Ability adaptive AAC: toward a modular, open-source framework

Technological advances in high-tech Augmentative and Alternative Communication (AAC) have expanded access for many individuals with complex communication needs; however, people with profound motor or cognitive impairments often remain underserved. Embodied communication is an essential component of human interaction, and prior research has demonstrated that individualized movements, postures, and vocalizations can be modeled computationally as communicative signals. Yet commercially available high-tech AAC systems generally treat embodied actions as mechanisms for accessing symbol-based interfaces rather than as communicative expressions in their own right. This design-oriented research note presents A3CP (Ability-Adaptive Augmentative Communication Platform), a modular AAC architecture for interpreting individualized, embodied communication through user-specific machine learning and communication partner clarification. An initial feasibility prototype was used to explore workflows for capturing and labeling embodied, communicative behaviors and training user-specific machine learning models. Findings from this exploratory work informed an architectural framework for coordinating multimodal input, uncertainty handling, communication partner clarification, and longitudinal user-specific model adaptation. Claims are limited to architectural specification and technical feasibility. This work provides an architectural foundation for future co-design and empirical investigation of AAC systems that can learn individualized, embodied communication.