Movement control relies on continuously mapping sensory inputs to movement parameters, allowing an animal to compare its posture with its behavioral goals.  This mapping must balance mechanical stability with the flexibility needed for goal-directed behavior in uncertain environments. In my lab, we investigate how self-generated visual signals, or visual feedback, contribute to this sensorimotor mapping by studying spontaneous, visually guided behaviors in Drosophila melanogaster. In this talk, I will discuss our efforts to reveal the functional organization of central multimodal circuits that process visual feedback to internally estimate self-motion and maintain body orientation as it navigates through space. We focus on how these distributed networks operate across multiple timescales to maintain stability while enabling goal-dependent locomotion.