Open AccessIntegration of Mobile Robot Control System Data Using the Extended Kalman Filter
Author(s) -
Petr Bezmen
Publication year - 2019
Publication title -
izvestiâ ûgo-zapadnogo gosudarstvennogo universiteta
Language(s) - English
Resource type - Journals
eISSN - 2686-6757
pISSN - 2223-1560
DOI - 10.21869/2223-1560-2019-23-2-53-64
Subject(s) - kalman filter , computer science , sensor fusion , mobile robot , fast kalman filter , robot , extended kalman filter , real time computing , computer vision , artificial intelligence
Purpose of research. The article deals with the adaptation of the algorithm of the extended Kalman filter for the integration of data from sensors of physical values of a mobile robot Methods. Integration of data is the process of information (data) fusion for determination or prediction of the state of an object. Integration provides increased robustness of robot control and accuracy of machine perception of information. This process is similar to repeated experiments in order to determine in direct and/or indirect ways the value of a physical quantity with the required accuracy. In the control system of a mobile robot, the integration of sensor data is carried out by one or more computing devices (for example, processors or microcontrollers) [1-5]. Results. Advances in digital signal processing and image processing are based on new algorithms, increasing the speed of data processing by computing devices and increasing the speed of access to data stored in storage (storage devices) and the capacity of the latter. Computing devices also perform averaging and filtering of signals of individual sensors and their further matching. The problem of sustainable integration and processing of information from different measuring devices can be solved with the help of the Kalman filter algorithm. The Kalman linear filter algorithm and, in particular, the extended Kalman filter algorithm perform a large amount of computation in the course of their work. In comparison with the linear Kalman filter, the extended Kalman filter significantly increases the requirements for the computing power of the onboard computer (computing device, computer) of amobile robot. Conclusion. The main effect of integration is to obtain fundamentally new information that cannot be obtained from individual sensors. This approach relieves data channels of large (excessive) data flows coming directly from the sensors, and reduces the requirements for computing power of the computing device of the upper level of the structure of the mobile robot control system.