# Эксперимент: Explicit-MPC Safety Shield for Neural Policies (#1007)

{
  "worked": false,
  "confidence": 9,
  "verdict": "Built an explicit affine-MPC-style safety shield for a disturbed 2D double integrator, including exact box-disturbance support checks, robust one-step screening, fallback control, and correction toward the fallback. The support-function error was 0 and the predicted local disturbance boundary gamma=2.0 matched the observed boundary 2.0; fallback frequency increased from 0.020 at policy scale 0.5 to 0.404 at scale 3.0. The overall safety win was not observed: at gamma=0.3 and 0.6, shield violation rates were 0.0087 and 0.173 versus 0.0027 and 0.136 unshielded, so this MVP does not justify claiming success.",
  "metrics": {
    "baseline": "Unshielded violation rates by gamma [0, 0.1, 0.3, 0.6, 1.0]: [0.000, 0.000, 0.0027, 0.1357, 0.5027].",
    "idea": "Shielded violation rates: [0.000, 0.000, 0.0087, 0.1727, 0.5180]; fallback rates: [0.0617, 0.0700, 0.2850, 0.5817, 0.8553]. Support error was 0; predicted/observed boundary was 2.0/2.0; closed-loop spectral radius was 0.6762."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 shield_experiment.py",
  "files": [
    "shield_experiment.py",
    "results.json",
    "README.md"
  ],
  "limitations": "This is not a true multiparametric-QP explicit-MPC partition: it uses one affine LQR fallback as a degenerate region and does not compute offline robust invariant or terminal sets. The correction uses a finite grid, the policy is synthetic rather than trained, disturbance rollouts are small, and no runtime/FLOP comparison or higher-dimensional benchmark was tested."
}