一种用于人体运动捕获的自适应混合滤波融合算法

发布时间：2018-02-16 11:06

  本文关键词： 传感器 信息融合 高斯牛顿算法 互补滤波 人体运动捕获　出处：《工程科学与技术》2017年05期 　论文类型：期刊论文

【摘要】：针对基于惯性传感器的人体运动捕获系统存在陀螺漂移和噪声干扰等问题,提出一种多元传感器信息融合的自适应混合滤波融合算法。算法首先利用快速高斯牛顿法对加速度计和磁力计数据进行姿态信息迭代估算,用四元数将参考坐标系中的加速度和磁场强度分量转换到载体坐标中,将转换后的值与当前时刻测量值的差值代入高斯牛顿迭代算法中用于四元数的实时值估计,通过确定搜索步长的最优值来缩短迭代次数,提高算法收敛速度。设计自适应的互补滤波器将高斯牛顿法解算的姿态信息作为观测矢量对陀螺漂移进行补偿,分别使用高通滤波器和低通滤波器处理陀螺仪数据和高斯牛顿算法优化过后的加速度计、磁力计数据。在互补滤波器中引入重力矢量及地磁参考矢量自适应调节滤波器参数用于实时调整不同算法的权重大小,融合后输出最终的姿态信息,实现最优估计。进行实验对比分析本算法和其他算法融合效果,结果表明,本算法有效降低陀螺累积误差、线性加速度及磁场对解算精度的干扰,磁干扰状态下误差为0.94°,自由运动状态下误差为1°。对比扩展卡尔曼滤波融合算法,本文算法执行时间缩短25%,有效提升了运动捕获系统的性能。
[Abstract]:Aiming at the problems of gyro drift and noise interference in the human motion capture system based on inertial sensor, An adaptive hybrid filtering fusion algorithm for multi-sensor information fusion is proposed. Firstly, the fast Gao Si Newton method is used to estimate the attitude information of accelerometers and magnetometers. The acceleration and magnetic field intensity components in the reference coordinate system are converted to the carrier coordinates by quaternions. The difference between the converted values and the measured values at the current time is substituted in the Gao Si Newton iterative algorithm for the real-time estimation of quaternions. By determining the optimal value of the search step size, the iteration times are shortened and the convergence rate of the algorithm is improved. An adaptive complementary filter is designed to compensate the gyro drift with the attitude information calculated by Gao Si Newton method as the observation vector. Using high-pass filter and low-pass filter respectively to process gyroscope data and the accelerometer optimized by Gao Si Newton algorithm, The parameters of gravity vector and geomagnetic reference vector are introduced into the complementary filter to adjust the weights of different algorithms in real time, and the final attitude information is output after fusion. The experimental results show that the proposed algorithm can effectively reduce the error of gyroscope accumulation, the interference of linear acceleration and magnetic field on the accuracy of the solution. The error is 0.94 掳in the state of magnetic interference and 1 掳in the state of free motion. Compared with the extended Kalman filter fusion algorithm, the execution time of this algorithm is shortened by 25%, which effectively improves the performance of the motion capture system.
【作者单位】： 重庆邮电大学光电信息感测与传输技术重庆市重点实验室;
【基金】：国家自然科学基金资助项目(51175535) 国际联合研究中心科技平台与基地建设项目资助(cstc2014gjhz0038) 重庆市基础与前沿研究计划资助项目(cstc2015jcyj BX0068) 重庆邮电大学博士启动基金资助项目(A2015-40);重庆邮电大学自然科学基金资助项目(A2015-49)
【分类号】：TN713;TP212
，

本文编号：1515364