Neuro-Fuzzy Modelling And Model-based Fault Detection On An Automonous Augriculture Robot

API (Autonomous Plant-care Implementation) is a sort of autonomous agriculture robot, developed by Aalborg University and Agriculture Science Institute in Denmark. The main task of API is to be used in agriculture autonomy. For example, API can be used in removing the weed in the field, which both relieves the hard-work of farmer and decreases the usage of the herbicide evidently. Therefore it can be said that the autonomous agriculture robot is the hot spot of agricultural modernization. As a complicated robot system, API system fuses many high-techs, such as the robotics, electric-mechanics, communication and computer science, robot vision and sensor fusion, fault detection etc.This thesis takes modeling of API and its fault detection as the two main parts. 1) The modeling of API based on Neuro-Fuzzy methodAPI robot is a complicated system of the strong nonlinearity, multi-varable and time variance. It takes time and pains to model by traditional mathematical approach. Moreover the model made under the assumptions will lose the flexibility to the real life. Because the neuro-fuzzy method is different from the analytically mathematical modeling method 'white box method' and the method totally depending on the experiment data 'black box method', the neuro-fuzzy method is a widely-used method, employing the measurement data and the...