Focusing on the high wind disturbance in the polar environment, an adaptive Kalman filter algorithm with radial basic function neural network was proposed to improve attitude information performance for the small unmanned aerial vehicle. Based on the unmanned aerial vehicle situation information and sensor information, system adjusts the weights of the measurement noise matrix and the estimated parameters in real time to get precise attitude information. Moreover, a vector field path following algorithm was proposed to improve small unmanned aerial vehicle performance in the following wind, upwind, turning, etc. Using the predefined trajectories as reference, system adjusts course in real time to realize precise path-following control. Finally, the effectiveness of the small unmanned aerial vehicle was proved by a series of simulations and tests. The adaptive Kalman filter can provide long time high precision attitude information for the small unmanned aerial vehicle, and the mean trajectories error is less than 10�fm in the environment test with 6 degree wind disturbance. The small unmanned aerial vehicle successfully realized the research tasks in south polar research.