In order to solve the shortcomings of traditional artificial spraying pesticides and more efficient prevention and treatment of diseases and pests, a pesticide spraying system based on sixteenrotor unmanned aerial vehicle (UAV) was designed. The sixteen-rotor UAV’s basic structure and attitude calculation method were explained. The whole system was full of 10kg, cruising speed can reach 5m/s, and the flight time was more than 10min. The traditional extended Kalman filter (EKF) attitude calculation method cannot meet the work requirements under the strong vibration condition of the large load UAV. The attitude angle calculation accuracy was not high and the attitude angle divergence was easily caused. A CPF-EKF algorithm based on 20 dimensional state quantity was proposed. The bias error of gyroscopes, accelerometers and magnetometers were added as the state quantity, which made the optimal estimation of the attitude angle of the three axes more accurate. Complementary filtering (CPF) was treated as an EKF detection module. When the EKF had the divergence trend, the EKF was reset, thus the EKF divergence was avoided simply and efficiently. Using actual flight data to verify the algorithm, the static experiment showed that the precision of the roll angle and the pitch angle of the algorithm were ±0.05°, the precision of the yaw angle was±0.2°. The dynamic experiment showed that the precision of the roll angle and the pitch angle of the algorithm were±0.1°, the precision of the yaw angle was±0.5°, and the algorithm had good real-time performance.