A Pattern Language for Head-Mounted AR in Museums: Parameterized, Composable, and Reproducible
Head-mounted AR (HMD-AR) is entering museums and exhibitions, yet creators such as AR developers and designers lack a reusable, parameterized, and verifiable interaction vocabulary for first-time visitors. We present a pattern language for cultural-heritage HMD-AR organized into three classes—PoI Guiding, Experience Indicating, and Experience Presenting—with nine sub-patterns that specify default safety bands and composition rules. Patterns map one-to-one to Unity prefabs on an instrumented runtime with unified logging. Using a Design Science Research (DSR) combined with Research-through-Design (RtD) approach, we instantiated two in-situ prototypes at European natural-history exhibits and evaluated them through creator panels (AR developers and cultural-heritage practitioners, N = 10) and visitor studies (N = 68) on HoloLens 2 and Quest 3. Experts judged the language clear and recommendable, and in-museum results met accepted thresholds for usability, workload, user experience, and presence, with stable behavioral logs under the studied operational conditions. Cross-site synthesis across two natural-history museum deployments showed directionally consistent outcomes and suggested that the interaction layer could be transferred with low effort within this scope: deployments required parameter localization, while core compositions and state machines remained unchanged. We contribute a parameterized, composable interaction language; aligned reference implementations and prototypes; creator- and user-side evidence; and a standardized mapping from constructs to metrics, measures, and decision rules for reproducible cross-study comparison. We release pattern cards, parameter documentation, and reusable Unity prefabs through a public GitHub repository to support transparent reuse, replication, and cross-study comparison.