Aiming at the problem that unmanned aerial vehicles （UAVs） are easy to overstep the boundary when rolling around the velocity vector axis， an adaptive sliding mode-based boundary protection control scheme is studied in this paper. Firstly， the attainable equilibrium set is improved based on dichotomy， and the command limitation method is introduced to realize online boundary solution and limited command generation. Then， to restrain the adverse impacts of uncertainties and disturbances to the closed-loop-system， a radial basis function （RBF） neural network is used to approximate the system uncertainties， and a nonlinear second-order disturbance observer is used to estimate the complex disturbances. Accordingly， an adaptive sliding mode-based boundary protection control law is designed to ensure the UAV’s safety and simultaneously realize the attitude tracking. Finally， the effectiveness of the adaptive sliding mode-based boundary protection control method is verified by numerical simulation.