Hybrid Navigation of Omnidirectional AMRs Using Dynamic Control and FWNN-BLS Compensation
##plugins.themes.bootstrap3.article.main##
Abstract
This paper proposes a novel hybrid navigation method, combining trackless and magnetically tracked navigation, for an omnidirectional AMR (OAMR), in order to achieve trackless navigation is the absence of no magnetic tape path, and switch to tracked navigation when a magnetic tape path is detected. For the trackless navigation, a dynamic motion controller is designed by integrating a backstepping PI kinematic control method and a fuzzy wavelet neural network (FWNN) augmented by a broad learning system (BLS), or abbreviated as FWNN-BLS. In the dynamic controller, the FWNN-BLS is employed to online learn uncertain dynamic behavior of the OAMR and then serve as a compensator, thus resulting in better performance and lower tracking errors during navigation. This dynamic controller works with the built Gmapping SLAM and A* global path algorithm to accomplish trackless navigation. For tracked navigation, the kinematic model of the OAMR is modified into a differential drive mode by setting the equal speeds of both wheels at the same side, and a PID controller is used with the magnetic guide sensor to carry out tracked navigation. The proposed hybrid navigation method is validated through simulations and experimental results, thus demonstrating its effectiveness and practicality.
Download Statistics
##plugins.themes.bootstrap3.article.details##
Dynamic control, fuzzy wavelet neural network- broad learning system, hybrid navigation, omnidirectional AMR
This work is licensed under a Creative Commons Attribution 4.0 International License.
Creative Commons CC BY 4.0