Modelling of An Agricultural Robot Applying Neuro-Fuzzy Inference System Approach

This paper emphasizes the modelling of an agricultural robot, API. It is expected to build a model by the Neuro-Fuzzy method according to its high nonlinearity, multivariate and time-variation. Firstly the neuro-fuzzy model is built by ANFIS algorithm. Particularly, the equivalent wheel is proposed in the paper. It decreases the number of the fuzzy rules sharply, hence to enhance the transparency of the ANFIS model. Secondly in the amount simulation, it is compared with the conventional mathematical model where the ANFIS model has the smaller error than the conventional mathematical model does. It can be concluded that the neuro-fuzzy modelling approach is much robust than the conventional mathematical model when the noise mixed in the input measurements.