Application of Neuro-Fuzzy Controller to Autonomous Agricultural Vehicle Operating on Unstructured Changing Terrain -Control Software Development-

Abstract

The control of an autonomous agricultural vehicle operating on unstructured changing terrain includes many objective diffi culties. One major diffi culty concerns the characteristics of the terrain condition that the vehicle should operate in. Problems ranged from the effects of varying terrain conditions on the autonomous vehicle sensors and traction performance through to the need to deal with the presence of unexpected situations. On unstructured changing terrain, many factors infl uence vehicle behavior such as terrain slope, lateral slippage, and so on. Therefore, it is necessary to develop a more suitable model for vehicle motion on these terrain conditions. In order to control the vehicle along a course on unstructured changing terrain, it was developed control software to enable more accurate control. The developed method to control the vehicle when operating on these conditions was Neuro-Fuzzy Controller. Result of the trained model could be described as follows; number of nodes was 193, number of fuzzy rules was 81, average testing error between simulation and ANFIS output was 0.76, while for experimental and ANFIS output was 1.61. It was concluded that the developed control system had a good accuracy to steer the vehicle.