Origins & History

HCDF (Hardware Configuration Descriptive Format) has been in development since 2020, emerging from real-world experience building autonomous systems where existing robot description formats could not capture the full picture. Over six years of work across DARPA-funded research programs, automotive Ethernet development, and open-source robotics, the need became clear: a single format that describes the complete cyber-physical system, not just kinematics (URDF) or simulation worlds (SDF), but the full stack including physical structure, sensors, actuators, networking, firmware, and device identity.

Motivation

Existing robot description formats (URDF, SDF, MJCF) describe the physical structure of robots (joints, geometry, inertial properties) but a real robot is far more than its kinematic tree. Modern robots have network topology connecting dozens of devices, wired and wireless connectivity across multiple bus types, deterministic scheduling for real-time control, link-level security, power distribution with batteries and fuel systems, firmware running on every device, sensor hardware with specific noise characteristics and sampling configurations, motors with electrical limits and thermal constraints, and human-machine interfaces. None of this can be described in URDF or SDF.

AI agents and autonomous configuration tools need a complete embodiment description, not just joint angles, but the full cyber-physical system: what sensors are connected via what networks, at what data rates, with what power budget, running what firmware. The gap between "robot description" and "system configuration" needed to be closed.

Existing formats describe physical structure but ignore networking, power, firmware, and device identity
Real robots have complex network topologies, multiple bus types, deterministic scheduling, and link security, none of which is describable in URDF/SDF
Sensors, motors, and energy systems have hardware-specific parameters (noise models, load curves, charging profiles) that simulation formats don't capture
AI agents need the complete picture (topology, data rates, power budgets, firmware versions) to reason about and configure a system
Hardware vendors need a verified way to distribute component descriptions that system integrators can compose into complete robots

Key Features & Innovations

First format combining kinematics + networking + firmware + discovery in one schema. HCDF puts physical structure, network topology, TSN scheduling, firmware identity, and device discovery into a single XSD-validated XML file.
SHA-based content addressing for model files and firmware. Every referenced artifact (GLB models, firmware images, stream profiles, included components) carries a SHA256 hash, enabling verified distribution and cache deduplication.
Stream profiles separate application-level data flows from hardware topology. A .streams.xml file defines traffic classes, bandwidth reservations, and Qbv schedules independently of the physical wiring.
Agent-ready design. HCDF is designed for autonomous configuration by CNC/TSN agents using YANG. An agent can read the HCDF file, determine what devices exist, what networks connect them, and what schedules are needed, then configure the network accordingly.
Extensible via XML namespaces and domain-versioned extensions. The reverse-DNS extension mechanism (org.ros2, org.gazebosim, org.cognipilot.imu-stability) allows vendor and community extensions without schema conflicts.

Ecosystem

Dendrite: 3D visualization and HCDF builder tool for CogniPilot systems. Fetches HCDF components from hcdf.cognipilot.org, renders GLB models, displays port locations and network topology.
MCUboot: Firmware versioning integrated via <software>/<hash> elements. Devices report firmware identity so agents can verify running firmware and trigger OTA updates when mismatches are detected.
hcdf.cognipilot.org: Remote HCDF component hosting with SHA-based caching. Hardware vendors publish components describing their boards; system integrators compose them into full robot descriptions.
Vendor distribution: Hardware vendors provide HCDF components describing their boards. SHA verification ensures integrity. Components compose into system-level descriptions via the <include> element.

Acknowledgments

HCDF development has been supported by and inspired by collaborations with:

DARPA: Research programs advancing autonomous systems
NXP Semiconductors: Automotive Ethernet, TSN, S32 platform
Purdue University: Robotics research
AFRL: Air Force Research Laboratory
Infineon Technologies
The Zephyr RTOS community
And the broader CogniPilot open-source community