Flying insects manage to avoid collisions with obstacles, approach landing sites, and control complex dynamic flight manoeuvres. Bumblebees navigate over large distances by learning and subsequently retrieving visual information about the environment. Given the small number of neurons in insect brains, extracting such information needs to rely on extremely efficient mechanisms. These mechanisms are tightly linked to the animal’s sophisticated locomotion and gaze strategies. With a combination of experimental and modelling approaches we want to elucidate the computational principles insects evolved to solve these tasks.