Modeling Interaction Semantics in Shared Autonomy Frameworks

A new decision framework for shared autonomy introduces an explicit layer for interaction semantics, modeling candidate actions as proposals that are evaluated before execution. This allows for more nuanced representation of how different decision models influence final outcomes in human-AI systems.