPED? Raise the damping and watch the wobble die.
▣ Stage animation: Two runs overlaid: low-damping trace oscillating around -0.10 (a red 'still moving |v|' meter pegged) vs a higher-damping trace easing smoothly to -0.10 (green 'settled |v|≈0'); a verdict card flips from 'reached but moving' to 'SETTLED'.
Build it up, step by step.
1. See the problem. Simulate the starter: it crosses -0.10 but the verdict is 'still moving' — low damping (0.2) lets it ring.
2. Add damping. Raise the joint's damping="0.2" toward 6 so energy bleeds out and the load eases to rest.
3. Read the verdict. PASS needs BOTH: settled AT -0.10 AND velocity ≈ 0.
How the Bench grades your run.
PASS WHEN PASS when the load is BOTH at -0.10 m (±0.006) AND has stopped (|velocity| < 0.02). Reading the full verdict.
- 'Reached but still moving' — it crossed -0.10 but is oscillating; raise the damping (toward 6) so it settles.
- Over-damped and sluggish — it creeps and hasn't reached -0.10 in the run time; ease the damping back down a bit.
- Settled at the wrong place — check mass (0.5) and stiffness (49); damping changes HOW it settles, not WHERE.
Bring back what you've already mastered.
- A sim verdict has two parts: the output is AT target AND has ____ (stopped / velocity ≈ 0).
- What physical role does damping play? (bleeds energy so oscillation dies and the system settles).
- Damping changes ____ the device settles, not ____ it settles (how; where).
What you must demonstrate to advance.
Add enough damping that the device settles (velocity ≈ 0) at the target — and correctly read the two-part settle verdict.
How this feeds your build.
Reading the settle verdict is exactly how the M5 capstone judges whether your designed device actually works under load.