blackridge-whaco 1.2.0

Wind-Harvesting Adaptive Controller for Operations — resource-aware distributed autonomy governor

WHACO — Wind-Harvesting Adaptive Controller for Operations

Bolt-on throttle governor that turns fuel waste into mission survival. Energy-constrained drone swarms lose 15-20% of missions to fuel exhaustion despite having sufficient reserves. WHACO creates fuel reserves during nominal operation and redistributes them after agent loss — delivering +9.8pp mission success with zero firmware changes.

Blackridge Autonomy LLC | v1.1.0 | February 2026 | blackridgeautonomy.com

---

The Problem

Multi-agent autonomous systems operating under hard energy constraints face a compounding failure mode: when agents are lost to attrition, comms degradation, or environmental stress, the remaining agents inherit expanded mission objectives but have no additional fuel budget. Fixed-throttle controllers cannot adapt — they run out of fuel and fail.

The Solution

WHACO is an 11-mode throttle governor that shapes burn rate in real time based on fuel pressure, wind state, and swarm connectivity. It runs as a companion-computer overlay on standard autopilots (PX4 via MAVLink OFFBOARD) — zero changes to firmware, zero changes to airframe, 15 microsecond compute overhead.

---

Validated Claims

Every claim has a traceable path: Claim → Test Script → Data Artifact → Quantitative Verdict.

#	Claim	Evidence	Effect Size
1	+9.8pp mission success over fixed-throttle (1,000 OOD environments)	TEST_REPORT.md	d=1.30
2	Mechanism is burn-rate shaping, not corridor exploitation	TEST_REPORT.md	Ablation + counterfactual
3	Generalizes across 1,000 out-of-distribution environments	TEST_REPORT.md	p<0.001
4	Scale-emergent multi-agent advantage at N>=50-75	TEST_REPORT.md	Crossover at N~75
5	PX4 SITL closed-loop validation: 7/7 scenarios pass	TEST_REPORT.md	7/7 PASS
6	15.1 us/call compute cost (1.8% of 50 Hz budget)	TEST_REPORT.md	1.8% overhead
7	Robust to +/-20% threshold perturbation	TEST_REPORT.md	No qualitative change
8	Robust to dropout timing (early/mid/late/burst)	TEST_REPORT.md	Burst amplifies advantage
9	EW modes (CLOAK/GHOST/SCUTTLE) with 6/6 RCS validation	VERSION.md	6.2 us/agent

See CLAIM_BOUNDARY.md for what does NOT hold — regime boundaries, explicit limitations, and honest failure modes.

---

Quick Start

Install

git clone https://github.com/morpheus18-glitch/whaco-rl.git
cd whaco-rl
pip install -e ".[all]"

Run Tests

make test-quick          # SDK + PX4 dry-run (~30s) → 2/2 PASS
make test-full           # Full 9-battery suite (~30 min)
make baqr-fast           # BAQR qualification (~3 min)

Use the SDK

from blackridge_governor import Governor

gov = Governor(fuel_budget=4.0)
thrust, mode_id, mode_name = gov.step(
    wind_x=-3.0, wind_y=0.5,
    fuel_remaining=2.8, dist_to_goal=45.0,
    groundspeed=6.2, throttle_pct=0.55,
)
print(f"Mode: {mode_name}, Thrust: {thrust}")
# Mode: CRUISE, Thrust: 0.55

---

Architecture

blackridge_governor/           SDK package
  governor.py                  11-mode throttle controller (15 us/call)
  ew_governor.py               EW swarm coordinator (formations, roles, 6.2 us/agent)
  demo.py                      SDK smoke test + reproduction runner

hil/                           PX4 SITL hardware-in-loop
  px4_sitl_harness.py          MAVLink companion controller (50 Hz)
  test_px4_harness_dryrun.py   128-check offline validation
  pegasus/                     Gazebo + UE5 4K visual simulation

results/                       Frozen publication data (immutable)
  pubA_200t/                   2,400 trials (PYTHONHASHSEED=42)
  baqr/v1.0/                   BAQR qualification artifacts

docs/                          Technical documentation
  baqr/                        BAQR scoring rubric, DARPA appendix
  FLIGHT_TEST_DESIGN.md        Phase II flight test protocol (98K)
  PATENT_STRATEGY.md           IP prosecution plan
  PROCUREMENT_PATHWAY.md       DoD procurement roadmap
  COST_MODEL.md                Operational economics

whaco_*.py                     43 test/benchmark scripts
reproduce_all.py               Single-command full reproduction

---

Test & Qualification Infrastructure

Suite	Script	What It Validates	Runtime
Stress (1-3)	whaco_stress_test.py	Structure index, compute, worst-case	~2 min
OOD (4-6)	whaco_ood_ablation_regret.py	1000-world generalization, ablation, regret	~5 min
Reviewers (7-9)	whaco_reviewer_killers.py	Calibration, counterfactual, significance	~3 min
Verifier	whaco_verifier.py	30 checks across 7 dimensions	~5 min
CTF Benchmark	whaco_ctf_tests.py	Multi-scale contested (2km/10km, 2D/3D)	~10 min
BAQR v1.0	whaco_baqr.py	6-tier formal qualification, 100-pt rubric	~8 min
Scale Sweep	whaco_scale_sweep.py	N=10-100 crossover analysis	~5 min
Kill Tests	whaco_kill_tests.py	Structural kill conditions (N=20, N=50)	~8 min
3D Benchmark	whaco_3d_benchmark.py	Altitude shear, climb penalty	~5 min
PX4 SITL	reproduce_all.py --with-sitl	7 closed-loop MAVLink scenarios	~10 min

All tests produce machine-readable JSON artifacts and publication-quality PNG plots. Full reproduction: PYTHONHASHSEED=42 python3 reproduce_all.py

---

BAQR v1.0 — Autonomy Qualification

BAQR (Blackridge Autonomy Qualification Range) is a formal benchmark suite for swarm autonomy evaluation.

python3 whaco_baqr.py --fast --all    # Quick (3 trials)
python3 whaco_baqr.py --all           # Full (15 trials)
python3 whaco_baqr.py --critique      # DARPA critique appendix

Tiers: T0 Regression → T1 Benign → T2 Contested → T3 Operational (LOW/MED/HIGH turbulence)

Awards: Gold (all tiers) | Silver (through T3-LOW) | Bronze (through T2) | Certified (T0+T1)

See docs/baqr/SCORING_RUBRIC.md for the 100-point rubric.

---

Documentation Map

Document	Audience	Purpose
TEST_REPORT.md	Technical reviewer	Formal evidence chain for all 9 claims
VALIDATION_PACKET.md	Program manager	6-tier CTF results with effect sizes
SBIR_TECHNICAL_VOLUME.md	Government	SBIR-formatted technical narrative
COMMERCIAL_READINESS.md	Investor	ICP analysis, pricing, GTM funnel
QUAD_CHART.md	Executive	1-page program office summary
CLAIM_BOUNDARY.md	Skeptic	Regime map + explicit limitations
INVENTION_DISCLOSURE.md	Legal	Patent novelty + prior art
VERSION.md	Developer	Canonical v1.1.0 spec
paper.md	Academic	Full technical paper (13 sections)
CHANGELOG.md	All	Version history

---

PX4 SITL Reproduction

Requires PX4 v1.14.3 + jMAVSim running in Docker:

# Terminal 1: Start PX4
bash hil/px4_sitl_setup.sh

# Terminal 2: Run validation
python3 reproduce_all.py --with-sitl

Expected: 7/7 scenarios PASS with telemetry in hil/telemetry/.

---

Docker

make docker-build          # Build image with smoke test
make docker-test           # Run quick tests in container

---

License

Copyright (c) 2026 Blackridge Autonomy LLC. All rights reserved. See LICENSE.

Third-party acknowledgments in NOTICE.

---

Citation

Blackridge Autonomy LLC. "WHACO: Wind-Harvesting Adaptive Controller for Operations —
Resource-Aware Distributed Control Governor with Scale-Emergent Attrition Resilience."
Technical Report TR-WHACO-2026-001, v1.1.0, February 2026.

---

Contact

Blackridge Autonomy LLC blackridgeautonomy.com [e-mail]