PAI-CW — Physical AI Reference Crosswalk

Engineering a Physical AI system requires two complementary representations: one for engineering questions, responsibilities, and evidence obligations, and another for typed facts and relations concerning the system and its task-relevant context. The Physical AI Reference System (PAI-RS) organizes the former through five Content Concerns—Task & Physical Context (T), Intelligent Capability (C), Capability Formation (F), Physical Grounding & Engagement (P), and Realization (R)—with cross-cutting Evidence, Verification & Assurance (V). The Physical AI Reference Framework (PAI-RF) organizes the latter through Scenario, Mechanism, Capability, Behavior, and Component Reference Domains, typed relations, cross-cutting concerns, and a Claim/Evidence/Assessment Overlay. PAI-CW V1.0 establishes a traceable, many-to-many crosswalk between these structures. It provides six Concern mappings, 28 directed relation-mapping records, coverage of all 13 normative PAI-RF Core relation typings, six Framework/Meta relation identities, and four Q relation identities. It also specifies identity and scope fields, rules for retaining non-equivalent source assertions, executable mapping records, and a structural validation procedure. The mapping rules address Capability Formation, the distinction between Physical Grounding and Physical Engagement, system boundaries, AI contribution attribution, evidence direction, and separate conformance claims. A hypothetical industrial workpiece-transfer example illustrates the method, while executable structural checks detect selected incorrect conversions. The reported tests establish structural consistency and error detection for the supplied package; assessments of physical performance, scientific validity, and source-system conformance require separate evidence.