Giving Sense to Inputs: Toward an Accessible Control Framework for Shared Autonomy