The low-cost MEMS sensor carried on the quadrotor UAV is susceptible to errors,such as installation,measurement and signal transmission,resulting in insufficient attitude output accuracy.For this question,a joint correction method for three-axis accelerometer 18 position and three-axis gyroscope was proposed based on Levenberg-Marquardt (L-M) algorithm.On the basis of considering sensor error,the effective output data of three sensors were collected and screened,the L-M algorithm was used to estimate the parameters of each sensor error model optimally.The experimental results show that after correction and compensation,the calibration indexes of accelerometer and gyroscope decrease by 98.96% and 74.33%,and the difference between the magnetometer data modulus and the local magnetic field intensity decreases significantly.The correction method can be used to effectively compensate the error of the four-rotor airborne MEMS sensor,and has good engineering application value.