Getting from one place to another in a directed manner is one of the defining abilities of complex animals. Based on sensory information, previous experience, and motivational state, animals generate an internal target heading and compare it to their current heading; any detected mismatch will cause corrective steering movements. Neurophysiological and anatomical data across insects implicate the central complex as the likely neural substrate for navigation. Using a combination of neurophysiology, light-microscopic anatomy, serial section block-face electron microscopy, behaviour, and computational modeling, we investigate species with sophisticated yet diverse navigation behaviours, including bees, ants, and migratory moths. We study how these animals compute their current heading, how they encode their momentary target directions (their ‘goals’), and how they compare these two angles to initiate steering. We have established an anatomically constrained model of path integration, developed hypotheses about a possible ancestral circuit layout, and identified additions for straight line orientation, exploratory navigation, and long-range migration. We have also begun to develop new hardware implementations of the core circuit. Using nanowire-based optical components, we aim to create an embodied version of the central complex that can serve as a fast, autonomous decision-making device operating with minimal energy expenditure, akin to an insect brain.