Abstract
In this paper, in order to contrive a method that changes a vehicle for farm and construction use into an autonomous vehicle, an attempt was made to create an optimal route in field by using genetic algorithm (GA). First, kinematics of the vehicle was described in a simulator that was a nonlinear dynamic model with neural networks (NN), and its accuracy was investigated using trained data and untrained data. Next, the method of creating an optimal route was contrived by using the NN simulator and GA. It was demonstrated that this method was effective to get an optimal route and the time series of steer angles. The optimal route was created under some boundary conditions and constraints. It was also possible to determine an obstacle avoiding route through this method.
