White-Box Neural Ensemble for Vehicular Plasticity: Quantifying the Efficiency Cost of Symbolic Auditability in Adaptive NMPC

TL;DR

This paper introduces a white-box neural ensemble architecture for adaptive NMPC in vehicles, balancing rapid regime adaptation and high auditability, while quantifying the significant efficiency cost of symbolic graph maintenance.

Contribution

It proposes a novel white-box adaptive NMPC framework using symbolic graphs for transparency, enabling fast adaptation and rigorous auditability in vehicular control.

Findings

Rapid adaptation time (~7.3 ms) under regime shifts

Near-ideal tracking fidelity during compound regime changes

Symbolic graph maintenance increases solver latency by 72-102X

Abstract

We present a white-box adaptive NMPC architecture that resolves vehicular plasticity (adaptation to varying operating regimes without retraining) by arbitrating among frozen, regime-specific neural specialists using a Modular Sovereignty paradigm. The ensemble dynamics are maintained as a fully traversable symbolic graph in CasADi, enabling maximal runtime auditability. Synchronous simulation validates rapid adaptation (~7.3 ms) and near-ideal tracking fidelity under compound regime shifts (friction, mass, drag) where non-adaptive baselines fail. Empirical benchmarking quantifies the transparency cost: symbolic graph maintenance increases solver latency by 72-102X versus compiled parametric physics models, establishing the efficiency price of strict white-box implementation.