Abstract: An intelligent vacuum cleaner system using MPU9250 nine-axis attitude sensor and STM32F103 which acquire and process the attitude data is designed.Aiming at the real-time requirement of embedded system and the necessity of accurate attitude detection in practical application, an algorithm is proposed – using the accelerometer and gyroscope to update quaternion respectively, and using the fast complementary filter to fuse the data.And then, the heading angle is corrected by the magnetic flux density which is processed by median filter based on the threshold value.Therefore, the system fuses data of sensors and estimates the attitude.To verify the feasibility and effectiveness of the algorithm, we applied it to the intelligent vacuum cleaner platform.The experimental results demonstrate that the intelligent vacuum cleaner using proposed algorithm realizes the accurate attitude control and can accurately select path.And path planning based on the attitude control can improve coverage of the indoor cleaning area for intelligent vacuum cleaner.