Paper 002 · EXP-SURG-003

When Parameter Repair Is Not Enough

Failure-Conditioned Model-Order Expansion for Embodied Control

Seonhye Gu

Independent personal research

Tier C · confirmatory complete Preregistered Isaac Sim 4.1 / ORBIT Not peer-reviewed

An embodied controller can fail because its parameters are wrong — or because its model is missing a state variable the dynamics actually depend on. This paper tests, in simulation, whether observed failure evidence can tell those two cases apart.

Central question

When persistent, structured task failure survives the best repair available inside the current model class, does a prepared increase in model order improve prediction-linked control — without nominal regression?

Approach

Protocol diagram: static-first seed selection, shared Episode 1 drift evidence, zero-order parameter repair, preregistered adequacy gate, then a four-arm Episode 2 comparison.
Figure 1. Preregistered protocol and arms.

In an Isaac Sim surgical-reach environment, a zero-order target model is exposed to persistent drift on a shared Episode 1 trace, then repaired within its own model class. A preregistered adequacy gate asks whether the remaining residual is persistent, directional, and better explained by a prepared constant-velocity alternative.

Episode 2 compares four arms — frozen zero order, repaired zero order, gated constant velocity, and a velocity oracle — holding the Cartesian controller fixed across 10 fresh seeds × 10 fresh drift conditions, with every seed–arm–condition cell in its own Isaac process.

Results

Complete confirmatory grid: 400/400 valid cells, no missing data, unexpected resets, or forbidden-region violations.
Endpoint B · repaired
zero order
C · gated
const. velocity
C − B
H=10 prediction error 18.401 mm 7.595 mm −10.806 mm95% CI [−11.360, −10.331]
Fixed-horizon final distance 18.905 mm 5.601 mm −13.304 mm95% CI [−13.599, −12.982]
20 mm resolution rate 76% 100% secondary
Static retention 10/10 10/10 non-inferior
Per-condition C minus B effect sizes across all ten fresh drift conditions, all favoring the expanded model.
Figure 3. C−B effects across all ten fresh conditions — favorable in 100/100 paired cells on both continuous endpoints.

The adequacy gate fired on 100/100 persistent-drift trials and 0/100 on each of the static, observation-noise, and single-impulse controls — the expansion activated when structure was present and stayed silent when it was not.

Gate specificity across drift and three control conditions, alongside static retention results for both models.
Figure 5. Gate specificity and static retention.

On figures: every panel here is plotted directly from the frozen confirmatory artifact, and nothing is a stand-in.

Claim boundary

Supported

In the specified Isaac target-drift family, structured failure after zero-order repair warrants a prepared velocity-state expansion that improves prediction-linked control without static regression.

Not supported

Unconstrained model invention, arbitrary world-model expansion, tissue or contact validity, hardware transfer, clinical efficacy, or operating-room deployment.