The study of the convergence rate of the algorithm extended Kalman filter — adaptive digital filter system

An important aspect of the information processing performing by a robot onboard computer is the convergence rate of the recurrent algorithms for data estimation and filtering used in a robot control system. This paper focuses on examining the convergence rate of the extended Kalman filter–adaptive digital filter (EKF–ADF) system algorithm. It explores the system’s performance as a digital filter for deterministic input signals, aiming to elucidate its potential applications in robotics. The criteria for determining the convergence rate are: the iterations’ number of the EKF–ADF system algorithm which required to achieve a steady-state value of the system operation results’ mean squared error and the iterations’ number of the EKF–ADF system algorithm which required to achieve a median value of the system operation results’ mean squared error. This paper evaluates and compares the signal filtering outcomes achieved by the EKF–ADF system with those obtained using EKF algorithms and various adaptive digital filters. In order to determine the influence of the ADF buffer memory organization in the EKF–ADF system on the signal filtering results and the convergence rate of the system algorithm, a simulation of this system operation with a different number of ADF buffer memory cells was carried out. The study’s findings indicate that the EKF–ADF system algorithm offers several key advantages, including a high convergence rate—exceeded only by the EKF algorithm itself—and effective noise suppression, even in the presence of substantial noise levels. By integrating the extended Kalman filter with the adaptive digital filter, the EKF–ADF system compensates for the errors associated with the extended Kalman filter’s operation, thereby enhancing the accuracy of system and process estimations. Inclusion of the EKF in the EKF–ADF system makes it possible to use nonlinear, but linearizable, models of the systems/processes estimated in robot control systems.
EDN: QZXTMB, https://elibrary/qzxtmb

References

[1] Urrea C., Agramonte R. Kalman filter: historical overview and review of its use in robotics 60 years after its creation. J. Sens., 2021, vol. 2021, art. 9674015, doi: https://doi.org/10.1155/2021/9674015

[2] Khodarahmi M., Maihami V. A review on Kalman filter models. Arch. Computat. Methods Eng., 2023, vol. 30, no. 1, pp. 727–747, doi: https://doi.org/10.1007/s11831-022-09815-7

[3] Lv M., Wei H., Fu X. et al. A loosely coupled extended Kalman filter algorithm for agricultural scene-based multi-sensor fusion. Front. Plant Sci., 2022, vol. 13, art. 849260, doi: https://doi.org/10.3389/fpls.2022.849260

[4] Feng S., Li X., Zhang S. et al. A review: state estimation based on hybrid models of Kalman filter and neural network. Syst. Sci. Control. Eng., 2023, vol. 11, no. 1, art. 2173682, doi: https://doi.org/10.1080/21642583.2023.2173682

[5] Schmidt S.F. Application of state-space methods to navigation problems. Adv. Control Syst., 1966, vol. 3, pp. 293–340, doi: https://doi.org/10.1016/B978-1-4831-6716-9.50011-4

[6] Bezmen P.A. Investigation of the operation of the extended Kalman filter supplemented by an adaptive digital filter for integrating data from a mobile robot control system. Izvestiya Yugo-Zapadnogo gosudarstvennogo universiteta [Proceedings of the Southwest State University], 2020, vol. 24, no. 1, pp. 68–89, doi: https://doi.org/10.21869/2223-1560-2020-24-1-68-89 (in Russ.).

[7] Bezmen P.A. Tsifrovoy filtr dlya nestatsionarnykh signalov [Digital filter for non-stationary signals]. Patent RU 2747199. Appl. 05.07.2020, publ. 29.04.2021. (In Russ.).

[8] Bezmen P.A. Studying the extended Kalman filter — adaptive digital filter system operation as the non-stationary signal filter. Izvestiya vysshikh uchebnykh zavedeniy. Mashinostroenie [BMSTU Journal of Mechanical Engineering], 2024, no. 10, pp. 9–19. EDN: ONANWI (in Russ.).

[9] Somefun C.T., Daramola S.A., Somefun T.E. Advancements and applications of adaptive filters in signal processing. JESA, 2024, vol. 57, no. 5, pp. 1259–1272, doi: https://doi.org/10.18280/jesa.570502

[10] Bychkov B.I., Romanovskiy A.S., Khartov V.Ya. Noise-free speech channel modeling for technical control systems. Radiooptika [Radio Engineering], 2016, no. 5, pp. 11–25, doi: https://doi.org/10.7463/rdopt.0516.0848125 (in Russ.).

[11] La Rosa A.B., Pereira P.T., Ücker P. et al. Exploring NLMS-based adaptive filter hardware architectures for eliminating power line interference in EEG signals. Circuits Syst. Signal Process., 2021, vol. 40, no. 5, pp. 3305–3337, doi: https://doi.org/10.1007/s00034-020-01620-6

[12] Khan A.A., Shah S.M., Raja M.A.Z. et al. Fractional LMS and NLMS algorithms for line echo cancellation. Arab. J. Sci. Eng., 2021, vol. 46, no. 4, pp. 9385–9398, doi: https://doi.org/10.1007/s13369-020-05264-1

[13] Kumar K., Pandey R., Karthik M.L.N.S. et al. Robust and sparsity-aware adaptive filters: a review. Signal Process., 2021, vol. 189, art. 108276, doi: https://doi.org/10.1016/j.sigpro.2021.108276

[14] Dzhigan V.I., Stempkovskiy A.L. LMS adaptive filtering algorithm: first or unique one for practical applications? Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh system [Problems of advanced micro- and nanoelectronic systems development], 2014, no. 4, pp. 159–166. (In Russ.).

[15] Kruse R.L., Ryba A.J. Data structures and program design in C++. Prentice-Hall, 1999. 717 p.

[16] Wang M. Design and implementation of asynchronous FIFO. Appl. Comput. Eng., 2024, vol. 70, pp. 215–221, doi: https://doi.org/10.54254/2755-2721/70/20241023

[17] Jianu M., Dăuş L. A note on the Heaviside step function. Proc. 18th Workshop on mathematics, Computer Science and Technical Education, 2021, vol. 4, pp. 31–38.

[18] Bezmen P.A. Nabor bibliotek «RFK-ATsF-ARS» realizatsii sistemy upravleniya sostoyaniem obekta [Set of “RFK-ACF-ARS” libraries for realization of the object state control system]. Software reg. cert. no. 2022663792 RU. Appl. 01.07.2022, publ. 20.07.2022. (In Russ.).