粒子滤波在微型飞行器姿态融合中的应用

Application of Particle Filter in Attitude Fusion of MAV

考虑廉价的惯导器件对周围环境很敏感,输出噪声未必是理想的高斯噪声。使用卡尔曼滤波器融合的姿态数据可能对环境的适应性不强。因此我们考虑利用粒子滤波器对一些廉价惯导器件输出的数据进行了融合以得到精度更高的角度、角速度数据,为飞行器的控制提供可靠的姿态数据。在实验阶段我们对比卡尔曼滤波器分析粒子滤波算法的性能,并用陀螺仪、加速度传感器(MPU6050)在实际的嵌入式系统上对算法进行实物验证。最终得到的结论是粒子滤波在有色噪声干扰条件下对噪声的强于较卡尔曼滤波,且在实际的嵌入式系统测试中粒子滤波的动态性更好。

Considering the low-priced inertial navigation devices is very sensitive to the environment and the output noise is not completely Gauss noise. Using Kalman filter to mix attitude data may not be strong adaptability to the environment. Therefore, this article considers to use particle filter for data of low-priced inertial navigation devices to estimate a higher accuracy angle and palstance, it provides reliable data for control the aircraft. At the experimental stage, the performance of the Kalman filter with particle filter algorithm is compared, and the gyro-acceleration sensor （MPU6050） is used to verify the algorithm in embedded system. Finally, it is concluded that the particle is stronger than that of Kalman filter in depress color noise, and the dynamic property of the particle filter is better than Kalman filter in embedded system test.