frontline-drones

A descriptive, citation-grade catalog of the drone landscape: frontline military systems, popular commercial platforms and the open autonomy ecosystem, plus a reference index of NVIDIA's openly-published robotics / perception models on Hugging Face.

This is a reference. It documents what publicly exists, with sources. It contains no operational, assembly, flight-control, guidance, or targeting instructions, and nothing about modifying any platform for weapons. See DISCLAIMER.md.

🔎 Example output

Sample result format (illustrative values — run on your own data for real findings):

{
  "drones": [
    {
      "id": "drone-001",
      "battery_level": 75,
      "location": {
        "latitude": 37.7749,
        "longitude": -122.4194
      },
      "status": "online"
    },
    {
      "id": "drone-002",
      "battery_level": 50,
      "location": {
        "latitude": 34.0522,
        "longitude": -118.2437
      },
      "status": "offline"
    }
  ]
}

How it's organized

flowchart LR
  MIL[military systems<br/>UAVs · loitering munitions · USVs] --> CAT[(catalog CSVs)]
  COM[commercial + open stacks<br/>DJI · PX4 · ArduPilot · MAVLink] --> CAT
  NV[NVIDIA HF models<br/>GR00T · Cosmos · Nemotron] --> CAT
  CAT --> VAL[stdlib validator<br/>CI selftest]
  CAT --> DOCS[docs · counter-UAS sensor selection]
  NV --> INST[install_models.py<br/>multi-step menu installer]

Loading

What's inside

File	Rows	Contents
data/military_drones.csv	24	Frontline military UAVs, loitering munitions, USVs — published specs + operators + source
data/commercial_platforms.csv	14	Commercial drones (DJI, Skydio, Autel, Parrot, senseFly) + open stacks (PX4, ArduPilot, MAVLink, ROS 2, QGroundControl, Pixhawk)
data/nvidia_hf_models.csv	19	NVIDIA's open Hugging Face models (Isaac GR00T, Cosmos, RADIO, SegFormer, Nemotron, Canary/Parakeet, NV-Embed)

Human-readable write-ups:

• docs/military-systems.md
• docs/commercial-and-open.md
• docs/nvidia-models.md
• docs/counter-uas-selection.md — defensive drone-detection sensor selection guide
• SOURCES.md · CHANGELOG.md

Validate

python scripts/validate.py     # stdlib only; checks schema, keys, and source URLs

Domains

Primary domain: Defense & Aerospace · JTF MERIDIAN division: IRONCLAD · INDIA

Topics: cognis defense aerospace defense-tech drone

Part of the Cognis Neural Suite — 300+ source-available tools organized across 12 domains under the JTF MERIDIAN command structure. See the suite on GitHub and jtf-meridian for how the pieces fit together.

Install & run the NVIDIA models

The NVIDIA HF index isn't just a list — install_models.py is a customizable, multi-step menu that downloads and runs a catalogued model for its documented purpose (segmentation, ASR, vision-language, embeddings, LLM, vision backbone):

python install_models.py        # menu: pick category -> model -> install / download / write a run-script

It generates real, runnable transformers snippets where the model runs there; for models that need a special runtime (NeMo for ASR, the Isaac/Cosmos toolkits for GR00T/Cosmos) it downloads the weights and points you to the model card rather than faking a snippet.

Scope: this is general-purpose ML tooling. It does not wire any model into drone control, navigation, or targeting — see DISCLAIMER.md.

Sourcing & honesty

Specs are publicly reported figures and vary between sources (especially range/warhead for one-way-attack drones); representative published values are given with a primary source per row. The NVIDIA index lists model ids, purpose, modality, license, and the Hugging Face URL — confirm license/gating on each model card, as NVIDIA revises terms frequently. This catalog was compiled from the sources in SOURCES.md; verify against the primary pages before citing.

Related Cognis tools — defense / drone / maritime OSINT cluster

• awesome-drone-warfare-osint — citation-grade dataset (8,300+ components / 195+ platforms) + compliance query CLI
• uaslog — counter-UAS telemetry/log analyzer (detection events, RF bands, tracks)
• maritimeint — AIS grey/dark-fleet detection: watchlist + sanctions cross-reference
• geoaoi-pro — MIL-STD-2525 symbology + AOI helpers · geolens — image geolocation
• stixgen · attackmap · ttphunt — threat-intel tooling
• edgemesh — run any model privately across your own hardware

300+ open security & OSINT tools → github.com/cognis-digital

Integrations

Forward frontline-drones's findings to STIX/MISP/Sigma/Splunk/Elastic/Slack/webhooks via cognis-connect. See INTEGRATIONS.md.

License

Code: MIT (LICENSE). Data & docs: CC BY 4.0 (LICENSE-DATA).

---

📡 Interop map — how this repo composes with the rest of the Cognis suite (private-AI backbone, agent language + cognition, domain intelligence).