Abstract: Abstract: Multi-rotor is an aircraft that can vertically takeoff, land and freely hover. It can be used as a platform to monitor field information near ground with remote sensing technology. It is a beneficial supplement to satellite remote sensing monitoring. Flight control is the core of the whole flight system and attitude detection is an important part of flight control. Therefore, real-time acquire accurate attitude information is the basis for its flight controlling. In this paper, taking multi-rotor as object, a multi-sensor attitude detection system was established. The system included STM32F103 as microprocessor, MPU6050 as three-axis gyroscope and accelerometer, HMC5883 as electronic compass. However, when vehicle accelerated horizontally, deviations among sensors would be produced because attitude angular could be measured by gyroscope timely, but accelerometer and electronic compass would lag due to the averaging. Moreover, accelerometer cannot distinguish any horizontal acceleration from gravity. In order to eliminate the measure deviations among above sensors, multi-sensor information fusion method with fuzzy-proportion integration (fuzzy-PI) deviation correction based on quaternion and its coordinate conversion was proposed. Fuzzy-proportion integration (fuzzy-PI) deviation correction method integrated precise performance of PI regulator and fast feature of fuzzy regulator. According to the vehicle at low or high speed motion, the deviations would be bigger or smaller. When the deviations and its integrations were less than the predetermined threshold value, it switched to PI regulator to obtain accurate measure. When they were greater than the predetermined threshold value, the algorithm switched to fuzzy regulator to correct the deviation as soon as possible. Attitude angle detected by fuzzy-PI deviation correction method, complementary filter algorithm and DMP program were validated by SGT320E multi-functional three-axis rotary platform. The measurement results were monitored and acquired through serial port. When roll angle ranged from ?40° to 40°, static angle measurements were made. Experiment result showed that the average measure error of roll angle and pitch angle with the proposed method was 1.213° and 1.072°, less than complementary filter algorithm. It could satisfy the accuracy requirement in aircraft attitude detection. Furthermore, time consumption per attitude angle solution with above 3 methods running on the proposed multi-sensor attitude detection hardware system was recorded through timer in STM32 and their solution time consumptions were less than 1ms, which proved that the hardware system can satisfy the quick response requirement of angle measure. Solution time of fuzzy-PI deviation correction method was 2 134 μs, less than the other tow methods because the hardware system with fuzzy-PI deviation correction method had a good real-time performance and was suitable for fast calculation measure. In addition, in long static measurement, the proposed method did not generate static drift or divergence and in dynamic measurement, roll angle and pitch angle tracking maximum error was 1.969° and 3.581° with measurement accuracy reaching 1°, which results were closed to DMP solver within MPU6050. The dynamic measure experiment demonstrated that dynamic correcting deviation among multi-sensor in the fuzzy-PI deviation correction method can effectively modify the deviation among sensors, fast and accurately measure vehicle movement. Our study showed that the multi-sensor hardware system based on fuzzy-PI deviation correction method with characteristics of higher solution frequency, higher accuracy, lower cost, can implement for detecting attitude angles of wide range types of vehicles.