From visible motion to physical differences#

Inspect recorded states, native observations and physical references together. Each plot links its archived experiment; playback reads records without advancing physics or changing scores or tolerances.

Replay geometry keeps physical proportions; inspect small changes in the curves. Dashed lines are the declared analytical/engineering references, not measured material truth. Historical protocols do not form a universal engine ranking.

Six-engine incline: shared conditions, different responses#

A 40 mm, 64 g cube; static, sliding and nominal-zero-friction conditions at three timesteps. Curves and replays retain the original passes, failures and invalid records.

Six-engine incline: shared conditions, different responses
Static figures and full data (no JavaScript required)Static figures and full data (no JavaScript required)

Conditions： Native incline cohorts: MuJoCo 3.15.0, SuperDex 1.0.0, Genesis 1.4.3, Newton Physics 1.6.1, PhysX 5.9.0 and Drake 1.57.0; each records contact law, precision and applicable profiles.

Observation： All six have positive, negative and failed evidence. Genesis floors nominal-zero contact friction to .01; SuperDex BFGS/SR1 execute Newton-equivalent steps under assembly every step.

Explanation： Established effective-parameter/execution-path differences change the interpretation of same-named profiles. Initial pass rates do not replace equal-budget tuned comparisons.

Advice： Filter candidates by sticking/sliding, low friction and contact scale; inspect invalid records, physical error and cost together.

Limits： Cohort protocols and admission scope remain separate. The full matrix maps research scope, not universal rank. Numerical work remains in #157/#159/#163/#165.

Selection: MuJoCo uses the first frozen profile, PGS/elliptic. Other published baselines are SuperDex Newton/AUTO/C1, Genesis Newton/elliptic/Signorini, Newton Physics XPBD, PhysX PGS and Drake SAP/kLagged/hydroelastic. Selection does not use pass counts; all other profiles remain in the full matrix.

Reference: an ideal nonrotating Coulomb block with continuous contact has sliding acceleration a=g(sinθ−μcosθ); rest requires the static-friction inequality. Contact switching, geometry and model differences limit this reference. Drake’s initial 1 μm gap and Genesis’s native floor on nominal zero friction remain. Close-ups use the same body-follow camera rule and scale; the shared ruler retains world displacement. Invalid-record curves are diagnostic only.

Evidence and all configurations: Complete solver matrix · Engine reports · Each replay dataset contains original scores and source hashes.

Normal response: change parameters or the solver?#

The same MuJoCo fixture moves from about 80 kN/m toward the synthetic 20 kN/m target. Inspect loaded-window indentation, then the complete unloading trace; success belongs to the stated checks and conditions.

Normal response: change parameters or the solver?
Static figures and full data (no JavaScript required)Static figures and full data (no JavaScript required)

Conditions： A 40 mm cube, 0.2 kg, 0.5 ms; 2/4/6/−1 N loading. The 20 kN/m response is a synthetic target.

Observation： MuJoCo initially measured about 80 kN/m and approached 20 kN/m after calibration. Fixed parameters gave 10.01/40.04 kN/m at 0.1/0.4 kg; the predeclared mass conversion gave 20.04/19.93 kN/m.

Explanation： Established in this fixture: reference-acceleration response depends on mass and impedance. Equal parameter names do not establish equal material stiffness across engines.

Advice： Check effective response on declared fitting loads, apply an explicit conversion hypothesis, and validate other loads and masses independently.

Limits： This constant-impedance, four-point face-contact fixture does not supply a universal geometry formula or measured material calibration. Historical PhysX identity remains missing under the #126 audit.

Effective parameters: shared solimp=[0.9,0.9,0.001,0.5,2]; initial/calibrated solref and full native settings are in effective_parameters and original run.json. The −1 N separation segment remains; a separate loaded-window plot exposes small indentation. Cost and transfer figures below belong to separate historical protocols and are not pooled with this case.

Evidence and all configurations: Normal protocol, all 17 results and reproduction · Transfer · Cost boundaries

Response and cost

Mass / size transfer failures

Pinch: 0.4 N slips, 0.8 N holds#

The same Genesis fixture and object, changing the finger-joint force limit. Compare complete closing, lifting, holding and release, with an open-gripper negative control retained.

Pinch: 0.4 N slips, 0.8 N holds
Static figures and full data (no JavaScript required)Static figures and full data (no JavaScript required)

Conditions： A native Genesis finite fixture, four force caps and ±2 mm initial offsets: 16 fixed cases. Separate MuJoCo cube-pinch and robot SDF historical scenes also exist.

Observation： Genesis cannot retain the object at 0.2/0.4 N caps; 0.8/10 N holds and releases. This result is limited to the reported fixture and drive.

Explanation： Established: drive caps change load capacity. Friction/torque balance starts diagnosis; small contact-force residuals alone do not establish grasp success.

Advice： Start with object load, each side’s normal force and effective friction, then check finite pads, slip and release alongside actual drive output.

Limits： This is not a unified cross-engine pinch ranking. #145 inertia/frames and #132→#134 shared-drive/594-episode research remain independently open.

Reference: ideal static flat-pad retention requires μΣN≥mg; here mg/μ≈1.256 N. The plotted Σ|Fx| is a pad-force projection proxy, not measured actuator output. The 60 mm height threshold during 2–3 s is one original check; all other checks remain.

Evidence and all configurations: All 16 cases, failures and costs · Historical solver audit

LIBERO: native task, timestep control and observation fix#

Same actions, initial state, Panda and native success; original 2 ms versus unqualified 1 ms control. Heldout task success stays 10/10, screens change from 5/10 to 3/10, and cost doubles. Policy effects of observation refresh remain unverified.

LIBERO: native task, timestep control and observation fix
Static figures and full data (no JavaScript required)Static figures and full data (no JavaScript required)

Conditions： Official cream-cheese-to-basket, Panda / robosuite 1.4.0 / MuJoCo 2.3.7 Newton elliptic; 20 Hz control, original 2 ms step.

Observation： Original and 1 ms heldout action executions both complete 10/10 tasks; numerical screens pass 5/10 and 3/10, with about double stepping cost. Five settling steps change one end-effector position component by 4.481 mm while the old observation is retained.

Explanation： Observation assignment omission is verified. Positive contact time constants couple to the timestep safety floor in pinned source. Smaller steps do not consistently help; the mechanism behind 12.163 mm floor overlap continues in #174.

Advice： Refresh observations first; distinguish controller goals, actuator output and contact forces. Retain failures and localize controlled experiments to the actual contact. No new physics profile is recommended.

Limits： Demonstration action execution only: no fixed-policy closed-loop effect, real-material accuracy or mid-episode restore qualification. Holdouts are ten demonstrations of one task, not new-task generalization.

Default: first frozen development demonstration, independent of outcome. Additional views show the first heldout screen regression and maximum baseline overlap, explicitly post-hoc diagnostic selections; all ten pairs remain in the full results. Video rendering never integrates physics.

The 1 mm dashed line is a declared numerical screen. Relative height includes rotation; normal force is actual contact readback. Native time = display time + 0.25 s; channels retain their pre/post-integration epochs.

Evidence and all configurations: Native task, all configurations, failures and reproduction · Remaining attribution #174