Focusing on the polar extreme environment, this paper proposed a small unmanned aerial vehicle for polar scientific research. To get precise attitude information, an adaptive Kalman filter algorithm is proposed to fuse GPS, gyro, accelerator sensor information. Moreover, an optimal fuzzy logic control algorithm is trained to adjust the heading angle in real time to improve system adaptability for the wind disturbance. Based on sensor information, the small unmanned aerial vehicle can keep a good performance in the following wind, upwind, turning, etc. Finally, the effectiveness of the small unmanned aerial vehicle is proved by a series of experiments and polar science research.