HMAA Repository

Authority-Governed Assured Autonomy Research Platform. Governance FSM with a bounded TLC model check of its discrete model, Dempster-Shafer trust fusion, 98 tests, 7 adversarial experiments, deterministic simulation. 42-file Python package + browser simulation dashboard.

hmaaai-governanceauthority-computationautonomous-systemsprovisional-application-filed
4,500+ lines 250KB+ HTML / JS / CSS
main 42+ files · Mar 7, 2026
Open Live
hmaa-simulation.html HMAA v2.4 authority computation engine Mar 7, 2026
LICENSE MIT License Mar 7, 2026
hmaa-simulation.html 4,500+ lines · 250KB+
  1<!DOCTYPE html>
  2<html lang="en">
  3<head>
  4<title>HMAA Simulation Dashboard</title>
  5
  6/* HMAA: Human-Machine Authority Architecture
  7 * U.S. provisional patent application No. 63/999,105
  8 * Filed: March 7, 2026
  9 * Inventor: Burak Oktenli (Sole)
 10 * Disclosure DOI: 10.5281/zenodo.18861653
 11 */
 12
 13// Full source: 4,500+ lines
README.md

HMAA: Human-Machine Authority Architecture

HMAA is an Authority-Governed Assured Autonomy Research Platform implementing a formally specified governance architecture for autonomous rovers (simulation; the rover build is in progress). Core pipeline: six physics-based sensor models produce Dempster-Shafer belief masses, Shafer discounting weights evidence by cross-sensor validation, fused trust drives an authority FSM with hysteresis, dwell enforcement, and oscillation lockout, whose discrete model has been checked with TLC at small bounds. Phase 1 baseline: 89/100 engineering score.

Provisional Patent Application Status

U.S. provisional patent application No. 63/999,105
Filed: March 7, 2026 · Receipt: 74759595
Status: Filed (provisional; not examined or published; no patent granted)
Disclosure DOI: 10.5281/zenodo.18861653

Key Features

  • 42-file Python package with pip-installable architecture
  • 98 automated tests (D-S engine, FSM, sensors, harness, conformance, twin, validation)
  • 7 adversarial experiments: camera occlusion, LiDAR spoofing, RF jamming, IMU drift, compound attack, cross-sensor, recovery
  • TLA+ specification of the discrete authority model (HMAA_Authority_FSM, Zenodo 10.5281/zenodo.21300977), checked with TLC at small bounds: 23,748 distinct states, depth 9, no counterexample; 4 of 8 properties exercised, 4 vacuous at these bounds (see Formal Coverage Status)
  • Dempster-Shafer trust fusion with Shafer discounting and cross-sensor validation
  • Authority FSM with hysteresis, dwell enforcement, oscillation lockout
  • CARA recovery mode manager with command restrictions
  • Predictive digital twin with risk zone detection
  • Deterministic simulation: the same seed is designed to reproduce the same SHA-256 trace hash within the same runtime configuration
  • Monte Carlo statistical analysis (N-trial)
  • Immutable trace artifacts with schema versioning and full-artifact hashing
  • Browser-based simulation dashboard (HTML)
  • GitHub Actions CI pipeline

Technical Specifications

  • Language: HTML, JavaScript, CSS (single-file architecture)
  • Dependencies: None (zero external libraries)
  • Runtime: Browser-based, client-side only
  • Lines of code: 3,228

Author

Burak Oktenli
Independent researcher
ORCID: 0009-0001-8573-1667
Contact: info@burakoktenli.com

License

Code: MIT License (the HMAA Zenodo record is also MIT). Copyright © 2026 Burak Oktenli · Independent researcher · ORCID 0009-0001-8573-1667 · Miami, Florida · Licenses vary by artifact