A Study about Kalman Filters Applied to Embedded Sensors

作     者：Valade, Aurelien Acco, Pascal Grabolosa, Pierre Fourniols, Jean-Yves 

作者机构：Univ Toulouse CNRS INSA LAAS F-31031 Toulouse France Inst Mediterraneen Enseignement & Rech Informat & F-66004 Perpignan France 

出 版 物：《SENSORS》 (传感器)

年 卷 期：2017年第17卷第12期

页      面：2810-2810页

核心收录：

学科分类：0710[理学-生物学] 071010[理学-生物化学与分子生物学] 0808[工学-电气工程] 07[理学] 0804[工学-仪器科学与技术] 0703[理学-化学] 

基　　金：TE Connectivity Institut Mediteraneen d'Enseignement et de Recherche en Informatique et Robotique (IMERIR) 

主　　题：smart sensors Kalman filters algorithm complexity IMU compensation 

摘      要：Over the last decade, smart sensors have grown in complexity and can now handle multiple measurement sources. This work establishes a methodology to achieve better estimates of physical values by processing raw measurements within a sensor using multi-physical models and Kalman filters for data fusion. A driving constraint being production cost and power consumption, this methodology focuses on algorithmic complexity while meeting real-time constraints and improving both precision and reliability despite low power processors limitations. Consequently, processing time available for other tasks is maximized. The known problem of estimating a 2D orientation using an inertial measurement unit with automatic gyroscope bias compensation will be used to illustrate the proposed methodology applied to a low power STM32L053 microcontroller. This application shows promising results with a processing time of 1.18 ms at 32 MHz with a 3.8% CPU usage due to the computation at a 26 Hz measurement and estimation rate.