The purpose of this study is to develop a method that is able to create a suboptimal path of an agricultural mobile robot. This work is an attempt to apply a control technique combining a neural network (NN) and a genetic algorithm (GA). A NN is applied to describe the motion of the agricultural mobile robot as a nonlinear system because it is able to identify the dynamics of complex systems with its high learning ability. To create a path using a simulator described by the NN, a GA, which is inspired by biological evolution and uses a process of variation and selection to search a solution space, is utilized as an optimization method. Using this simulator and the GA, the time series of the steer angles, which were the control input, were optimized, and consequently an optimal work path of the mobile robot was created. The technique explored here should be applicable to a wide variety of nonlinear control problems in agriculture. (C) 1997 Elsevier Science B.V.
