Design Research On Multisensor Data Fusion In ARM Based Strapdown Attitude Reference System

An Unmanned Aerial Vehicle (UAV) is a self-propelled aircraft that can fly autonomously and carry out multiple tasks. In order to realize UAV’s intelligent and autonomous flight, onboard flight attitude reference system is one of the key techniques that we should solve. UAV’s performance in executing an assignment depends largely on the stability of the flight attitude. Due to the presence of motion inertia and environment impact, the aircraft in flight is a highly unstable platform, and a tiny tilting angle can result in a considerable deviation of the aircraft’s movement contrail.To measure and compensate the kinetic parameters of onboard inertia devices and improve the UAV’s autonomous flight performance to a great extent, the dissertation has devised a aircraft attitude reference system based on ARM with multi-sensor data fusion methods. The embedded attitude reference system is divided into two parts:signal capturing unit and attitude control unit. The former consists of posture angle sensors and angular velocity sensors, which supply the aircraft’s real-time attitude information. Low-pass filtering algorithm and Kalman filtering theory are also applied to inhibit the measurement error of inertial sensors and ensure the measuring accuracy of signal acquisition unit. The latter generates steering-engines driving signal which serves as the actuator in the control of fins of rudders to make adjustments and corrections to aircraft’s flight attitude timely and accurately.Firstly, this paper introduces an ARM implementation including the overall hardware structure and specific sensor selection. The realization principles and methods of the attitude signal acquisition part are analyzed in detail. Then the method of generating the PWM control waveform based on ARM internal timers is present. And then we port the real-time kernel μ C/OS-II to Samsung ARM920TDMI and develop embedded multitask system, which is capable of dispatching attitude data acquisition channels and controlling PWM output channels.Secondly, We design the linear Kalman filtering algorithm to process each single sensor signals to separate the white Gaussian noise from the useful signals and get the best estimation of angles or angular rates on the least mean-square error rule. In order to improve the robustness, we apply multi-sensors to collect the attitude signals in parallel. Federated Kalman filter, optimally weighted least squares fusion algorithm and extended Kalman filter are designed to process and fuse the data from each separate filter. Eventually, the accuracy and reliability of measurement system can be guaranteed, even if one or some transducers are ineffective.Finally, the data acquisition in LabVIEW and hardware in-the-loop simulation for the attitude control system that are used to compare with Matlab simulation are discussed. Wave-display and three-dimensional dynamic state display are realized on computer to conduct experimental investigation. According to the experimental results, we could review the system’s design for the optimal system.The design of embedded attitude reference system is based on hardware modularization and various kinds of communication interfaces with ARM9micro processor. The integrate system has several advantages such as low power consumption, low cost, reduced size and better performance. The software is programmed in C++, and the optimization and refinement of algorithms are easy to be accomplished. Many experiments justify that the system can judge the aircraft’s attitude angles accurately and balance the aircraft in flight.