ROBOT DESCRIPTION

Describing a robot is not the same as describing its embodiment.

Traditional robot descriptions are excellent at representing structure, geometry and simulation-relevant properties. An AI system also needs explicit semantics about capability, limits, provenance and operational context.

CITATION RECORDCanonical source ↗
MATURITYREFERENCE SYNTHESIS
LAST VERIFIED22 Sep 2026
SOURCE VERSIONROS URDF + OpenUSD / UsdPhysics + OpenRGD import boundary
EVIDENCE BOUNDARYConcept synthesis; source description is treated as evidence and never as implicit runtime or actuation authorization.

STRUCTURAL DESCRIPTION

URDF answers foundational questions extremely well.

URDF is an XML robot description format centered on links and joints. It represents the robot model as a tree and can encode topology, inertial properties, joint limits, origins, dynamics and related structural information.

Those facts are foundational. They answer questions such as: what is connected to what, how can it move, what are the limits, and what physical properties were authored?

SCENE + PHYSICS

OpenUSD can represent richer composed worlds and physics.

UsdPhysics extends the OpenUSD scene graph with rigid bodies, joints, collision and simulation-oriented properties. It is useful for complex scenes and physically meaningful simulation assets.

OpenRGD does not attempt to replace the OpenUSD composition engine. Its current lightweight USDA importer intentionally supports only a narrow, auditable subset and refuses to pretend it has resolved semantics that require the full SDK.

THE SEMANTIC GAP

Structure alone does not tell an AI what the body means operationally.

01

Identity

Which physical robot or profile does this description belong to?

02

Capability

What tasks and actions are explicitly declared or supported?

03

Operational limits

What boundaries apply beyond raw joint ranges?

04

Safety context

Which constraints must remain explicit before execution?

05

Provenance

Which facts came from source evidence, calibration or later enrichment?

06

Lifecycle

How has the effective embodiment changed over time?

OPENRGD IMPORT

Extract evidence. Do not manufacture missing knowledge.

OpenRGD imports only source-supported Foundation evidence from supported URDF and USDA inputs. Missing values remain missing. Import does not create identity, safety policy, alignment, cognition, HAL bindings or hardware authorization.

01URDF / USDA
02Evidence extraction
03Partial Foundation
04Explicit enrichment
05Integrity check
06Grounding context

KEY PRINCIPLE

A description is evidence, not permission.

Even a complete structural description cannot prove that a robot is safe to operate or that a controller is correctly bound to hardware. OpenRGD preserves that distinction deliberately.

PRIMARY SOURCES