Multi-scenario data driven fuzzy TSK nonholonomic mobile robot modelling

In this paper the problem of multi-scenario data driven fuzzy parameter estimation is considered. Experimental data are used from a small scale differentially steered four-wheel mobile robot “PROMETHEUS”. In particular two key modes of operation were identified and the multi-model parameters were obtained using the subtractive clustering approach. The two modes of the mobile robot operation were blended using a suitable blending function. The robotic vehicle modes structure was of a 1-st order multivariate Takagi-Sugeno-Kang. The parameter estimation process also included a noncasual filtering approach which resulted in a reduced number of TSK rules.