Control with unstructured uncertainties refers to controlling systems where not only the parameters are unknown but also the way the parameters appear in the dynamics. This problem becomes pivotal in autopilots, where a unified control architecture is sought for aerial/ground/marine vehicles with different structures. By only making use of basic Euler–Lagrange (EL) properties valid in most mechanical systems independent of their specific structure, this brief proposes an adaptive design that does not rely on structural knowledge of the uncertainties. The proposed adaptive method, here validated in the ArduPlane module of ArduPilot, applies also to other modules like ArduCopter, ArduRover, and ArduSub. Enhanced performance with respect to state-of-the-art methods addressing unstructured and state-dependent uncertainties is verified.