# Эксперимент: Constraint Shield for Learned Interaction Dynamics (#489)

{
  "worked": false,
  "confidence": 9,
  "verdict": "Built a short-horizon SLSQP predictive safety shield around a 1D nominal policy, with actuator and predicted-force constraints, plus reproducible target and stiffness sweeps. The exact static force-boundary prediction and zero shield-violation behavior were confirmed, and high-command baseline violations of 0.78–0.84 were eliminated. However, nominal dynamic violation onset was observed at target 0.08 versus the naive prediction 0.05, and the stiffness-scaling check missed its strict tolerance by 5.7e-05; under the required criterion that all derived predictions be confirmed, worked=false.",
  "metrics": {
    "baseline": "At target 0.20 and stiffness 10: RMSE=0.1337, violation_rate=0.84, peak_force=2.0. High-target baseline violation rates were 0.78–0.84.",
    "idea": "At target 0.20 and stiffness 10: RMSE=0.2102, violation_rate=0.0, peak_force=0.238. Across tested target and stiffness sweeps, violation_rate was 0.0 and minimum reported feasibility rate was 1.0."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 shield_experiment.py && /home/maxwelhelp/main/bin/python3 verification_report.py",
  "files": [
    "shield_experiment.py",
    "verification_report.py",
    "results.json",
    "verification_report.json",
    "run_output.txt",
    "verification_output.txt"
  ],
  "limitations": "The predictor and force model are hand-coded rather than learned from data, the environment is a simple deterministic 1D wall, and no neural-policy training, model mismatch, uncertainty calibration, disturbances, latency, or hardware/MuJoCo evaluation was tested. The optimizer uses generic SLSQP rather than a real-time constrained QP/NLP implementation."
}