Physical AI Reference System (PAI-RS) — Definition

Physical AI Reference System (PAI-RS) V1 — Definition establishes a system-level reference structure for identifying the core engineering questions, relationships, responsibilities, and evidence obligations involved in understanding, constructing, comparing, and verifying heterogeneous Physical AI systems. Its Canonical Reference Core comprises five content Concerns: Task & Physical Context (T), Intelligent Capability (C), Capability Formation (F), Physical Grounding & Engagement (P), and Realization (R), linked by Typed Relations. Evidence, Verification & Assurance (V) operates across those Concerns. Concerns denote engineering questions and responsibilities, not fixed components of a system. The Definition distinguishes intelligent Capability from the conditions and processes through which that Capability is formed, maintained, or used. It separates Formation Dependency, Maintenance/Adaptation Dependency, and Runtime Dependency, as well as Physical Grounding from Physical Engagement. A three-level reference structure organizes the Canonical Reference Core, Engineering Contracts, and Profiles & Informative Views. Coupled engineering and assurance planes connect Concerns to responsibilities, representations and realizations, and connect Claims to verification objects, evidence requirements and results. Profiles retain implementation flexibility while informative technology views remain time-indexed and revisable. This document is the definitional authority for PAI-RS V1 and includes six supporting SVG figures. A separately published Specification (RS02) develops implementation, interface, verification and conformance contracts. The two publications have independent DOIs and version histories.