基于四元数和卡尔曼滤波的两轮车姿态稳定方法

摘要

针对自平衡两轮车姿态角的在线估计问题,采用四元数的姿态解算算法,利用Levenberg-Marquardt非线性最小二乘法对自平衡两轮车陀螺仪信号拟合,建立了随机漂移误差数学模型,应用卡尔曼滤波融合陀螺仪和加速度计输出的信号,补偿了陀螺仪输出角速度的随机漂移误差,得到了自平衡两轮车姿态的最优估计.实验结果表明,这种姿态估计算法是有效的,有利于车体的自平衡控制.%Based on the on-line estimation of the two-wheeled self-balance vehicle's attitude angles, an attitude solution algorithm based on quaternion is used. According to the output of gyroscope, the random drift error model is fitted by Levenberg-Marquardt nonlinear least-square method. Using Kill man filter to fuse the data from gyroscope and aecelerometer,an optimal estimation for the attitude of the two-wheeled self-balance vehicle is achieved after the compensation of the angular velocity drift error of gyroscope. The results of experiments show that the attitude estimation algorithm is effective. It is beneficial to the self-control of the two-wheeled self-balance vehicle.